Minerals Engineering 24 (2011) 1595–1602
Contents lists available at SciVerse ScienceDirect
Minerals Engineering journal homepage: www.elsevier.com/locate/mineng
Extraction of lithium from micaceous waste from china clay production E. Siame, R.D. Pascoe ⇑ Camborne School of Mines, College of Engineering, Mathematics and Physical Sciences, University of Exeter, Cornwall Campus, Penryn, Cornwall TR10 9EZ, UK
a r t i c l e
i n f o
Article history: Received 7 June 2011 Accepted 22 August 2011 Available online 16 September 2011 Keywords: Industrial minerals Froth flotation Roasting Leaching Waste processing Lithium
a b s t r a c t The granites of South-West England are a potential source of lithium which is generally found within the mica mineral, zinnwaldite. It is mainly found in the central and western end of the St. Austell granite. When kaolin extraction occurs in these areas a mica-rich waste product is produced which is currently disposed of in tailings storage facilities. In this study a tailings sample containing 0.84% Li2O was upgraded by a combination of froth flotation, using dodecylamine as the collector, and wet high intensity magnetic separation (WHIMS) to 2.07% Li2O. The concentrate was then roasted with various additives, including limestone, gypsum and sodium sulphate, over a range of temperatures. The resulting products were then pulverised before being leached with water at 85 °C. Analysis of these products by XRD revealed that the water-soluble sulphates, KLiSO4 and Li2KNa(SO4)2, were produced under specific conditions. A maximum lithium extraction of approximately 84% was obtained using gypsum at 1050 °C. Sodium sulphate produced a superior lithium extraction of up to 97% at 850 °C. In all cases iron extraction was very low. Preliminary tests on the leach solution obtained by using sodium sulphate as an additive have shown that a Li2CO3 product with a purity of >90% could be produced by precipitation with sodium carbonate although more work is required to reach the industrial target of >99%. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Lithium is important for a number of uses, including production of batteries, glass and ceramics. It is also used in the production of aluminium, preparation of greases, rubbers, alloys and pharmaceuticals. In 2008 lithium battery production represented 70% of the total rechargeable battery market (USGS, 2010) which includes mobile phones, laptop computers and power tools. The use of lithium for batteries has been increasing by more than 20% per year (USGS, 2010). The use of lithium-ion batteries in hybrid electric vehicles, plug-in hybrid and pure electric vehicles could see further significant increases in lithium production. Forecasts indicate that the demand for lithium in the next 5 years is expected to increase by approximately 60% from 102,000 t to 162,00 t of lithium carbonate or equivalent (LCE), with batteries representing more than 40,000 t of the perceived growth (Hykawy, 2010). The primary source of lithium is from continental brines which typically contain 0.06–0.15% Li followed by pegmatites. The principal lithium minerals from pegmatite, with their theoretical maximum lithium content, are shown in Table 1. Most of the lithium minerals from pegmatite are used for glass and ceramic production. Lithium chemicals, such as lithium carbonate, are normally produced from brines because of the lower ⇑ Corresponding author. Tel.: +44 1326 371838. E-mail address: r.d.pascoe@exeter.ac.uk (R.D. Pascoe). 0892-6875/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.mineng.2011.08.013
costs involved. China does however produce some lithium carbonate using imported spodumene as a feed stock. In a future scenario in which brine deposits could not meet the demand for lithium carbonate the deficit would have to be made up by increasing the use of lithium minerals from pegmatite. Zinnwaldite is one mineral that could be exploited as a lithium source in the future. The high iron content of the mineral combined with the relatively low Li2O content makes it relatively unattractive at the current time. The processing of lithium minerals from pegmatites involves both comminution and physical separation techniques such as gravity concentration, froth flotation and magnetic separation (Bale and May, 1989; Amarante et al., 1999). A novel comminution technique involving the application of high voltage pulses has been shown to improve the liberation of spodumene (Brandt and Haus, 2010). Once a lithium mineral concentrate has been produced it is typically roasted followed by leaching the products with either acid or water. This is an energy intensive chemical process. The total cost of the process is significantly affected by the requirements for mining, fine grinding, physical separation, high temperature roasting and evaporation. A number of lithium extraction processes have been reported for spodumene, petalite, lepidolite (Wietelmann and Bauer, 2008; Dresler et al., 1998) and zinnwaldite concentrates (Alex and Suri, 1996; Jandova and Vu, 2008; Jandova et al., 2009, 2010). The method used for the extraction of lithium from zinnwaldite by Jandova and Vu (2008) and Jandova et al. (2009, 2010) involved roasting of the concentrate with
1596
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
Table 1 Lithium minerals from pegmatite (from Harben (2002) and USGS (2010)). Mineral
Formula
% Li2O
Resource
Spodumene Petalite Lepidolite Amblygonite Bikitaite Eucryptite Montebrasite Jadarite Zinnwaldite
LiAlSi2O6 LiAlSi4O10 K(Li,Al)3(Si,Al)4O10(OH,F)2 LiAlPO4(F,OH) LiAlSi2O6H2O LiAlSiO4 Li2O Al2O3 2SiO2 LiNaSiB3O7(OH) KLiFeAl(AlSi3)O10(F OH)2
8.0 4.9 6.2 10.3 11.8 11.9 7 3.16 2.5-5
Australia, China, Canada, Zimbabwe, Portugal and Finland Zimbabwe, Namibia and Canada Zimbabwe Zimbabwe Zimbabwe Zimbabwe Canada Serbia – at feasibility stage No current exploitation
limestone and gypsum to produce a water-soluble lithium salt. Roasting with gypsum/Ca(OH)2 produced a lithium extraction of over 90% between 900 and 980 °C. With limestone a peak leach extraction of 90% occurred at 825 °C, with a significant drop-off above that temperature possibly caused by increased crystallinity of eucryptite (Jandova and Vu, 2008). Eucryptite normally requires leaching with strong acid for lithium leach extraction. The mineral is known to exist as different polymorphs that have varying stability ranges depending on temperature. Co-extraction of rubidium was around 25% for gypsum/Ca(OH)2 and 90% for the limestone at the temperature that gave optimum lithium extraction. Alex and Suri (1996) used pugging of zinnwaldite at 700 °C with sulphuric acid to give lithium extraction of 90%. Unfortunately there was significant co-extraction of both iron and aluminium under these conditions. The lithium potential of the St. Austell granite, situated in Cornwall (UK) was investigated by the British Geological Survey (Hawkes et al., 1987). An area of approximately 8 km2 100 m depth was identified as containing a resource of 3.3 million tonnes of lithium. The china clay operations at Rostowrack and Treleavour Downs, run by Goonvean Ltd., fall within this resource area. Using the china clay production figures for these operations, it has been estimated that approximately 100,000 tonnes per year of micaceous residues (the hydrocylcone underflow product), assaying around 0.84% Li2O, are available for treatment. A sample of micaceous residues from these areas was used in this study. Additional resource would be available by reprocessing material held in nearby tailings dams, although a lower Li2O content would be expected as a result of mixing with residues from operations in lower Li areas. The advantage of using the hydrocylcone residue from an operating plant is that no additional mining costs are accrued. In addition the minerals within the residue are typically fine and well liberated therefore saving on the grinding costs that would result from the processing of other pegmatite deposits. Previous studies have mentioned the physical separation of waste materials to produce a mica concentrate. Jandova et al. (2010) used dry magnetic separation whereas Hawkes et al. (1987) considered both froth flotation, for recovery of a mica concentrate, and dry magnetic separation for the separation of the various mica minerals. In these investigations a limited size range was used and no quantitative data on lithium and rubidium recovery was presented. In this study we have investigated the efficiency of froth flotation and magnetic separation for separation of both lithium and rubidium from the hydrocyclone underflow. Where possible we have linked mineralogy with separator performance. Following production of a lithium–mica concentrate the effectiveness of the roast/water leach procedure has been investigated using the reagent systems considered by Jandova and Vu (2008) and Jandova et al. (2009, 2010) with the addition of sodium sulphate (Na2SO4). The mineral phase changes that occur during roasting have been followed using X-ray diffraction (XRD) and differential thermal analysis (DTA).
2. Experimental 2.1. Materials The material used in this test work was obtained from Goonvean Ltd., St. Austell, UK. The sample was collected from the underflow of a group of 250 mm diameter hydrocyclones, which were processing material for china clay production, mined from the Trelavour Downs and Rostowrack pits. The overflow is further processed into china clay products ready for sale. The underflow, which consists of fine mica-rich sand, is a waste product that is discharged to a nearby tailing dam (see Fig. 1). The hydrocyclone underflow was further classified using a 50 mm laboratory hydrocyclone operated at a pressure of 276 kPa in order to remove the majority of the 10 lm fraction which is known to be rich in kaolinite. Fig. 2 shows the particle size distribution of this fraction obtained using a Malvern laser-sizer (Mastersizer MAF 5000). The de-slimed hydrocyclone underflow was then homogenised before being riffled into 1.2 kg lots. These lots were the feed samples for the froth flotation experiments. 2.2. Analytical procedures Semi-quantitative information on the mineralogy of the feed, separation products and new materials formed on roasting were produced by X-ray diffraction (XRD) using a Siemens Diffractometer D5000. X-ray fluorescence (XRF), using a Bruker S4 Pioneer, with the boric jacket preparation method was used for elemental analysis of solid samples. Quantitative elemental analysis of specific mineral grains was undertaken using a JEOL JAX-8200 electron microprobe. Atomic absorption spectrometry (AAS), using a Unicam SP 9 spectrometer, was used for the determination of lithium
Feed from mining operation
Coarse classification using bucket wheel de-sander
Sand to contoured tips
Classification using hydrocyclones
Mica-rich residue to tailings dam
Kaolinite-rich product for further classification and refining Fig. 1. Simplified flow diagram of china clay production by Goonvean Ltd.
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
100
Cumulative % Undersize
90 80 70 60 50 40 30 20 10 0
1
10
100
1000
Particle Size (μm) Fig. 2. Particle size distribution of the de-slimed hydrocylone underflow.
Table 2 Estimated mineralogy of the hydrocyclone underflow sample from Goonvean Ltd. clay processing plant. Mineral
Chemical Formula
wt%
Quartz Muscovite Zinnwaldite Kaolinite K-Feldspar Topaz Apatite Total
SiO2 KAl2 (AlSi3)O10 (F,OH)2 KLiFeAl (AlSi3)O10 (F OH)2 Al2Si2O5 (OH)4 KAlSi3O8 Al2SiO4 (F,OH)2 Ca5(PO4)3(OH,F,Cl)
17 27 20 15 19 1 1 100
and rubidium. AAS analysis was performed on a solution produced by digesting the solid sample in a mixture of perchloric and hydrofluoric acid in a PTFE beaker, followed by evaporation until a solid residue remained. The residue was then dissolved in a solution of hydrochloric and boric acid, diluted and then analysed. The elemental analysis (from XRF and AAS) and XRD data were combined to predict the mineral composition of the feed sample which is shown in Table 2. In order to help understand what reactions were occurring during roasting, thermal analysis was carried out using a Stanton Redcroft STA 780. The instrument provides simultaneous thermogravimetric (TG) and differential thermal analysis (DTA) data which can be used to identify exothermic and endothermic reactions that occurred during the roasting stage. The DTA analysis was conducted with samples of 8 mg heated at a rate of 10 °C/min. 2.3. Physical separation to produce a Li-rich concentrate The mica fraction from the de-slimed hydrocyclone underflow was initially concentrated using froth flotation. The use of wet high intensity magnetic separation (WHIMS) was then investigated as a means of separating the individual mica minerals. A schematic of the overall separation process is shown in Fig. 3. 2.4. Froth flotation Flotation was conducted in a Denver D-12 laboratory flotation machine equipped with a 3.5 dm3 Minnovex cell. The impeller speed and air flow rate were set at 1400 rpm and 8 dm3/min respectively. A flotation test sample of 1.2 kg was mixed with 2.8 dm3 tap water and conditioned at 1400 rpm to give 30% solids by weight. Dodecylamine (98% purity, supplied by Sigma–Aldrich
1597
Company Ltd., UK) was used as the cationic collector. A stock solution of 1% w/v was made up and this was added to give an addition rate of 150–500 g/t for the mica flotation. The pulp was then conditioned for 5 min at the required pH prior to flotation for a period of 8 min. The pH was adjusted either by addition of dilute H2SO4 or NaOH. Initial tests revealed that Fe recovery (obtained from XRF measurement) provided a good indicator of both lithium and rubidium recovery. As Fe analysis was easier to perform it was used to evaluate flotation performance. The effect of collector concentration and pH on recovery is shown in Fig. 4. Each data point represents the average data from three experiments. The flotation concentrate grade was relatively consistent for all tests at 4.5–5% Fe2O3. Analysis of flotation results did not indicate a significant link between iron recovery and grade for both pH and collector dosage over the variable range tested. In order to produce a bulk magnetic product for further work a series of flotation tests were performed using 500 g/t of collector at pH 2.5. The two products in this case were analysed for lithium and rubidium by AAS as well as iron by XRF in order to calculate recoveries. The results are presented in Table 3. In order to determine the variation in composition by size a sample of the concentrate was screened at 150, 125, 106, 90, 75, 63 and 53 lm and the fractions analysed by AAS. The results are shown in Fig. 5. It can be seen from Fig. 5 that the grade of lithium, rubidium and iron increase as the particle size increases. The results indicate that the zinnwaldite tends to occur in coarser fractions. Previous studies have shown that the type of mica present in a china clay sample can vary significantly with size (Hawkes et al., 1987). One other contributing factor could be that entrainment of nonmica minerals increases with decreasing particle size. The information suggests that fine screening could provide further enrichment and that coarser residues from the processing plant (sand fractions from the bucket-wheel de-sanders) could potentially contain a higher lithium content in the mica than the hydrocyclone underflow product. 2.5. Wet high intensity magnetic separation The mica flotation concentrate was subjected to wet high intensity magnetic separation (WHIMS) using a batch Rapid Magnetic Ltd. separator using a matrix with a 1 mm gap to recover the high lithium/iron mica minerals. The magnetic field was adjustable up to a maximum of 2.06 Tesla. The test procedure involved reprocessing the non-magnetic product from the first test a further two times through the separator and combining the three magnetic products. This gave the maximum recovery possible but would be likely to adversely affect the grade. Fig. 6 shows the weight, Li2O, Rb2O and Fe2O3 recovery to the magnetic fraction as a function of magnetic field strength. The recoveries of the three components are almost identical suggesting that they are all present within the same mineral. The performance of the separator was then determined as a function of particle size as shown in Fig. 7. It can be seen that the mass recovery increases with increased particle size. Li2O recovery as a function of magnetic field strength (see Fig. 8) however is similar for each size fraction. This can be explained by the higher Li2O content of the coarser fractions. Similar trends were observed for iron and rubidium recoveries as a function of size and magnetic field strength. At 1.95T the Li2O content of the magnetic fraction increased from 1.48% for the 38 lm fraction to 2.94% for the +150 lm fraction. The Fe2O3 content increased from 4.55% to 10.06% for the same sizes. The highest Fe2O3 content in a non-magnetic fraction occurred at the +150 lm fraction at 4.25% reducing to 1.32% in
1598
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
De-slimed hydrocyclone underflow from china clay processing
Quartz, feldspar and kaolinite Froth flotation
Mica Concentrate Wet High Intensity Magnetic Separation
“Non-magnetic” mica
Paramagnetic mica Fig. 3. Mica separation flow sheet.
80
90
70
80
Recovery to Magnetic (%)
Fe2O3 Recovery to Concentrate (%)
100
70 60 50 40 30 20 10 0
60 50 40 30 20 10
0
100
200
300
400
500
600
0
Collector Dosage (g/t) pH 2.5
0
Li2O
59.3 40.7 100.00
Content (%)
Recovery (%)
Rb2O
Fe2O3
Li2O
Rb2O
Fe2O3
1.45 0.03 0.87
0.55 0.14 0.38
4.47 0.51 2.86
98.6 1.4 100.0
85.2 14.8 100.0
92.8 7.2 100.0
8
Composition (%)
7
2.5
Rb2O
Fe2O3
Weight
60 50 40 30 20
6
10
5
0
4
1
1.5
2
2.5
Magnetic Field (Tesla)
3 2 1 0
2
70
Li2O
Weight Recovery (%)
Concentrate Tailings Feed
1.5
Fig. 6. WHIMS recovery as a function of magnetic field strength.
Table 3 Analysis of bulk flotation feed and products. Weight (%)
1
Magnetic Field (Tesla)
Fig. 4. Flotation recovery of Fe2O3 as a function of collector dosage and pH.
Product
0.5
pH 5
pH 3.5
+150 microns
+106 microns
+75 microns
+53 microns
+38 microns
-38 microns
Fig. 7. WHIMS weight recovery to magnetic as a function of particle size and magnetic field strength.
0
20
40
60
80
100
120
140
160
Size (μm) Li2O
Rb2O
Fe2O3
Fig. 5. Variation in lithium, rubidium and iron oxides content in flotation concentrate by particle size.
the 38 lm fraction. The higher Fe2O3 content produced at the coarser sizes can be explained by the increasing contribution of gravitational forces in the separation zone resulting in a lowering in magnetic separator efficiency (Kelly and Spottiswood, 1982). Following these initial tests it was decided to maximise lithium recovery to the magnetic fraction by operating a three stage sepa-
1599
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
Li2O Recovery in Magnetic (%)
90
peratures for 60 min. The typical particle size of the additives was less than 14 lm. In the experiments with limestone the ratio of paramagnetic mica concentrate to limestone was 5:2. For both gypsum and sodium sulphate the weight ratio was 2:1 for the roasting temperature optimisation. The roasting temperatures ranged from 250 to 1100 °C. Optimisation of the ratio of materials was undertaken once the optimum roasting temperature had been identified. This was done using gypsum (ratio from 2 to 10:1) and sodium sulphate (ratio from 2 to 7:1).
80 70 60 50 40 30 20 10 0
2.7. Leaching process 1
1.2
1.4
1.6
1.8
2
2.2
Magnetic Field (Tesla) +150 microns +53 microns
+106 microns +38 microns
+75 microns
Fig. 8. Li2O recovery by WHIMS as a function of magnetic field strength.
ration at the maximum field strength of 1.95 Tesla. Approximately 3 kg of paramagnetic-mica concentrate was produced by repeating this process a number of times with the WHIMS. Table 4 gives the elemental analysis and calculated recoveries produced by the magnetic separation process. Microprobe analysis indicated that two main mica minerals were present in the magnetic separation products; zinnwaldite and muscovite. The composition of both minerals was quite variable but the two minerals were quite distinct when considering SiO2 and the calculated Li2O content (see Fig. 9). It can be seen that the muscovite contains 0–0.6% Li2O whereas the zinnwaldite contains 2.6–5.0% Li2O. Both non-magnetic and magnetic products include both minerals with the zinnwaldite making up 44% of the mica grains analysed in the non-magnetic fraction and 85% in the magnetic fraction.
2.6. Roasting process In preparation for the roasting process the paramagnetic mica concentrate was pulverised in a Tema mill with a tungsten carbide pot for 3 min before being mixed with the limestone, gypsum or sodium sulphate. Fig. 10 shows the particle size distribution of raw and pulverised paramagnetic mica concentrate. A predetermined weight of pulverised paramagnetic mica concentrate (90% <100 lm) was mixed with either limestone (CaCO3 supplied by BDH Laboratory Supplies, UK, at >99% purity), gypsum (CaSO4 2H2O, supplied by Fisher Scientific Ltd. at >98% purity) or sodium sulphate (Na2SO4, supplied by Hopkin and Company, UK, at >99.5% purity) and put in a ceramic crucible before being roasted in a laboratory Carbolite furnace (CWF 1200) at the selected tem-
The roasted products were pulverised before leaching using 3 min in the Tema mill. The typical size distribution achieved is given in Fig. 10. A 10 g sample of the pulverised product was then leached in 100 cm3 de-ionised water in a stirred-glass reaction vessel placed in a water bath maintained at 85 °C. Initial experiments with all products showed that if the lithium were soluble then 10 min leaching achieved almost full extraction. The standard leaching procedure was undertaken for 30 min to ensure maximum extraction. After leaching the samples were filtered, the leach solution (filtrate) collected and the residues dried. The lithium, rubidium and iron analysis of both products were determined by AAS. 3. Results 3.1. Limestone addition The influence of roasting temperature on lithium, rubidium and iron extraction is shown in Fig. 11. Experiments were conducted at a mica:limestone weight ratio from 2.5 to 5:1. At the 5:1 ratio used by Jandova and Vu (2008) the lithium extraction was <1%. The results shown in Fig. 12 are for a 5:2 ratio, which produced some lithium extraction. With these experiments the good lithium and rubidium leach extraction achieved by Jandova and Vu (2008) could not be replicated. This may be due to formation of sparingly soluble, crystalline eucryptite, which was the only lithium compound identified by XRD. As these results were disappointing the focus then turned to other possible additives. 3.2. Gypsum addition The influence of roasting temperature, using gypsum as the reactant, on lithium extraction is shown in Fig. 12. It can be seen that highly water-soluble lithium compounds are formed when roasting occurs at temperatures above 900 °C. A maximum lithium extraction with water of 84% was achieved after roasting at 1050 °C. Rubidium extraction was much lower at 14%. Iron coextraction was below the detection level under all conditions.
Table 4 Chemical analysis of the WHIMS products at a magnetic field of 1.95 Tesla. Fraction Magnetic Non-mag Head
Magnetic Non-mag Head
wt%
SiO2
50.3 49.7 100.0
Content (%) 40.1 27.8 50.6 32.0 45.3 29.9
50.3 49.7 100.0
Al2O3
Recovery (%) 44.5 46.8 55.5 53.2 100 100
Fe2O3
TiO2
MgO
CaO
K2O
Na2O
P2O5
F
Li2O
Rb2O
LOI
7.4 2.3 4.9
0.27 0.11 0.19
0.11 0.33 0.22
0.07 0.1 0.08
9.56 6.84 8.21
0.15 0.14 0.15
0.06 0.08 0.07
3.36 1.27 2.32
2.07 0.76 1.42
0.74 0.37 0.56
3.4 6.4 4.9
76.7 23.3 100
71.3 28.7 100
25.2 74.8 100
41.5 58.5 100
58.6 41.4 100
52.0 48.0 100
43.2 56.8 100
72.8 27.2 100
73.4 26.6 100
66.9 33.1 100
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
51
100
50
90
49
80
Metal Extraction (%)
Si2O (wt %)
1600
48 47 46 45 44 43 42
70 60 50 40 30 20
0
1
2
3
4
5
10
6
0
Li2O (wt %)
0
200
Muscovite (found in non-magnetic) Muscovite (found in magnetic) Zinnwaldite (found in non-magnetic) Zinnwaldite (found in magnetic)
400
600
800
1000
1200
Roast Temperature (ºC) Li
Fig. 9. Relationship between SiO2 and calculated Li2O content of the mica.
Rb
Fe
Fig. 12. Effect of roasting temperature on the water leach extraction of lithium, rubidium and iron from a mica:gypsum mix of weight ratio 2:1.
100
Cumulative Undesize (%)
90
ΔT
80 70 60 50 40 30 20 10 0
0
20
40
60
80
100
120
140
160
Particle size (μm) magnetic mica
0
200
400
600
800
1000
1200
Temperature (ºC)
pulverised magnetic mica
pulverised roasted product
Fig. 13. DTA profile for mica–gypsum (2:1) mixture.
Fig. 10. Particle size distributions of magnetic mica before and after pulverising and a typical pulverised roasted product.
10
Metal Extraction (%)
9 8 7 6 5 4 3 2 1 0
0
200
400
600
800
1000
1200
Roast Temperature (ºC) Li
Rb
Fe
Fig. 11. Effect of roasting temperature on the water leach extraction of lithium, rubidium and iron from a mica:limestone mix of weight ratio 5:2.
The roasting process was also investigated using DTA as shown in Fig. 13. The DTA profile shows two endothermic peaks. A peak be-
low 200 °C represents loss of water from the gypsum to form anhydrite. A second, smaller, endothermic peak at around 900 °C can be linked to the formation of new mineral phases. XRD profiles of the paramagnetic mica and the mica–gypsum mixtures roasted at 800 °C and 1050 °C were studied. It was observed that the XRD peaks from the original mica minerals were still present at 800 °C while at 1050 °C these had been replaced by a series of new peaks from the new mineral phases created. The new phases identified included lithium potassium sulphate (KLiSO4), uvarovite-aluminian [Ca3(Cr0.85Al0.15)2(SiO4)3] and cuspidine (Ca4Si2O7F2/3CaO 2SiO2 CaF2). The effect of the paramagnetic mica:gypsum ratio on lithium and rubidium extraction efficiency was then investigated at 1050 °C. The results are shown in Fig. 14. Increasing the mica:gypsum ratio resulted in a steady decrease in both lithium and rubidium extraction. A further experiment was carried out using the mica flotation concentrate (i.e. prior to magnetic separation. Using a mica:gypsum weight ratio of 2:1 and a roasting temperature of 1050 °C the average lithium and rubidium extractions from replicate tests were 63% and 18% respectively. This suggested a lower lithium extraction results from the non-magnetic mica fraction. This was confirmed with identical tests using the non-magnetic fraction
1601
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602
90 80
ΔT
Metal Extraction (%)
70 60 50 40 30 20 10 0
2:1
3:1
4:1
5:1
6:1
7:1
8:1
9:1
0
10:1
200
600
800
1000
1200
Temperature (ºC)
Mica:gypsum ratio Li
400
Fig. 16. DTA curves for mica–sodium sulphate mixture (2:1).
Rb
Fig. 14. Effect of mica:gypsum ratio on the water leach extraction of lithium, rubidium and iron following roasting at 1050 °C.
100
Metal Extraction (%)
90 100 90
Metal Extraction (%)
80 70 60
80 70 60 50 40 30 20 10
50
0
40
2:1
3:1
5:1
7:1
Mica: Sodium sulphate ratio
30
Li
Rb
20 Fig. 17. Effect of mica:sodium sulphate weight ratio on the water leach extraction of lithium, rubidium and iron following roasting at 850 °C.
10 0
0
200
400
600
800
1000
1200
Roasting Temperature (ºC) Li
Rb
Fe
Fig. 15. Effect of roasting temperature on the water leach extraction of lithium, rubidium and iron from a mica:sodium sulphate mix of weight ratio 2:1.
which gave average lithium and rubidium extractions of 15% and 2% respectively.
3.3. Sodium sulphate addition The influence of roasting temperature on lithium, rubidium and iron extraction efficiency, using sodium sulphate as the reactant, is shown in Fig. 15. The experiments were conducted in duplicate and the mean values are shown in Fig. 15. The mica:sodium sulphate ratio was 2:1 for the initial experiments. It can be seen from Fig. 15 that lithium extraction was approximately 90% for temperatures from 850 to 1050 °C. Rubidium extraction was highest at 1050 °C at 23% while iron co-extraction was very low at all conditions tested. The roasting process was also investigated using DTA as shown in Fig. 16. The endothermic effect at about 200 °C was due to the polymorphic transition of the sodium sulphate. The small endothermic
peak shown at 800 °C is indicative of the melting of the sodium sulphate and reaction with the mica. XRD analysis of the product of roasting at 750 °C showed peaks corresponding to the natural mica minerals. The product produced at 850 °C however showed new peaks as a result of the formation of lithium potassium sodium sulphate (Li2KNa(SO4)2), anorthite (CaAl2Si2O8), fluoroedenite (NaCa2(Mg,Fe)5Si7AlO22F2) and residual thenardite (Na2SO4). The effect of the mica:sodium sulphate weight ratio on lithium extraction is shown in Fig. 17. It can be seen that % extraction decreases as the mica:sodium sulphate ratio increases. Despite the reduction in extraction the ratio of 5:1 gives a reasonable recovery while significantly reducing the sodium sulphate requirement. Assuming formation of Li2KNa(SO4)2 a ratio of 5:1 corresponds to the stoichiometric sulphate addition required to combine with the lithium in the original mica. Rubidium extraction decreased from 17% to 7% as the ratio was increased from 2 to 7:1. Iron coextraction was negligible in all experiments.
4. Discussion The flotation of the micaceous residue produced a Li2O recovery of 98% while rejecting 41% of the feed. Further upgrading by WHIMS was less successful with 73% recovery while rejecting 50% of the flotation concentrate. This gave an overall physical sep-
1602
E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595â&#x20AC;&#x201C;1602
Table 5 Cost of bulk roasting additives from Industrial Minerals (2011) and Lines (2009). Additive
Approximate cost (ÂŁ/t)
Sodium sulphate Limestone (as used for ďŹ&#x201A;ue gas desulphurisation- FGD) Gypsum (natural) Gypsum (produced by FGD)
123 7 20 7
aration recovery of 72% while rejecting 71% of the micaceous residues. The extraction efďŹ ciency achieved with limestone was relatively poor and has been linked to the formation of eucryptite. Lithium extraction following roasting with gypsum at 1050 °C was slightly lower than that obtained with the sodium sulphate at 850 °C. Comparing these results to those obtained by Jandova and Vu (2008) and Jandova et al. (2009) with gypsum and limestone it can be observed that the sodium sulphate method gave a similar lithium extraction efďŹ ciency but at a lower roasting temperature. In this experimental work analytical grade reagents were used. Bulk commodities, with higher impurity levels, would be required on an industrial scale. Table 5 gives the approximate cost of bulk additives for the roasting process (Industrial Minerals (2011), Lines (2009)). Limestone is relatively cheap but was not very effective for the mica concentrate tested. Sodium sulphate produced the best performance but it is the most expensive additive. Synthetic gypsum from ďŹ&#x201A;ue gas desulphurisation (FGD) offers a relatively pure product (typically 95% CaSO4 2H2O) at a ďŹ ne size and relatively low cost compared to mined gypsum. One potential drawback of this material is the low level of mercury that may be found in FGD gypsum (Kairies et al., 2006). The presence of mercury in products made from this material, such as plasterboard, has caused some environmental concerns. It would be expected that mercury would be released during the roasting process and therefore scrubbing of the off-gas would be essential. This work has identiďŹ ed additives that could be used to produce water-soluble lithium compounds on roasting. The precipitation process ideally requires a lithium concentration of at least 9 g dm 3 to achieve acceptable Li2CO3 precipitation efďŹ ciency (Jandova et al. (2009)). The speciďŹ cation for lithium carbonate is >99% with low levels of alkaline impurities. From initial experiments with sodium sulphate as an additive we have produced >90% purity but more work is required to reach the industrial target. Each potential additive will generate a different range of impurities and in future work we plan to focus attention on the precipitation step. Given the annual resources available, the Li2O recovery achieved in the physical separation stage and assuming an 80% recovery to a saleable Li2CO3 product, the potential production would be 2600 t/year. This represents approximately 2.5% of current world production. Recovery of mica from some of the coarser plant residues could increase potential production considerably. 5. Conclusions The conclusions from this research can be summarised as follows: Froth ďŹ&#x201A;otation using dodecylamine upgraded the micaceous feed from 0.84 to 1.45% Li2O at a lithium recovery of 98%. Further upgrading of the ďŹ&#x201A;otation concentrate by WHIMS produced a 2.07% Li2O concentrate at a recovery of 73%. The magnetic fraction was identiďŹ ed as being predominantly zinnwaldite.
Microprobe measurements on individual zinnwaldite grains gave calculated Li2O concentrations ranging from 2.5% to 5%. The coarser grain size of the magnetic mica contained approximately 3% Li2O. Roasting of the zinnwaldite concentrate with limestone did not produce the desired lithium and rubidium extraction found by Jandova et al. (2010). There was evidence of eucryptite formation, but this lithium mineral is not very soluble in water. Roasting of the zinnwaldite concentrate with both gypsum and sodium sulphate produced maximum lithium extractions of 84% and 90% respectively. Rubidium extraction was much lower at 14% and 23% respectively. The soluble lithium species KLiSO4 and Li2KNa(SO4)2 were identiďŹ ed by XRD from the products produced after roasting with gypsum and sodium sulphate respectively. The temperature at which the water-soluble species formed could be linked to an endothermic peak from the DTA analysis. This occurred at 900 °C with the gypsum and 800 °C with the sodium sulphate. The lower cost of gypsum suggests this may be the most attractive additive despite the higher operating temperature required. Synthetic gypsum, produced by ďŹ&#x201A;ue gas desulphurisation is a relatively attractive material given its low cost and reasonable purity. Further work needs to be done to determine how impurities in the mica and in the additives inďŹ&#x201A;uence the precipitation and ultimate lithium carbonate concentration.
Acknowledgements The authors wish to thank the Goonvean Ltd., UK, for supplying the china clay waste samples and the Commonwealth Scholarship Commission for the sponsorship provided. References Alex, P., Suri, A.K., 1996. Processing of low grade zinnwaldite (lithiumâ&#x20AC;&#x201C;mica) concentrate. Light Metals, 1165â&#x20AC;&#x201C;1168. Amarante, M.M., Botelho de Sousa, A., Machado Leite, M., 1999. Processing a spodumene ore to obtain lithium concentrates for addition to glass and ceramic bodies. Minerals Engineering 12 (4), 433â&#x20AC;&#x201C;436. Bale, M.D., May, A.V., 1989. Processing of ores to produce tantalum and lithium. Minerals Engineering 2 (3), 299â&#x20AC;&#x201C;320. Brandt, F., Haus, R., 2010. New concepts for lithium mineral processing. Minerals Engineering 23 (2), 659â&#x20AC;&#x201C;661. Dresler, W., Jena, B.C., Reilly, I.G., LafďŹ n, S., Egab, E., 1998. The extraction of lithium carbonate from a pegmatite. Light Metals, The Minerals, Metals & Materials Society 12 (12), 1303â&#x20AC;&#x201C;1308. Harben, W.P., 2002. Lithium Minerals and Compounds, The Industrial Minerals Handbook,, Industrial Minerals Information Ltd., pp. 184â&#x20AC;&#x201C;192. Hawkes, J.R., Harris, P.M., DangerďŹ eld, J., Strong, G.E., Davis, A.E., Nancarrow, P.H.A, 1987. The Lithium Potential of the St. Austell granite, British Geological Survey (BGS) Report, vol. 19, no. 4, pp. 1â&#x20AC;&#x201C;54. Hykawy, J., 2010. Looking at lithium: discussing market demand for lithium in electronics. Materials World â&#x20AC;&#x201C; Ceramics and lithium 18 (5), 34â&#x20AC;&#x201C;35. Industrial Minerals, 2011. <www.indmin.com/price> (accessed 26.01.11). Jandova, J., Vu, H.N., 2008. Processing of Zinnwaldite Wastes to Obtain Lithium and Rubidium Compounds, Global Symposium on Recycling, Waste Treatment and Clean Technology, pp. 923â&#x20AC;&#x201C;930. Jandova, J., Vu, H.N., Belkova, T., Dvorak, P., Kondas, J., 2009. Obtaining Li2CO3 from zinnwaldite wastes. Ceramics â&#x20AC;&#x201C; Silikaty 53 (2), 108â&#x20AC;&#x201C;112. Jandova, J., Dvorak, P., Vu, H.N., 2010. Processing of zinnwaldite waste to obtain Li2CO3. Hydrometallurgy 103, 12â&#x20AC;&#x201C;18. Kairies, C.L., Schroeder, K.T., Cardone, C.R., 2006. Mercury in gypsum produced from ďŹ&#x201A;ue gas desulfurization. Fuel 85, 2530â&#x20AC;&#x201C;2536. Kelly, E.G., Spottiswood, D.J., 1982. Introduction to Mineral Processing. John Wiley & Sons, New York, p. 285. Lines, M., 2009. FGD: capturing mineral opportunities. Industrial Minerals (October), 64â&#x20AC;&#x201C;69. USGS, 2010. 2008 Minerals Handbook â&#x20AC;&#x201C; Lithium. USGS. Wietelmann, U., Bauer, R.J., 2008, . Ullmannâ&#x20AC;&#x2122;s Encyclopedia of Industrial Chemistry, seventh ed., vol. 20. Wiley-VCH, pp. 33-60.