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 a b s t r a c t

Article history:Received 7 June 2011Accepted 22 August 2011Available online 16 September 2011

Keywords:Industrial mineralsFroth flotationRoastingLeachingWaste processingLithium

0892-6875/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.mineng.2011.08.013

⇑ Corresponding author. Tel.: +44 1326 371838.E-mail address: r.d.pascoe@exeter.ac.uk (R.D. Pasc

The granites of South-West England are a potential source of lithium which is generally found within themica 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 currentlydisposed of in tailings storage facilities. In this study a tailings sample containing 0.84% Li2O wasupgraded by a combination of froth flotation, using dodecylamine as the collector, and wet high intensitymagnetic 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 productswere then pulverised before being leached with water at 85 �C. Analysis of these products by XRDrevealed that the water-soluble sulphates, KLiSO4 and Li2KNa(SO4)2, were produced under specific condi-tions. 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 extractionwas very low.

Preliminary tests on the leach solution obtained by using sodium sulphate as an additive have shownthat a Li2CO3 product with a purity of >90% could be produced by precipitation with sodium carbonatealthough more work is required to reach the industrial target of >99%.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Lithium is important for a number of uses, including productionof batteries, glass and ceramics. It is also used in the production ofaluminium, preparation of greases, rubbers, alloys and pharmaceu-ticals. In 2008 lithium battery production represented 70% of thetotal rechargeable battery market (USGS, 2010) which includesmobile phones, laptop computers and power tools. The use of lith-ium for batteries has been increasing by more than 20% per year(USGS, 2010). The use of lithium-ion batteries in hybrid electricvehicles, plug-in hybrid and pure electric vehicles could see furthersignificant increases in lithium production. Forecasts indicate thatthe demand for lithium in the next 5 years is expected to increaseby approximately 60% from 102,000 t to 162,00 t of lithium car-bonate or equivalent (LCE), with batteries representing more than40,000 t of the perceived growth (Hykawy, 2010).

The primary source of lithium is from continental brines whichtypically contain 0.06–0.15% Li followed by pegmatites. The princi-pal lithium minerals from pegmatite, with their theoretical maxi-mum lithium content, are shown in Table 1.

Most of the lithium minerals from pegmatite are used for glassand ceramic production. Lithium chemicals, such as lithium car-bonate, are normally produced from brines because of the lower

ll rights reserved.

oe).

costs involved. China does however produce some lithium carbon-ate using imported spodumene as a feed stock. In a future scenarioin which brine deposits could not meet the demand for lithium car-bonate the deficit would have to be made up by increasing the useof lithium minerals from pegmatite. Zinnwaldite is one mineralthat could be exploited as a lithium source in the future. The highiron content of the mineral combined with the relatively low Li2Ocontent makes it relatively unattractive at the current time.

The processing of lithium minerals from pegmatites involvesboth comminution and physical separation techniques such asgravity concentration, froth flotation and magnetic separation(Bale and May, 1989; Amarante et al., 1999). A novel comminutiontechnique involving the application of high voltage pulses has beenshown to improve the liberation of spodumene (Brandt and Haus,2010). Once a lithium mineral concentrate has been produced it istypically roasted followed by leaching the products with eitheracid or water. This is an energy intensive chemical process. The to-tal cost of the process is significantly affected by the requirementsfor mining, fine grinding, physical separation, high temperatureroasting and evaporation. A number of lithium extraction pro-cesses have been reported for spodumene, petalite, lepidolite(Wietelmann and Bauer, 2008; Dresler et al., 1998) and zinnwal-dite concentrates (Alex and Suri, 1996; Jandova and Vu, 2008; Jan-dova et al., 2009, 2010). The method used for the extraction oflithium from zinnwaldite by Jandova and Vu (2008) and Jandovaet al. (2009, 2010) involved roasting of the concentrate with

Table 1Lithium minerals from pegmatite (from Harben (2002) and USGS (2010)).

Mineral Formula % Li2O Resource

Spodumene LiAlSi2O6 8.0 Australia, China, Canada, Zimbabwe, Portugal and FinlandPetalite LiAlSi4O10 4.9 Zimbabwe, Namibia and CanadaLepidolite K(Li,Al)3(Si,Al)4O10(OH,F)2 6.2 ZimbabweAmblygonite LiAlPO4(F,OH) 10.3 ZimbabweBikitaite LiAlSi2O6H2O 11.8 ZimbabweEucryptite LiAlSiO4 11.9 ZimbabweMontebrasite Li2O�Al2O3�2SiO2 7 CanadaJadarite LiNaSiB3O7(OH) 3.16 Serbia – at feasibility stageZinnwaldite KLiFeAl(AlSi3)O10(F�OH)2 2.5-5 No current exploitation

Coarse classification using bucket wheelde-sander

Classification using hydrocyclones

Sand to contoured tips

Mica-rich residue to tailings dam

Kaolinite-rich product for further classification and refining

Feed from mining operation

Fig. 1. Simplified flow diagram of china clay production by Goonvean Ltd.

1596 E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602

limestone and gypsum to produce a water-soluble lithium salt.Roasting with gypsum/Ca(OH)2 produced a lithium extraction ofover 90% between 900 and 980 �C. With limestone a peak leachextraction of 90% occurred at 825 �C, with a significant drop-offabove that temperature possibly caused by increased crystallinityof eucryptite (Jandova and Vu, 2008). Eucryptite normally requiresleaching with strong acid for lithium leach extraction. The mineralis known to exist as different polymorphs that have varying stabil-ity ranges depending on temperature. Co-extraction of rubidiumwas around 25% for gypsum/Ca(OH)2 and 90% for the limestoneat the temperature that gave optimum lithium extraction. Alexand Suri (1996) used pugging of zinnwaldite at 700 �C with sul-phuric acid to give lithium extraction of 90%. Unfortunately therewas significant co-extraction of both iron and aluminium underthese conditions.

The lithium potential of the St. Austell granite, situated in Corn-wall (UK) was investigated by the British Geological Survey (Haw-kes et al., 1987). An area of approximately 8 km2 � 100 m depthwas identified as containing a resource of 3.3 million tonnes of lith-ium. The china clay operations at Rostowrack and TreleavourDowns, run by Goonvean Ltd., fall within this resource area. Usingthe china clay production figures for these operations, it has beenestimated that approximately 100,000 tonnes per year of mica-ceous residues (the hydrocylcone underflow product), assayingaround 0.84% Li2O, are available for treatment. A sample of mica-ceous residues from these areas was used in this study. Additionalresource would be available by reprocessing material held in near-by tailings dams, although a lower Li2O content would be expectedas a result of mixing with residues from operations in lower Liareas. The advantage of using the hydrocylcone residue from anoperating plant is that no additional mining costs are accrued. Inaddition the minerals within the residue are typically fine and wellliberated therefore saving on the grinding costs that would resultfrom the processing of other pegmatite deposits.

Previous studies have mentioned the physical separation ofwaste 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 con-centrate, and dry magnetic separation for the separation of the var-ious mica minerals. In these investigations a limited size range wasused and no quantitative data on lithium and rubidium recoverywas presented.

In this study we have investigated the efficiency of froth flota-tion and magnetic separation for separation of both lithium andrubidium from the hydrocyclone underflow. Where possible wehave linked mineralogy with separator performance. Followingproduction of a lithium–mica concentrate the effectiveness of theroast/water leach procedure has been investigated using the re-agent systems considered by Jandova and Vu (2008) and Jandovaet al. (2009, 2010) with the addition of sodium sulphate (Na2SO4).The mineral phase changes that occur during roasting have beenfollowed using X-ray diffraction (XRD) and differential thermalanalysis (DTA).

2. Experimental

2.1. Materials

The material used in this test work was obtained from Goonv-ean Ltd., St. Austell, UK. The sample was collected from the under-flow of a group of 250 mm diameter hydrocyclones, which wereprocessing material for china clay production, mined from theTrelavour Downs and Rostowrack pits. The overflow is further pro-cessed into china clay products ready for sale. The underflow,which consists of fine mica-rich sand, is a waste product that is dis-charged to a nearby tailing dam (see Fig. 1).

The hydrocyclone underflow was further classified using a50 mm laboratory hydrocyclone operated at a pressure of276 kPa in order to remove the majority of the �10 lm fractionwhich is known to be rich in kaolinite. Fig. 2 shows the particle sizedistribution of this fraction obtained using a Malvern laser-sizer(Mastersizer MAF 5000). The de-slimed hydrocyclone underflowwas then homogenised before being riffled into 1.2 kg lots. Theselots 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 wereproduced by X-ray diffraction (XRD) using a Siemens Diffractome-ter D5000. X-ray fluorescence (XRF), using a Bruker S4 Pioneer,with the boric jacket preparation method was used for elementalanalysis of solid samples. Quantitative elemental analysis of spe-cific mineral grains was undertaken using a JEOL JAX-8200 electronmicroprobe. Atomic absorption spectrometry (AAS), using a Uni-cam SP 9 spectrometer, was used for the determination of lithium

0

10

20

30

40

50

60

70

80

90

100

1 10 100 1000

Cum

ulat

ive

% U

nder

size

Particle Size (μm)

Fig. 2. Particle size distribution of the de-slimed hydrocylone underflow.

Table 2Estimated mineralogy of the hydrocyclone underflow sample from Goonvean Ltd. clayprocessing plant.

Mineral Chemical Formula wt%

Quartz SiO2 17Muscovite KAl2 (AlSi3)O10 (F,OH)2 27Zinnwaldite KLiFeAl (AlSi3)O10 (F�OH)2 20Kaolinite Al2Si2O5 (OH)4 15K-Feldspar KAlSi3O8 19Topaz Al2SiO4 (F,OH)2 1Apatite Ca5(PO4)3(OH,F,Cl) 1Total 100

E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602 1597

and rubidium. AAS analysis was performed on a solution producedby digesting the solid sample in a mixture of perchloric and hydro-fluoric acid in a PTFE beaker, followed by evaporation until a solidresidue remained. The residue was then dissolved in a solution ofhydrochloric and boric acid, diluted and then analysed. The ele-mental analysis (from XRF and AAS) and XRD data were combinedto predict the mineral composition of the feed sample which isshown in Table 2.

In order to help understand what reactions were occurring dur-ing roasting, thermal analysis was carried out using a Stanton Red-croft STA 780. The instrument provides simultaneous thermo-gravimetric (TG) and differential thermal analysis (DTA) datawhich can be used to identify exothermic and endothermic reac-tions that occurred during the roasting stage. The DTA analysiswas 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 underflowwas initially concentrated using froth flotation. The use of wet highintensity magnetic separation (WHIMS) was then investigated as ameans of separating the individual mica minerals. A schematic ofthe overall separation process is shown in Fig. 3.

2.4. Froth flotation

Flotation was conducted in a Denver D-12 laboratory flotationmachine equipped with a 3.5 dm3 Minnovex cell. The impellerspeed and air flow rate were set at 1400 rpm and 8 dm3/minrespectively. A flotation test sample of 1.2 kg was mixed with2.8 dm3 tap water and conditioned at 1400 rpm to give 30% solidsby weight. Dodecylamine (98% purity, supplied by Sigma–Aldrich

Company Ltd., UK) was used as the cationic collector. A stock solu-tion of 1% w/v was made up and this was added to give an additionrate of 150–500 g/t for the mica flotation. The pulp was then con-ditioned for 5 min at the required pH prior to flotation for a periodof 8 min. The pH was adjusted either by addition of dilute H2SO4 orNaOH. Initial tests revealed that Fe recovery (obtained from XRFmeasurement) provided a good indicator of both lithium andrubidium recovery. As Fe analysis was easier to perform it wasused to evaluate flotation performance. The effect of collector con-centration and pH on recovery is shown in Fig. 4. Each data pointrepresents the average data from three experiments. The flotationconcentrate grade was relatively consistent for all tests at 4.5–5%Fe2O3.

Analysis of flotation results did not indicate a significant linkbetween iron recovery and grade for both pH and collector dosageover the variable range tested. In order to produce a bulk magneticproduct for further work a series of flotation tests were performedusing 500 g/t of collector at pH 2.5. The two products in this casewere analysed for lithium and rubidium by AAS as well as ironby XRF in order to calculate recoveries. The results are presentedin Table 3.

In order to determine the variation in composition by size asample of the concentrate was screened at 150, 125, 106, 90, 75,63 and 53 lm and the fractions analysed by AAS. The results areshown in Fig. 5.

It can be seen from Fig. 5 that the grade of lithium, rubidiumand iron increase as the particle size increases. The results indicatethat the zinnwaldite tends to occur in coarser fractions. Previousstudies have shown that the type of mica present in a china claysample can vary significantly with size (Hawkes et al., 1987).One other contributing factor could be that entrainment of non-mica minerals increases with decreasing particle size. The informa-tion suggests that fine screening could provide further enrichmentand that coarser residues from the processing plant (sand fractionsfrom the bucket-wheel de-sanders) could potentially contain ahigher lithium content in the mica than the hydrocyclone under-flow product.

2.5. Wet high intensity magnetic separation

The mica flotation concentrate was subjected to wet high inten-sity magnetic separation (WHIMS) using a batch Rapid MagneticLtd. separator using a matrix with a 1 mm gap to recover the highlithium/iron mica minerals. The magnetic field was adjustable upto a maximum of 2.06 Tesla. The test procedure involved repro-cessing the non-magnetic product from the first test a furthertwo times through the separator and combining the three mag-netic products. This gave the maximum recovery possible butwould be likely to adversely affect the grade. Fig. 6 shows theweight, Li2O, Rb2O and Fe2O3 recovery to the magnetic fractionas a function of magnetic field strength.

The recoveries of the three components are almost identicalsuggesting that they are all present within the same mineral. Theperformance of the separator was then determined as a functionof particle size as shown in Fig. 7. It can be seen that the massrecovery increases with increased particle size. Li2O recovery as afunction of magnetic field strength (see Fig. 8) however is similarfor each size fraction. This can be explained by the higher Li2O con-tent of the coarser fractions.

Similar trends were observed for iron and rubidium recoveriesas a function of size and magnetic field strength.

At 1.95T the Li2O content of the magnetic fraction increasedfrom 1.48% for the �38 lm fraction to 2.94% for the +150 lm frac-tion. The Fe2O3 content increased from 4.55% to 10.06% for thesame sizes. The highest Fe2O3 content in a non-magnetic fractionoccurred at the +150 lm fraction at 4.25% reducing to 1.32% in

De-slimed hydrocyclone underflow fromchina clay processing Quartz, feldspar and kaolinite

Mica Concentrate

“Non-magnetic” mica

Paramagnetic mica

Froth flotation

Wet High IntensityMagnetic Separation

Fig. 3. Mica separation flow sheet.

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600

Fe2O

3 R

ecov

ery

to C

once

ntra

te (

%)

Collector Dosage (g/t)

pH 2.5 pH 3.5 pH 5

Fig. 4. Flotation recovery of Fe2O3 as a function of collector dosage and pH.

Table 3Analysis of bulk flotation feed and products.

Product Weight (%) Content (%) Recovery (%)

Li2O Rb2O Fe2O3 Li2O Rb2O Fe2O3

Concentrate 59.3 1.45 0.55 4.47 98.6 85.2 92.8Tailings 40.7 0.03 0.14 0.51 1.4 14.8 7.2Feed 100.00 0.87 0.38 2.86 100.0 100.0 100.0

0 20 40 60 80 100 120 140 160

Com

posi

tion

(%)

Size (μm)

0

8

7

6

5

4

3

2

1

Li2O Rb

2O Fe

2O

3

Fig. 5. Variation in lithium, rubidium and iron oxides content in flotationconcentrate by particle size.

0

10

20

30

40

50

60

70

80

0 0.5 1 1.5 2 2.5

Rec

over

y to

Mag

netic

(%

)

Magnetic Field (Tesla)

WeightLi2O Rb2O Fe2O3

Fig. 6. WHIMS recovery as a function of magnetic field strength.

0

10

20

30

40

50

60

70

2.521.51

Wei

ght R

ecov

ery

(%)

Magnetic Field (Tesla)

+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 andmagnetic field strength.

1598 E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602

the �38 lm fraction. The higher Fe2O3 content produced at thecoarser sizes can be explained by the increasing contribution ofgravitational forces in the separation zone resulting in a loweringin magnetic separator efficiency (Kelly and Spottiswood, 1982).

Following these initial tests it was decided to maximise lithiumrecovery to the magnetic fraction by operating a three stage sepa-

0

10

20

30

40

50

60

70

80

90

1 1.2 1.4 1.6 1.8 2 2.2

Li 2O

Rec

over

y in

Mag

netic

(%

)

Magnetic Field (Tesla)

+150 microns +106 microns +75 microns

+53 microns +38 microns

Fig. 8. Li2O recovery by WHIMS as a function of magnetic field strength.

E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602 1599

ration at the maximum field strength of 1.95 Tesla. Approximately3 kg of paramagnetic-mica concentrate was produced by repeatingthis process a number of times with the WHIMS. Table 4 gives theelemental analysis and calculated recoveries produced by the mag-netic separation process.

Microprobe analysis indicated that two main mica mineralswere present in the magnetic separation products; zinnwalditeand muscovite. The composition of both minerals was quite vari-able but the two minerals were quite distinct when consideringSiO2 and the calculated Li2O content (see Fig. 9).

It can be seen that the muscovite contains 0–0.6% Li2O whereasthe zinnwaldite contains 2.6–5.0% Li2O. Both non-magnetic andmagnetic products include both minerals with the zinnwalditemaking up 44% of the mica grains analysed in the non-magneticfraction and 85% in the magnetic fraction.

2.6. Roasting process

In preparation for the roasting process the paramagnetic micaconcentrate was pulverised in a Tema mill with a tungsten carbidepot for 3 min before being mixed with the limestone, gypsum orsodium sulphate. Fig. 10 shows the particle size distribution ofraw and pulverised paramagnetic mica concentrate. A predeter-mined weight of pulverised paramagnetic mica concentrate (90%<100 lm) was mixed with either limestone (CaCO3 supplied byBDH Laboratory Supplies, UK, at >99% purity), gypsum (Ca-SO4�2H2O, supplied by Fisher Scientific Ltd. at >98% purity) or so-dium sulphate (Na2SO4, supplied by Hopkin and Company, UK, at>99.5% purity) and put in a ceramic crucible before being roastedin a laboratory Carbolite furnace (CWF 1200) at the selected tem-

Table 4Chemical analysis of the WHIMS products at a magnetic field of 1.95 Tesla.

Fraction wt% SiO2 Al2O3 Fe2O3 TiO2 MgO C

Content (%)Magnetic 50.3 40.1 27.8 7.4 0.27 0.11 0Non-mag 49.7 50.6 32.0 2.3 0.11 0.33 0Head 100.0 45.3 29.9 4.9 0.19 0.22 0

Recovery (%)Magnetic 50.3 44.5 46.8 76.7 71.3 25.2Non-mag 49.7 55.5 53.2 23.3 28.7 74.8Head 100.0 100 100 100 100 100

peratures for 60 min. The typical particle size of the additiveswas less than 14 lm.

In the experiments with limestone the ratio of paramagneticmica concentrate to limestone was 5:2. For both gypsum and so-dium sulphate the weight ratio was 2:1 for the roasting tempera-ture optimisation. The roasting temperatures ranged from 250 to1100 �C. Optimisation of the ratio of materials was undertakenonce the optimum roasting temperature had been identified. Thiswas done using gypsum (ratio from 2 to 10:1) and sodium sulphate(ratio from 2 to 7:1).

2.7. Leaching process

The roasted products were pulverised before leaching using3 min in the Tema mill. The typical size distribution achieved is gi-ven in Fig. 10. A 10 g sample of the pulverised product was thenleached in 100 cm3 de-ionised water in a stirred-glass reaction ves-sel placed in a water bath maintained at 85 �C. Initial experimentswith all products showed that if the lithium were soluble then10 min leaching achieved almost full extraction. The standardleaching procedure was undertaken for 30 min to ensure maxi-mum extraction. After leaching the samples were filtered, the leachsolution (filtrate) collected and the residues dried. The lithium,rubidium and iron analysis of both products were determined byAAS.

3. Results

3.1. Limestone addition

The influence of roasting temperature on lithium, rubidium andiron extraction is shown in Fig. 11. Experiments were conducted ata mica:limestone weight ratio from 2.5 to 5:1. At the 5:1 ratio usedby Jandova and Vu (2008) the lithium extraction was <1%. The re-sults shown in Fig. 12 are for a 5:2 ratio, which produced some lith-ium extraction.

With these experiments the good lithium and rubidium leachextraction achieved by Jandova and Vu (2008) could not be repli-cated. This may be due to formation of sparingly soluble, crystal-line eucryptite, which was the only lithium compound identifiedby XRD. As these results were disappointing the focus then turnedto other possible additives.

3.2. Gypsum addition

The influence of roasting temperature, using gypsum as thereactant, on lithium extraction is shown in Fig. 12. It can be seenthat highly water-soluble lithium compounds are formed whenroasting occurs at temperatures above 900 �C. A maximum lithiumextraction with water of 84% was achieved after roasting at1050 �C. Rubidium extraction was much lower at 14%. Iron co-extraction was below the detection level under all conditions.

aO K2O Na2O P2O5 F Li2O Rb2O LOI

.07 9.56 0.15 0.06 3.36 2.07 0.74 3.4

.1 6.84 0.14 0.08 1.27 0.76 0.37 6.4

.08 8.21 0.15 0.07 2.32 1.42 0.56 4.9

41.5 58.6 52.0 43.2 72.8 73.4 66.958.5 41.4 48.0 56.8 27.2 26.6 33.1100 100 100 100 100 100 100

42

43

44

45

46

47

48

49

50

51

0 1 2 3 4 5 6

Si2O

(w

t %)

Li2O (wt %)

Muscovite (found in non-magnetic)

Zinnwaldite (found in non-magnetic)Muscovite (found in magnetic)

Zinnwaldite (found in magnetic)

Fig. 9. Relationship between SiO2 and calculated Li2O content of the mica.

0102030405060708090

100

0 20 40 60 80 100 120 140 160

Cum

ulat

ive

Und

esiz

e (%

)

Particle size (μm)

magnetic mica pulverised magnetic mica

pulverised roasted product

Fig. 10. Particle size distributions of magnetic mica before and after pulverising anda typical pulverised roasted product.

0123456789

10

0 200 400 600 800 1000 1200

Met

al E

xtra

ctio

n (%

)

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.

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000 1200

Met

al E

xtra

ctio

n (%

)

Roast Temperature (ºC)

Li 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.

0 200 400 600 800 1000 1200

ΔT

Temperature (ºC)

Fig. 13. DTA profile for mica–gypsum (2:1) mixture.

1600 E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602

The roasting process was also investigated using DTA as shown inFig. 13. The DTA profile shows two endothermic peaks. A peak be-

low 200 �C represents loss of water from the gypsum to form anhy-drite. A second, smaller, endothermic peak at around 900 �C can belinked to the formation of new mineral phases.

XRD profiles of the paramagnetic mica and the mica–gypsummixtures roasted at 800 �C and 1050 �C were studied. It was ob-served that the XRD peaks from the original mica minerals werestill present at 800 �C while at 1050 �C these had been replacedby 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 cuspi-dine (Ca4Si2O7F2/3CaO�2SiO2�CaF2).

The effect of the paramagnetic mica:gypsum ratio on lithiumand rubidium extraction efficiency was then investigated at1050 �C. The results are shown in Fig. 14. Increasing the mica:gyp-sum ratio resulted in a steady decrease in both lithium and rubid-ium extraction.

A further experiment was carried out using the mica flotationconcentrate (i.e. prior to magnetic separation. Using a mica:gyp-sum weight ratio of 2:1 and a roasting temperature of 1050 �Cthe average lithium and rubidium extractions from replicate testswere 63% and 18% respectively. This suggested a lower lithiumextraction results from the non-magnetic mica fraction. This wasconfirmed with identical tests using the non-magnetic fraction

0

10

20

30

40

50

60

70

80

90

2:1 3:1 4:1 5:1 6:1 7:1 8:1 9:1 10:1

Met

al E

xtra

ctio

n (%

)

Mica:gypsum ratio

Li Rb

Fig. 14. Effect of mica:gypsum ratio on the water leach extraction of lithium,rubidium and iron following roasting at 1050 �C.

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000 1200

Met

al E

xtra

ctio

n (%

)

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.

0 200 400 600 800 1000 1200

ΔT

Temperature (ºC)

Fig. 16. DTA curves for mica–sodium sulphate mixture (2:1).

0102030405060708090

100

2:1 3:1 5:1 7:1

Met

al E

xtra

ctio

n (%

)

Mica: Sodium sulphate ratio

Li Rb

Fig. 17. Effect of mica:sodium sulphate weight ratio on the water leach extractionof lithium, rubidium and iron following roasting at 850 �C.

E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602 1601

which gave average lithium and rubidium extractions of 15% and2% respectively.

3.3. Sodium sulphate addition

The influence of roasting temperature on lithium, rubidium andiron extraction efficiency, using sodium sulphate as the reactant, isshown in Fig. 15. The experiments were conducted in duplicateand the mean values are shown in Fig. 15. The mica:sodium sul-phate ratio was 2:1 for the initial experiments. It can be seen fromFig. 15 that lithium extraction was approximately 90% for temper-atures from 850 to 1050 �C. Rubidium extraction was highest at1050 �C at 23% while iron co-extraction was very low at all condi-tions tested. The roasting process was also investigated using DTAas shown in Fig. 16.

The endothermic effect at about 200 �C was due to the polymor-phic transition of the sodium sulphate. The small endothermic

peak shown at 800 �C is indicative of the melting of the sodium sul-phate and reaction with the mica.

XRD analysis of the product of roasting at 750 �C showed peakscorresponding to the natural mica minerals. The product producedat 850 �C however showed new peaks as a result of the formationof lithium potassium sodium sulphate (Li2KNa(SO4)2), anorthite(CaAl2Si2O8), fluoroedenite (NaCa2(Mg,Fe)5Si7AlO22F2) and residualthenardite (Na2SO4).

The effect of the mica:sodium sulphate weight ratio on lithiumextraction is shown in Fig. 17. It can be seen that % extraction de-creases as the mica:sodium sulphate ratio increases. Despite thereduction in extraction the ratio of 5:1 gives a reasonable recoverywhile significantly reducing the sodium sulphate requirement.Assuming formation of Li2KNa(SO4)2 a ratio of 5:1 corresponds tothe stoichiometric sulphate addition required to combine withthe lithium in the original mica. Rubidium extraction decreasedfrom 17% to 7% as the ratio was increased from 2 to 7:1. Iron co-extraction was negligible in all experiments.

4. Discussion

The flotation of the micaceous residue produced a Li2O recoveryof 98% while rejecting 41% of the feed. Further upgrading byWHIMS was less successful with 73% recovery while rejecting50% of the flotation concentrate. This gave an overall physical sep-

Table 5Cost of bulk roasting additives from Industrial Minerals (2011) and Lines (2009).

Additive Approximate cost (£/t)

Sodium sulphate 123Limestone (as used for flue gas

desulphurisation- FGD)7

Gypsum (natural) 20Gypsum (produced by FGD) 7

1602 E. Siame, R.D. Pascoe / Minerals Engineering 24 (2011) 1595–1602

aration recovery of 72% while rejecting 71% of the micaceousresidues.

The extraction efficiency achieved with limestone was rela-tively poor and has been linked to the formation of eucryptite. Lith-ium extraction following roasting with gypsum at 1050 �C wasslightly lower than that obtained with the sodium sulphate at850 �C. Comparing these results to those obtained by Jandovaand Vu (2008) and Jandova et al. (2009) with gypsum and lime-stone it can be observed that the sodium sulphate method gave asimilar lithium extraction efficiency but at a lower roastingtemperature.

In this experimental work analytical grade reagents were used.Bulk commodities, with higher impurity levels, would be requiredon an industrial scale. Table 5 gives the approximate cost of bulkadditives for the roasting process (Industrial Minerals (2011), Lines(2009)). Limestone is relatively cheap but was not very effective forthe mica concentrate tested. Sodium sulphate produced the bestperformance but it is the most expensive additive. Synthetic gyp-sum from flue gas desulphurisation (FGD) offers a relatively pureproduct (typically 95% CaSO4�2H2O) at a fine size and relativelylow cost compared to mined gypsum. One potential drawback ofthis material is the low level of mercury that may be found inFGD gypsum (Kairies et al., 2006). The presence of mercury inproducts made from this material, such as plasterboard, has causedsome environmental concerns. It would be expected that mercurywould be released during the roasting process and therefore scrub-bing of the off-gas would be essential.

This work has identified additives that could be used to producewater-soluble lithium compounds on roasting. The precipitationprocess ideally requires a lithium concentration of at least9 g dm�3 to achieve acceptable Li2CO3 precipitation efficiency (Jan-dova et al. (2009)). The specification for lithium carbonate is >99%with low levels of alkaline impurities. From initial experimentswith sodium sulphate as an additive we have produced >90% pur-ity but more work is required to reach the industrial target. Eachpotential additive will generate a different range of impuritiesand in future work we plan to focus attention on the precipitationstep.

Given the annual resources available, the Li2O recoveryachieved in the physical separation stage and assuming an 80%recovery to a saleable Li2CO3 product, the potential productionwould be 2600 t/year. This represents approximately 2.5% of cur-rent world production. Recovery of mica from some of the coarserplant residues could increase potential production considerably.

5. Conclusions

The conclusions from this research can be summarised asfollows:

� Froth flotation using dodecylamine upgraded the micaceousfeed from 0.84 to 1.45% Li2O at a lithium recovery of 98%.� Further upgrading of the flotation concentrate by WHIMS

produced a 2.07% Li2O concentrate at a recovery of 73%. Themagnetic fraction was identified as being predominantlyzinnwaldite.

� Microprobe measurements on individual zinnwaldite grainsgave calculated Li2O concentrations ranging from 2.5% to 5%.The coarser grain size of the magnetic mica contained approxi-mately 3% Li2O.� Roasting of the zinnwaldite concentrate with limestone did not

produce the desired lithium and rubidium extraction found byJandova et al. (2010). There was evidence of eucryptite forma-tion, 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 of84% and 90% respectively. Rubidium extraction was much lowerat 14% and 23% respectively.� The soluble lithium species KLiSO4 and Li2KNa(SO4)2 were iden-

tified by XRD from the products produced after roasting withgypsum 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 thesodium sulphate.� The lower cost of gypsum suggests this may be the most attrac-

tive additive despite the higher operating temperature required.Synthetic gypsum, produced by flue gas desulphurisation is arelatively attractive material given its low cost and reasonablepurity.� Further work needs to be done to determine how impurities in

the mica and in the additives influence the precipitation andultimate lithium carbonate concentration.

Acknowledgements

The authors wish to thank the Goonvean Ltd., UK, for supplyingthe china clay waste samples and the Commonwealth ScholarshipCommission for the sponsorship provided.

References

Alex, P., Suri, A.K., 1996. Processing of low grade zinnwaldite (lithium–mica)concentrate. Light Metals, 1165–1168.

Amarante, M.M., Botelho de Sousa, A., Machado Leite, M., 1999. Processing aspodumene ore to obtain lithium concentrates for addition to glass and ceramicbodies. Minerals Engineering 12 (4), 433–436.

Bale, M.D., May, A.V., 1989. Processing of ores to produce tantalum and lithium.Minerals Engineering 2 (3), 299–320.

Brandt, F., Haus, R., 2010. New concepts for lithium mineral processing. MineralsEngineering 23 (2), 659–661.

Dresler, W., Jena, B.C., Reilly, I.G., Laffin, S., Egab, E., 1998. The extraction of lithiumcarbonate from a pegmatite. Light Metals, The Minerals, Metals & MaterialsSociety 12 (12), 1303–1308.

Harben, W.P., 2002. Lithium Minerals and Compounds, The Industrial MineralsHandbook,, Industrial Minerals Information Ltd., pp. 184–192.

Hawkes, J.R., Harris, P.M., Dangerfield, 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. 1–54.

Hykawy, J., 2010. Looking at lithium: discussing market demand for lithium inelectronics. Materials World – Ceramics and lithium 18 (5), 34–35.

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 andClean Technology, pp. 923–930.

Jandova, J., Vu, H.N., Belkova, T., Dvorak, P., Kondas, J., 2009. Obtaining Li2CO3 fromzinnwaldite wastes. Ceramics – Silikaty 53 (2), 108–112.

Jandova, J., Dvorak, P., Vu, H.N., 2010. Processing of zinnwaldite waste to obtainLi2CO3. Hydrometallurgy 103, 12–18.

Kairies, C.L., Schroeder, K.T., Cardone, C.R., 2006. Mercury in gypsum produced fromflue gas desulfurization. Fuel 85, 2530–2536.

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), 64–69.

USGS, 2010. 2008 Minerals Handbook – Lithium. USGS.Wietelmann, U., Bauer, R.J., 2008, . Ullmann’s Encyclopedia of Industrial Chemistry,

seventh ed., vol. 20. Wiley-VCH, pp. 33-60.