Hydrometallurgy 138 (2013) 93–99

Contents lists available at SciVerse ScienceDirect

Hydrometallurgy
journal homepage: www.elsevier.com/locate/hydromet

Recovery of magnesium from Uyuni salar brine as high purity

magnesium oxalate
Khuyen Thi Tran a, Tri Van Luong a, Jeon-Woong An b, Dong-Jun Kang b, Myong-Jun Kim a, Tam Tran a,⁎
a
Department of Energy & Resources Engineering, Chonnam National University, Gwangju, South Korea
b
Technology Research Institute, Korea Resources Corp., Seoul, South Korea

a r t i c l e i n f o a b s t r a c t

Article history: Containing 10.2 Mt Li, Salar de Uyuni is known to be the richest resource of Li in the world. A high Mg/Li mass
Received 1 February 2013 ratio of 21.2:1 of the Uyuni salar brine used in this study is a significant factor hindering the effective lithium
Received in revised form 3 May 2013 recovery. Stabcal modelling was first conducted to study the conditions and chemical speciation of various
Accepted 28 May 2013
species involved in the selective precipitation of Mg and Ca oxalate. Along with the addition of oxalic acid,
Available online 27 June 2013
the effect of pH was then studied in order to determine optimal conditions to selectively remove Ca and
Keywords:
achieve high Mg yield subsequently. At an Oxalate/Ca molar ratio of 6.82:1 and pH b1, ~ 80% of Ca could be
Uyuni salar brine removed from brine without co-precipitation of Mg oxalate. A NaOH/Mg/Oxalate molar ratios of 1.95:1:1
Oxalic acid to 3.21:1:1.62 in the range of pH 3–5.5 was used for the Mg precipitation. A recovery of N95% Mg was
Selective recovery of magnesium obtained (precipitate containing mostly Mg oxalate) together with the K and Li losses of up to 35% from
Magnesium oxalate the original brine. Washing would remove Li, K contaminants and the co-precipitated sodium sulphate and
Calcium oxalate oxalate. Their absence from the final precipitate was confirmed by XRD analysis. The high purity (99.5%
grade) Mg precipitate obtained could be used as a precursor for MgO production.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction sulphur dioxide at b 50 °C and pH 1–2, then purified by precipitation

to remove most metallic impurities. Mg was recovered as Mg(OH)2 at
Magnesium and its compounds have been used in many applications pH 9.5–10.5. Elsner and Rothon (1998) extracted magnesium from a
such as insulation, agriculture, construction, in chemical and other in- magnesite ore by leaching with HNO3 or HCl. This was followed by puri-
dustries (USGS, 2012). High purity MgO is specifically used in food fication at pH 5–6 using NH4OH (or MgO) and H2O2 (or Cl2) and then
and pharmaceutical production while Mg(OH)2 is a major component precipitation to produce Mg(OH)2. The slurry obtained was thereafter
in the manufacture of fire retarding reagents (Hull et al., 2011). Accord- heated at 150–185 °C, to yield high purity Mg(OH)2. zdemir et al.
ing to the International Magnesium Association, the production of mag- (2009) produced MgCl2.6H2O (91% purity) from magnesite tailings
nesium around the world surpasses 429,000 tonnes/year and steadily (39.03% MgO) by leaching with HCl at first and then crystallization by
increases annually due to the demand of magnesium alloys in the auto- evaporation of the leach liquor at 40 °C.
mobile industry (Intlmag, 2012). Magnesium is found in minerals such Yildirim and Akarsu (2010) leached a dolomite ore containing 20.6%
as magnesite, dolomite, brucite, serpentinite, etc. and is mostly recov- MgO. After removing Ca as calcium carbonate, the purified magnesium
ered from seawater, brines and bitterns. Seawater contains about chloride solution was then treated via pyrohydrolysis to yield HCl gas
1.3 g/L Mg while the resource of magnesium-bearing brines and bit- and MgCl2.6H2O, which was thereafter decomposed to MgO and HCl.
terns around the world is estimated to be in the billions of tonnes The final MgO product had a purity of 98.86% and HCl gas was then
(USGS, 2012). High contents of magnesium have been found in salar recycled. Recently, Nduagu et al. (2012) treated a serpentinite ore
brines located in South America (An et al., 2012) and other salt lakes containing 21.8% Mg to produce high purity Mg(OH)2. Mixtures of the
around the world (Table 1). The high Mg content in Uyuni brine implies ore with ammonium sulphate at a mass ratio of 0.5–0.7 were first roasted
it is of high value which should be recovered as a co-product with Li. at 400–550 °C. Calcines were then slurried into water to obtain a
Calcined, dead-burned and fused magnesia are produced from mag- magnesium sulphate solution. After purification, Mg(OH)2 was recovered
nesite whereas chemical grade Mg(OH)2 used as fire retarding additives, at pH 10–11. During the process, NH3 gas obtained from roasting is
pharmaceuticals and food additives has been mostly produced from recycled to subsequent precipitation stages. Pyrometallurgy was also
brines and bitterns. Lalancette (1978) extracted Mg from magnesium- used to recover Mg from a hot slag (1300 °C) containing 16–20% MgO
bearing silicate. During processing, the ores were first digested with (Bodenstein and Farrell, 1985) or a magnesium alloy waste (Faure, 1995).
Robinson et al. (1943) patented a simple scheme for recovering
⁎ Corresponding author. Tel.: +82 62 530 1726; fax: +82 62 530 0462. magnesium from sea water as Mg(OH)2 using lime at pH 10.5 and
E-mail address: tamtran@jnu.ac.kr (T. Tran). 25 °C. A patented process for producing MgO from brines and bitterns

0304-386X/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.hydromet.2013.05.013
94 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99

Table 1
Composition (wt.%) of different brines (Dave and Ghosh, 2005, 2012).

Salar brine Element

Mg Ca Li Na K B Cl SO4 Mg/Li ratio

Uyuni (Bolivia) 0.65 0.05 0.035 8.75 0.72 0.02 15.69 0.85 18.6
Atacama (Chile) 0.96 0.03 0.150 7.60 1.85 0.06 16.04 1.65 6.40
Hombre Muerto (Argentina) 0.085 0.05 0.062 9.79 0.62 0.04 15.80 0.853 1.37
Cauchari (Argentina) 0.131 0.03 0.052 9.55 0.47 0.08 14.86 1.62 2.52
Rincon (Argentina) 0.303 0.06 0.033 9.46 0.66 0.04 16.06 1.02 9.29
Maricunga (Chile) 0.69 1.31 0.104 7.14 0.75 0.05 16.06 0.06 6.63
Silver Peak (USA) 0.03 0.02 0.023 6.20 0.53 0.01 10.06 0.71 1.43

was developed by Panda and Mahapatra (1983) using calcined magne- types to uptake Mg from an artificial seawater containing 0.11 g/L Mg
site, dolomite, calcite or limestone to precipitate Mg(OH)2 yielding 98% and 0.02 g/L Ca. Telzhensky et al. (2011) used nanofiltration mem-
pure MgO after drying and calcination. In another study, Martinac et al. branes DS-5 DL to accumulate Mg from seawater (1.43 g/L Mg) with a
(1996) used dolime to precipitate Mg from seawater containing 2.42 g/L recovery of 64% and a new solution containing 3.5 g/L Mg was
MgO to produce 98–99% pure Mg(OH)2. The Mg(OH)2 yielded was heat- produced.
ed at 105 °C before being sintered at 950 °C to obtain MgO. Dave and As shown in Table 1 and An et al.'s study (2012), the Mg/Li mass ratio
Ghosh (2005) used hydrated lime to precipitate Mg(OH)2 from a bittern in salar de Uyuni brine is 18–20:1 compared to ~6.4:1 and ~1.4:1 in
containing 5.25% w/v Mg at pH 7.0–7.5 as the by-product during bro- salar de Atacama and salar de Hombre Muerto brines respectively. In
mine processing. In cases where sulphate is present in the liquors, gyp- our previous study (An et al., 2012) a process was proposed to remove
sum (CaSO4.2H2O) is normally formed as a co-precipitate with Mg(OH)2 Mg from the Uyuni salar brine by precipitation using lime. Such a step
during lime addition (An et al., 2012; Karidakis et al., 2005). This produces a mixture of Mg(OH)2 and gypsum which requires further
co-precipitation would cause serious contamination of the final MgO processing to recover Mg values. The high level of Ca in the brine
product. would contaminate the Mg products if not removed first. This study
Phosphate has also been used to precipitate Mg from synthetic solu- therefore aims to recover Ca and Mg separately using oxalate and to pro-
tions (Cao et al., 2007; Golubev et al., 1999, 2001). Earlier in 1999, duce Mg oxalate as a precursor for high purity MgO production.
Golubev and co-authors reported that an amorphous precipitate phase
of Ca–Mg phosphates ((Ca,Mg)4H(PO4)3.xH2O) was detected when
Na2HPO4 was added into artificial seawater containing 0–2.48 g/L Mg 2. Experimental
and 0.4 g/L Ca. In wastewater treatment processes, phosphate was com-
monly used to remove magnesium as struvite (MgNH4PO4.6H2O) (Lee et 2.1. Materials
al., 2003; Stratful et al., 2001).
NaOH has been widely used to precipitate Mg(OH)2 from Mg- Salar de Uyuni brine (200 L) from Bolivia was used as the raw mate-
containing liquors. Lartey (1997) precipitated magnesium hydroxide rial for this study. Its composition was determined by ICP-MS (for cations)
from a bittern (4.24% Mg) using NaOH to produce MgO after calcination. and IC (for anions) as shown in Table 2. Compared to other brines quoted
The formation of nano-sized particles and poor crystallinity of Mg(OH)2 in Table 1, the raw material used in this study has a Mg/Li mass ratio of
when NaOH was used for precipitation caused difficulties during filtra- 21.2:1 much higher than those from other locations.
tion (Baird et al., 1988; Henrist et al., 2003). To solve this problem,
Alamdari et al. (2008) added Mg(OH)2 seeds (5–40 μm) during Mg pre-
cipitation from a bittern containing 30 g/L Mg to produce a product 2.2. Equipment and chemicals
having particle sizes of 5–60 μm after 1 h crystallization.
In many studies, magnesium could be recovered from mother The composition of the original brine before treatment and its com-
liquors as magnesium chloride salts. Aubry and Bichara (1979) used ponents after calcium and magnesium precipitation were analysed by
1,4-dioxane to precipitate magnesium as MgCl2.6H2O.(C2H4)O2 from a ICP-MS (Agilent A5500) and IC (Metrohm 883 Basic IC Plus). The miner-
brine containing 250–340 g/L MgCl2. After filtration, the salt was heated alogy of products obtained from precipitation was determined by X-ray
at 125 °C to liberate dioxane gas, which was recycled. Hamzaoui et al. diffraction (XRD) analysis (Cu-tube60kv 50 Ma, Phillips) whereas their
(2003) investigated a route to collect Mg from a synthetic brine constituents were also specified by X-ray Fluorescence (XRF) using an
containing 120 g/L Mg and 0.24 g/L Li using ammonium oxalate to instrument from Rigaku (ZSX Primus II). In addition, Thermal Gravimet-
form magnesium oxalate. It was indicated that up to 70% of Li was ric Analysis (TGA) and Differential Thermal Analysis (DTA) using Seiko
co-precipitated if Oxalate/Mg molar ratios of ≥0.57 were used. In a re- TG/DTA 200 were used to investigate weight changes and thermal
cent study, An et al. (2012) developed a two-stage scheme for recover- transformations of materials as a function of temperature, respectively.
ing Li from Uyuni salar brine (Bolivia) containing 16–17 g/L Mg and Hot plates were used for all experiments and all chemicals used were
0.3–3 g/L Ca. In the first stage, lime or recycled dolime (CaO.MgO) of analytical grade. Precipitants used in the study include oxalic acid
was used to remove Mg and the resultant precipitate is a mixture of (analytical grade) and sodium hydroxide (used for pH adjustment).
Mg(OH)2 and CaSO4.2H2O. Ca and the residual Mg were then removed
from the brine using Na2C2O4. Table 2
Solvent extraction and ion-exchange were also applied to recover Major components (g/kg) of the original brine and after the removal of Ca (stage 1) at
Mg from mother liquors and brines. Lee and Jackson (1970) patented Oxalate:Ca molar ratio of 6.82:1 and Mg (stage 2) at Oxalate:Mg molar ratio of 1.62:1
(Mg/Li mass ratio of the original brine is 21.2:1).
a process to extract Mg from a brine containing 1 g/L Mg using alkyl sul-
phates and sulphonates. After extraction, Mg was stripped from the Components Ca Mg Na K Li Cl SO4
mixture by a water-soluble salt such as NaCl. Later, Mackay (1977) Stage 1 Original brine 0.476 14.2 81.4 17.4 0.67 150 11
used 1-dodecylphenyl-3-trifluoromethyl-1,3-propanedione to selec- After Ca removal at pH 1 0.118 14.4 82.6 17.4 0.67 147.9 8.3
tively recover Mg from synthetic seawater containing 1.4 g/L Mg, Stage 2 Before Mg removal 0.025 14.6 86.3 16.3 0.67 160.1 20.6
0.37 g/L Ca and other elements (pH 6–9). Muraviev et al. (1996) sug- After Mg removal 0.003 0.83 67.6 10.6 0.41 121.0 13.0
at pH 4
gested a new method using carboxylic resins of acrylic and methacrylic
 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99 95

a Ca(T) b Mg(T)
MgC2O4:2H2O
0.3 12

Mg(OH)2
8.5

Mg , g/kg
0.1 CaC2O4:H2O MgCl+
Ca , g/kg
0.05 6.2 MgC2O4:2H2O
+ Mg(OH)2
Ca2+
0.02 4.5
Mg2+
0.009 CaCl2(a) 3.3
CaCl+
0 4 8 12 0 4 8 12
pH pH

c 3 C(T) Concentration (g/kg) used for Stabcal:

CaC2O4:H2O
1 - Ca : 0.476
CaC2O4:H2O - Mg: 14.2
0.3
+ Na2C2O4:H2O - Na: 84.0 (including NaOHaddition)
0.1
C, g/kg

- K: 17.4
0.03 MgC2O4:2H2O - Li: 0.67
+ CaC2O4:H2O - B: 0.70
0.01
HC2O4- - C: 1.9
0.003 - S : 3.66
0.001 C2O42- - Cl: 150
0.0003 Mg(C2O4)22-
0 4 8 12
pH
Fig. 1. Stabcal simulation at 25 °C for oxalate precipitation during Stage 1 (Oxalate/Ca molar ratio of 6.82:1) showing concentration profile vs pH for (a) Ca, (b) Mg and (c) C (C2O4)
formed in Stage 1.

a Mg(T) b Mg(T)
10 MgC2O4:2H2O + Mg(OH)2 10
MgC2O4:2H2O + Mg(OH)2
3 MgCl+ 3 MgCl+
Mg(OH)2

MgC2O4:2H2O MgC2O4:2H2O

Mg(OH)2
1 1
Mg , g/kg
Mg , g/kg

Mgequil.= 0.2 g/kg

0.3 0.3 Mgequil.= 0.2 g/kg at pH >3.3
at pH >4.7
0.1 0.1
Mg2+ Mg2+
0.03 MgSO4(a) 0.03
MgSO4(a)
0.01 0.01
Mg(C2O4)22-
Mg(C2O4)22-
0.003 0.003
0 4 8 12 0 4 8 12
pH pH

c Na(T) d Na(T)
110 Na2SO4:10H2O
120
Na2SO4:10H2O Na2C2O4:H2O + Na2SO4:10H2O
100 Na+
110
Na , g/kg

Na , g/kg

Na2C2O4:H2O
92 + Na2SO4:10H2O 99 Na+

85 90

79 81

73 73
0 4 8 12 0 4 8 12
pH pH

Fig. 2. Stabcal simulation at 25 °C for Mg oxalate and Na oxalate precipitation (Stage 2) at Oxalate/Mg molar ratios of (a, c) 1:1 and (b, d) 1.62:1. Concentration (g/kg) used:
(a, c): Ca: 0.025; Mg: 14.6; Na: 113.2 (including NaOH addition) K: 16.3; Li: 0.67; B: 0.70; Cl: 160.1; S: 6.86; C: 14.47) (b, d): Ca: 0.025; Mg: 14.6; Na: 130.2 (including NaOH
addition) K: 16.3; Li: 0.67; B: 0.70; Cl: 160.1; S: 6.86; C: 23.39).
96 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99

100 Table 3
Comparison the Mg amount (g Mg per kg brine) formed by Stabcal prediction and the
experimental data, measured as difference (Δ) between original and equilibrium Mg
80
concentrations.
% Ca Recovery

pH NaOH/Mg/Oxalate Δ[Mg]Stabcal Δ[Mg]Exp.

60
Molar ratio g/kg g/kg

40 4 1.72:1:1 14.0 12.6

4 1.86:1:1.2 14.1 14.1
3 3.16:1:1.62 14.4 13.4
20 4 3.18:1:1.62 14.4 13.8
5.5 3.21:1:1.62 14.4 14.3
0
0 60 120 180 240
were incorporated in the final precipitates. Typical compositions of the
Time (min)
brines before and after the precipitation of Ca and Mg are shown in
Fig. 3. Removal of Ca at ambient temperature during Stage 1 at Oxalate/Ca molar ratio Table 2.
of 6.82:1, using NaOH/Oxalate molar ratio of 1.41:1 for adjustment to pH 1. Reproduc-
ibility of data: ±3.5% maximum. 3.1. Stabcal modelling

2.3. Experimental techniques Fig. 1 shows the concentration-pH profiles of Ca and Mg

representing the precipitation of Ca and Mg oxalate from the brine
A weighed volume of 1 or 2 L of the brine was used in each exper- at an addition of oxalic acid equivalent to Oxalate/Ca molar ratio of
iment at ambient temperature (20–22 °C). A two-blade radial impel- 6.82:1 over the range of pH 0–12. The simulation results show that
ler set at a slow stirring rate of 200 rpm was used to suspend the Ca could be precipitated in acidic zone (pH 0–1) using oxalic acid
slurry during precipitation tests. Samples (2 mL) were taken at differ- (Fig. 1a), reaching an equilibrium concentration of b0.02 g/kg Ca
ent times to detect the steady state of the reaction. Oxalic acid was (from the original level of 0.476 g/kg). This Ca equilibrium concentra-
first added to the stirred brine and 5 M NaOH was continuously tion will be further reduced at pH N 8.5. Stabcal predicts the formation
added to adjust the pH. The total amount of NaOH added was mea- of CaC2O4.H2O in a wide range of pHs shown in Fig. 1a &c.
sured in each test to calculate the molar ratio of NaOH used with re- The precipitation of Mg oxalate (Fig. 1b) would not take place
spect to Mg and Oxalate. Different molar ratios of Oxalate/Ca and until pH N 1. At 6.82:1 Oxalate/Ca molar ratio, Mg starts precipitating
NaOH/Mg/Oxalate were tested to determine the optimal conditions at pH N 1 and its concentration is reduced to just above 12 g/kg (from
for Ca removal and Mg recovery. After precipitation, the solid was 14.2 g/kg) at pH ~ 2. Hence there is a window of pH in the range of
separated from the residual brine by vacuum filtration. The recovered pH 0–1 (Fig. 1a), when only Ca could be precipitated to b 0.02 g/kg
precipitate (15 g) was washed in 150 mL of distilled water at 20 °C. as Ca oxalate in Stage 1 and removed from the brine before Stage 2
The final washed precipitate was then dried in an oven at 100 °C be- is started with further addition of oxalic acid and NaOH (to adjust
fore being subjected to composition analysis. An accurately weighted to pH N2) for Mg oxalate precipitation. Above pH 9, Mg oxalate will
amount of 1 g of the product was dissolved in 500 mL of HNO3 (0.2% decompose to form Mg(OH)2 as shown in Fig. 1b. The free oxalate
v/v) and the resulting solutions were subjected to chemical analysis will then combine with Na to form sodium oxalate as shown in
by ICP-MS (for cations) or IC (for anions). Fig. 1c, which plots the oxalate concentration (as g/kg C) of various
species (Ca, Mg and Na oxalates) over the range of pH 0–12.
3. Results and discussion For Stage 2 of the precipitation when both oxalic acid and NaOH are
added, the Stabcal simulation of this process shows that at Oxalate/Mg
The investigation was first based on the Stabcal simulation (software molar ratio of 1:1 (Fig. 2a), the equilibrium concentration of Mg could
from Huang, 2008) of the chemical speciation and precipitation process- be reduced to ~0.2 g/kg at pH N 4.7 from the original 14.6 g/kg Mg. A fur-
es to remove Ca and Mg separately as oxalate. Precipitation experiments ther increase of Oxalate/Mg ratio past 1:1 (for example at ratio 1.62:1
were subsequently conducted to determine rate and equilibrium data. shown in Fig. 2b) would not render a higher equilibrium Mg concentra-
The concentrations of Na, K and Li in the brine were also reduced due tion (~0.2 g/kg Mg) at pH N 3.3. Mg should precipitate as MgC2O4.2H2O
to dilution (by neutralizing 5 M NaOH) although as later shown some under favourable conditions (Oxalate/Mg molar ratio of 1:1, pH N 3) as
a result. As extra sodium is added to the saturated salar brine for pH
100
100
80
% Mg Recovery

80
% Mg Recovery

60
Oxalate/Mg 60
40
molar ratio
1.00:1 40
20
pH 3
1.20:1
20 pH 4
1.62:1
0 pH 5.5
0 30 60 90 120 0
Time (min) 0 30 60 90 120
Time (min)
Fig. 4. Recovery of Mg at ambient temperature during Stage 2 at different oxalates to
Mg molar ratios at pH 4, using NaOH/Mg/oxalate molar ratios of 1.72:1:1; 1.86:1:1.2; Fig. 5. Recovery of Mg at ambient temperature at different pH's using NaOH/Mg/Oxalate
3.18:1:1.62. molar ratios of 3.16:1:1.62 (pH 3); 3.18:1:1.62 (pH 4); 3.21:1:1.62 (pH 5.5).
 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99 97

Table 4 3.3. Stage 2 — magnesium oxalate precipitation

Chemical analysis of major components (wt.%) of magnesium oxalate products. Pure
MgC2O4.2H2O should have 16.38% Mg and 59.33% C2O4.
The precipitation of Mg oxalate can be represented by the follow-
NaOH/Mg/oxalate, molar ratio; pH Mg Ca Na K Li C2O4 ing reaction:
1.95:1:1; pH 3 Before washing 17.99 0.02 0.31 0.62 0.13 61.83 2þ 2−
After washing 17.37 0.03 b0.05 0.33 0.05 62.64 Mg þ C2 O4 þ 2H2 O ➔ MgC2 O4 :2H2 O:
3.16:1:1.62; pH 3 Before washing 11.82 0.03 12.97 0.50 0.06 59.22
After washing 16.87 0.03 0.15 0.26 0.04 60.54 Fig. 4 shows the Mg concentration profile during precipitation by
3.18:1:1.62; pH 4 Before washing 12.44 0.03 9.93 0.54 0.09 59.17 first adding oxalic acid (at three Oxalate/Mg molar ratios of 1:1, 1.2:1
After washing 17.20 0.04 0.10 0.24 0.05 65.25 and 1.62:1) and then pH controlled to pH 4 by slowly adding NaOH. Un-
3.21:1:1.62; pH 5.5 Before washing 12.20 0.03 12.62 0.70 0.09 63.47 like Ca, Mg precipitation reached a steady state within 60 min. The total
After washing 16.66 0.04 0.16 0.22 0.05 65.04
NaOH added was measured and reported as NaOH/Oxalic acid molar
ratio from now on. The result indicated that almost 2:1 stoichiometry
of NaOH/Oxalic acid were used to adjust the liquor pH to pH 4. Results
adjustment, Stabcal also predicts the formation of sodium sulphate at low
indicate that a 20% excess oxalic acid is required to remove N 90% Mg.
pHs as shown in Fig. 2c & d. At 1:1 Oxalate/Mg ratio, the presence of so-
A further addition to 62% excess oxalic acid would not improve much
dium oxalate in the precipitate is also detected at pH N 4.6. As more oxa-
the Mg recovery and as a matter of fact could cause much Na oxalate
late is added (at Oxalate/Mg ratio of 1.62:1), more NaOH is also used for
to be precipitated as predicted by Stabcal (Fig. 2). The removal of Mg
neutralization leading to the formation of sodium oxalate at a higher
as measured by the difference of original and equilibrium concentra-
pH 2.1.
tions of Mg measured after 2 h is closed to those predicted by Stabcal
Stabcal simulation therefore identifies favourable conditions for
as shown in Table 3.
selective Ca removal at pH b1 (Stage 1), and no great changes of Ca
The effect of pH on the precipitation of Mg oxalate at Oxalate/Mg
equilibrium were detected for Oxalate/Ca ratios in the range from
molar ratio of 1.62:1 is shown in Fig. 5. Equilibria observed were also
5:1 to 10:1. For Stage 2, Mg would be recovered as MgC2O4.2H2O,
reached within 60 min. Results indicate that in this range of pH 3–5.5
best at pH N4.7 to render residual Mg levels in the treated salar
and at NaOH/Mg/Oxalate molar ratios of (3.16–3.21):1:1.62, the recov-
brine to b 0.2 g/kg Mg.
ery of Mg is in the range of 90–95%.

3.2. Stage 1 — selective calcium oxalate precipitation 3.4. Characterisation of magnesium oxalate precipitation

The precipitation of Ca oxalate can be represented by the reaction: During the precipitation of Mg oxalate, Na, Li and K losses were
experienced in all experiments. The compositions of different Mg ox-
alate precipitates recovered at different conditions and their final
2þ 2− products after washing are tabulated in Table 4. Sodium contamina-
Ca þ C2 O4 þ H2 O ➔ CaC2 O4 :H2 O:
tion could be mostly washed from 10–13% to 0.1%, whereas Li and K
impurities could be reduced from 0.5–0.7% K and 0.06–0.13% Li
The precipitation of calcium oxalate (as CaC2O4.H2O) by oxalic acid to b0.3% K and b 0.05% Li, respectively. As shown in Table 2, the con-
at pH 1 using NaOH (at NaOH/Oxalate molar ratio of 1.41:1, closed to centration of sulphate ion of the final liquor after precipitation
2:1 for complete neutralization) was conducted at Oxalate/Ca molar decreased, indicating the presence of sodium sulphate in the precip-
ratio of 6.82:1 in triplicate and the results of the tests were plotted in itate. The loss of chloride could be due to the incorporation of con-
Fig. 3. Although Stabcal predicts that 95% of Ca could be removed centrated NaCl in the precipitate. This salt however was easily
(from 0.476 g/kg to b0.02 g/kg Ca as shown in Fig. 1a), in practice removed during washing.
only ~80% Ca could be removed after 4 h of reaction, indicating the pre- By mass balancing the Li and K contents in both the residual liquors
cipitation was reaching equilibrium slowly. Reproducibility of the tests and samples taken, the losses of Li and K could be calculated (Fig. 6). As
was confirmed to within ±3.5% maximum. Nevertheless, as long as the shown in Fig. 6, the losses of Li and K seem to be independent of the li-
residual Ca is maintained at b 0.05 g/kg Ca after Stage 1, its incorpora- quor pH. However, a higher Li (possibly as Li oxalate) loss was experi-
tion into the Mg oxalate precipitate at Stage 2 would be less than 0.5% enced at higher Oxalate/Mg ratios (up to 35% cumulatively at 1.62:1
by weight, even if all Ca were co-precipitated with Mg in Stage 2. Oxalate/Mg ratio), although the solubility of Li oxalate is relatively

Fig. 6. Losses of Li and K from the liquors during Mg oxalate precipitation, showing dependence on Oxalate/Mg ratios but not on liquor pH.
98 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99

high at 66 g/L or 0.65 mol/L at 20 °C (Merck Millipore International, 100

2013). This loss was also observed by Hamzaoui et al. (2003) who 146.2oC
-24.3% 2
explained the intercalation of Li into the monoclinic structure of Mg ox-

DrTG(mg/min)
alate as the reason. For K, up to 35% loss was experienced, although at a 80

TGA (%)
lower Oxalate/Mg ratio (1:1), K seemed to re-equilibrate to re-dissolve 0
176.7oC 377.2oC
and liberate free oxalate for Mg to further precipitate. 60 223.4oC
Fig. 7 shows the XRD patterns of various precipitates produced at dif-
488.3oC -2
ferent conditions. At 1:1 Oxalate/Mg ratio the product produced is not -47.2%
contaminated (Fig. 7c for product before washing and Fig. 7d for product 40
after extensive water washing). However as excess oxalate was added at -4
Oxalate/Mg ratio of 1.62:1, more sodium hydroxide was needed for pH
adjustment (corresponding to NaOH/Oxalate ratio of 3.21:1.62 or 0 200 400 600
1.98:1) could cause the contamination due to the less soluble sodium Temperature (oC)
oxalate found in the final product (Fig. 7a). The contaminated sodium
oxalate was removed (Fig. 7b) after extensive washing with water. Fig. 8. TG-DTA profiles (at heating rate of 10 °C/min) of magnesium oxalate obtained
from precipitation at ambient temperature using NaOH/Mg/Oxalate molar ratio of
The carefully washed Mg oxalate products were then subjected to
1.95:1:1 after washing (at room temperature) using distilled water.
TGA–DTA analysis and the typical curve obtained is shown in Fig. 8.
The first weight loss from 176.7 °C corresponds to the reaction of
MgC2O4.2H2O releasing water to become MgC2O4 (theoretical loss is
24.29%). The average loss of water measured for 4 products yielded co-precipitation of Na2C2O4. Consequently, after washing Na2C2O4 was
at different conditions (24.1 ± 0.2)% is very closed to the theoretical mostly removed and the product obtained showed the high purity of
value and is similar to that found in a previous study (23.02%) by Mg oxalate (~99.5%) as confirmed by XRD and composition analyses.
Zhang et al. (2006). The second measured weight loss (47.1 ± 0.2)% TGA–DTA analysis showed a weight reduction of 23.81%–24.31% in the
is from 377.2 °C due to CO and CO2 liberation, corresponding to the first step, corresponding to the removal of water (24.29% theoretically)
decomposition of MgC2O4 to MgO (theoretical loss is 48.55%). Final from MgC2O4.2H2O (to form MgC2O4). Approximately 46.77–47.20%
Mg oxalate products were analysed as having 16.66–17.37% Mg, weight loss in the second step was observed, corresponding to the de-
corresponding to theoretical Mg% of 16.38% in pure MgC2O4.2H2O composition of MgC2O4 to MgO due to the loss of CO and CO2 (48.55%
product. A slightly higher Mg analysis compared to theory Mg% theoretically). The Mg oxalate produced is suitable to be used as a pre-
could be due to analytical error. cursor for the production of MgO by roasting.

4. Conclusion
Acknowledgement
Despite being the richest Li resource (10.2 Mt) Uyuni salar brine
containing a high Mg concentration (Mg/Li mass ratio ~21.2:1) is caus- This work was supported by a grant from the Energy Efficiency &
ing difficulties for Li production. The high Mg content represents a sig- Resources of the Korea Institute of Energy Technology Evaluation
nificant metal value which should be recovered with Li. Consequently, and Planning (KETEP), Ministry of Knowledge Economy, Korea (no.
Ca and Mg had to be removed from the brine by using oxalic acid before 2010T100100408).
the production of Li. At the Oxalate/Ca molar ratio of 6.82:1 and pH b 1,
~80% Ca could be precipitated from the brine without co-precipitating References
much Mg in the solid. Over 95% Mg could be obtained at an Oxalate/
Alamdari, A., Rahimpour, M.R., Esfandiari, N., Nourafkan, E., 2008. Kinetics of magnesium
Mg molar ratios of 1:1 to 1.62:1 and in the range of pH 3–5.5. Up to
hydroxide precipitation from sea bittern. Chem. Eng. Process. 47, 215–221.
35% of Li and K were lost in all experiments. At a lower Oxalate/Mg An, J.W., Kang, D.J., Tran, K.T., Kim, M.J., Lim, T., Tran, T., 2012. Recovery of lithium from
molar ratio of 1:1, K co-precipitated then seemed to re-equilibrate to Uyuni salar brine. Hydrometallurgy 117–118, 64–70.
release oxalate. The NaOH addition for pH adjustment also caused the Aubry, A., Bichara, M., 1979. Recovery of magnesium chloride from brines, US Patent
4,162,297 (24 July 1979).
Baird, T., Braterman, P.S., Cochrane, H.D., 1988. Magnesium hydroxide precipitation as
studied by gel growth methods. J. Cryst. Growth 91 (4), 610–616.
Bodenstein, J.J., Farrell, L.J., 1985. Magnesium production. US Patent 4,543,122
MgC2O4.2H2O (Comp: 28-0625)
(24 September 1985).
Na2C2O4 (Comp: 20-1149)
Cao, X., Harris, W.G., Josan, M.S., Nair, V.D., 2007. Inhibition of calcium phosphate precipitation
Database: MDI Jade 6.5
under environmentally-relevant conditions. Sci. Total. Environ. 383 (1–3), 205–215.
Dave, R.H., Ghosh, P.K., 2005. Enrichment of bromine in sea-bittern with recovery of
other marine chemicals. Indian Acad. Sci. 44 (9), 2903–2907.
Elsner, D., Rothon, R.N., 1998. Magnesium process, US Patent 5,843,389 (1 December 1998).
10000 Faure, P., 1995. Process for the recovery of magnesium from magnesium alloys waste.
US Patent 5,476,529 (19 December, 1995).
Golubev, S.V., Pokrovsky, O.S., Savenko, V.S., 1999. Unseeded precipitation of calcium
a and magnesium phosphates from modified seawater solutions. J. Cryst. Growth
Intensity

205 (3), 354–360.

5000 b Golubev, S.V., Pokrovsky, O.S., Savenko, V.S., 2001. Homogeneous precipitation of magnesium
phosphates from seawater solutions. J. Cryst. Growth 223 (4), 550–556.
c Hains, D., Reidel, F., et al., 2012. Technical report on the Salar de Maricunga lithium project
region III Chile, Technical Report NI 43-101 prepared for Li3 Energy — April 17, 2012,
d p. 1.7 (accessed on 31 January 2013, http://www.li3energy.com/projects/ni-43-101/
0 2012NI_43-101Report_LI3_Energy.pdf).
10 20 30 40 50 60 Hamzaoui, A.H., M'nif, A., Hammi, H., Rokbani, R., 2003. Contribution to the lithium
2 Theta recovery from brine. Desalination 158 (1–3), 221–224.
Henrist, C., Mathieu, J.P., Vogels, C., Rulmont, A., Cloots, R., 2003. Morphological study of
magnesium hydroxide nanoparticles precipitated in dilute aqueous solution. J. Cryst.
Fig. 7. XRD patterns of magnesium oxalate obtained from precipitation at ambient Growth 249 (1–2), 321–330.
temperature using (a) NaOH/Mg/oxalate molar ratios of 3.21:1:1.62 at pH 5.5 and Huang, H., 2008. Stabcal Software. Montana Tech. University of Montana, MT, USA.
(b) after washing with distilled water; (c) NaOH/Mg/oxalate molar ratios of 1.95:1:1 Hull, T.R., Witkowski, A., Hollingbery, L., 2011. Fire retardant action of mineral fillers.
at pH 3 and (d) after washing with distilled water (Intensity (Y) scale is for graph (d). Polym. Degrad. Stab. 96 (8), 1462–1469.
 K.T. Tran et al. / Hydrometallurgy 138 (2013) 93–99 99

Intlmag – International Magnesium Association, 2012. Accessed on 10 January 2013, Nduagu, E., Bjorklof, T., Fagerlund, J., Warna, J., Geerlings, H., Zevenhoven, R., 2012.
http://www.intlmag.org/. Production of magnesium hydroxide from magnesium silicate for the purpose
Karidakis, T., Agatzini-Leonardou, S., Neou-Syngouna, P., 2005. Removal of magnesium from of CO2 mineralisation. Part 1: application to Finnish serpentinite. Miner. Eng.
nickel laterite leach liquors by chemical precipitation using calcium hydroxide and the 30, 75–86.
potential use of the precipitate as filler material. Hydrometallurgy 76 (1–2), 105–114. zdemir, M., Çakir, D., Kipçak, İ., 2009. Magnesium recovery from magnesite tailings by
Lalancette, J.M., 1978. Recovery of magnesium from magnesium silicates. US Patent acid leaching and production of magnesium chloride hexahydrate from leaching
4,124,683 (7 November 1978). solution by evaporation. Int. J. Miner. Process. 93 (2), 209–212.
Lartey, R.B., 1997. Recovery of useful chemicals from local salt bitterns. J. Appl. Sci. Panda, J.D., Mahapatra, S.K., 1983. Process for the production of magnesium oxide from
Technol. 2 (1–2), 77–84. brine or bittern. US Patent 4,370,422 (25 January 1983).
Lee, J.M., Jackson, L., 1970. Extraction of magnesium values from brines. US Patent Robinson, H.A., Friedrich, R.E., Spencer, R.S., 1943. Magnesium hydroxide from seawater.
3,488,703 (6 January 1970). US Patent 2,405,055 (30 June 1943).
Lee, S.I., Weon, S.Y., Lee, C.W., Koopman, B., 2003. Removal of nitrogen and phosphate Stratful, I., Scrimshaw, M.D., Lester, J.N., 2001. Conditions influencing the precipitation
from wastewater by addition of bittern. Chemosphere 51 (4), 265–271. of magnesium ammonium phosphate. Water Res. 35 (17), 4191–4199.
Mackay, K.D., 1977. Process for the recovery of magnesium, US Patent 4,051,223 (27 Telzhensky, M., Birnhack, L., Lehmann, O., Windler, E., Lahav, O., 2011. Selective separation
September 1977). of seawater Mg2+ ions for use in downstream water treatment processes. Chem. Eng.
Martinac, V., Labor, M., Petric, N., 1996. Effect of TiO2, SiO2 and Al2O3 on properties of J. 175, 136–143.
sintered magnesium oxide from seawater. Mater. Chem. Phys. 46 (1), 23–30. USGS – US Geological Survey, 2012. Mineral Commodity Summaries – Magnesium.
Merck Millipore International, 2013. Accessed on 31 January 2013, http://www. USGS, Virginia, pp. 96–99 (January 2012, ISBN 978-1-4113-3349-9).
merckmillipore.com/chemicals/di-lithium-oxalate/MDA_CHEM-822085/p_MtWb. Yildirim, M., Akarsu, H., 2010. Preparation of magnesium oxide (MgO) from dolomite
s1Lzn0AAAEWjuEfVhTl. by leaching–precipitation–pyrohydrolysis process. Physicochem. Probl. Miner.
Muraviev, D., Noguerol, J., Valiente, M., 1996. Separation and concentration of calcium Process. 44, 257–272.
and magnesium from sea water by carboxylic resins with temperature-induced Zhang, J.J., Ren, N., Bai, J.H., 2006. Non-isothermal decomposition reaction kinetics of
selectivity. React. Funct. Polym. 28 (2), 111–126. the magnesium oxalate dehydrate. Chin. J. Chem. 24 (3), 360–364.