Hydrometallurgy 143 (2014) 1–11

Contents lists available at ScienceDirect

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

Lithium recovery from aqueous solution by sorption/desorption

Julien Lemaire a, Lenka Svecova a,b,⁎, Fanny Lagallarde a, Richard Laucournet c, Pierre-Xavier Thivel a,b
a
Laboratoire d'Electrochimie et de Physico-chimie des Matériaux et des Interfaces, UMR 5279 CNRS-Grenoble INP-UdS-UJF, 1130 rue de la piscine, 38402 Saint-Martin d'Hères, France
b
Réseau sur le Stockage Electrochimique de l'Energie (RS2E), FR CNRS 3459, France
c
CEA Grenoble, LITEN/DEHT/LPCE, 17 rue des martyrs, 38054 Grenoble, France

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

Article history: The Li-ion batteries are currently the technology of choice for the future generation of Hybrid and Electric
Received 12 July 2013 Vehicles. New generation of these batteries based on less noble materials are currently in development. That is
Received in revised form 11 October 2013 why the new approaches to their recycling are necessary and lithium recovery will become economically inter-
Accepted 4 November 2013
esting. Indeed, this study deals with lithium separation from solution and is part of a larger project dealing
Available online 11 November 2013
with the recycling of the future generation of Li-ion batteries. In this study, the sorption/desorption technique
Keywords:
was selected to evaluate lithium separation from lithium aqueous solutions. Four types of commercially available
Lithium materials were used as sorbents: Amberlite IR 120 resin, molecular sieve 13X, aluminosilicate MCM 41 and
Sorption activated carbons. Under the experimental conditions used in this study, the maximum lithium uptake ranging
Ion exchange between 20 and 25 mg Li g−1 has been reached for Amberlite IR 120 resin and molecular sieve 13X. All equilib-
Desorption rium and kinetics data have been described by a single-site ion exchange model.
Model © 2013 Published by Elsevier B.V.

1. Introduction batteries and accumulators must be properly treated and recycled. More-
over the minimum efficiency of the recycling process is given in the direc-
Even though the Li recovery from end-of-life Li-ion batteries is not tive, i.e. 50% (by weight) of Li-ion batteries should be recycled.
considered currently as an economically interesting operation, the rise Pyrometallurgical processes are traditionally used when recycling of
of electric vehicle technology could change this view point. The conse- metals is done on a large scale, but the processes are expensive, energy-
quence could be a higher pressure on the Li cost and therefore on the consuming and are not adaptable for lithium recycling (Georgi-
battery makers and an increasing price for the users (Grosjean et al., Maschler et al., 2012). Henceforth the hydrometallurgical methods of
2012). From a geopolitical point of view, there is an inherent risk re- recycling are generally privileged. Typically, the hydrometallurgical
garding the accessibility of lithium since this metal is available mainly way consists of three steps: material leaching, metals separation from
in specific geographic regions (Argentina, Bolivia and Chile) and solution and metals recovery at solid state.
an instability in these regions can greatly affect the lithium supply Battery recycling is a real environmental, scientific and technologic
(Chagnes and Pospiech, 2013; Grosjean et al., 2012). Fortunately, the challenge. At first, appropriate collecting and sorting chains are to be
situation could be improved by considering Li recycling (Tarascon, established. Next the battery discharging and dismantling should be de-
2010). Nevertheless, up to date the Li recovery remained marginal in veloped; this is rather a technological challenge as the batteries arrive in
the Li-ion battery recycling and only high valuable metals (Co, Ni) are different charging state to the dismantling facility. Further the plastics,
recovered from used Li-ion batteries (Xu et al., 2008). In rare cases solvents, ferromagnetic metals and the electrode active materials
when the lithium was indeed recovered, it was always at the very end should be separated according to Fig. 1 (Chagnes and Pospiech, 2013;
of the recovery process (Castillo et al., 2002; Contestabile et al., 1999; Xu et al., 2008). Next, the electrode active materials have to be treated
Georgi-Maschler et al., 2012; Zhang et al., 1998). However, new gener- by leaching. The obtained leachate is further treated by different mass
ation of materials for Li-ion batteries without expensive metals are cur- transfer operations in order to separate the valuable elements, which
rently being developed such as lithium iron phosphate (LFP) and are finally recovered in solid form, either as a metal or as salt.
lithium manganese oxide (LiMn2O4) and so new recycling approaches The majority of authors dealing with Li-ion battery recycling have
should be adopted in order to recover and re-use the lithium (Chagnes optimized the leaching step with the scope of cobalt recovery from
and Pospiech, 2013). And last but not least many governments have LiCoO2, the active material widely used in nomad applications Typically,
made compulsory to collect and to recycle batteries and accumulators a strong mineral acid combined (or not) with H2O2 at 50 °C–80 °C
in order to prevent environmental pollution. For example the European allows the quantitative leaching of Co in a solution (Contestabile et al.,
battery directive (2006/66/EC) lays down the requirements that all 2001; Kang et al., 2010; Mantuano et al., 2006; Provazi et al., 2011;
R.-C. Wang et al., 2009; Zhang et al., 1998). Even if the objective of
⁎ Corresponding author. Tel.: +33 4 76 82 66 91; fax: +33 4 76 82 67 77. these studies was to optimize the cobalt leaching, the authors confirm
E-mail address: [email protected] (L. Svecova). a quantitative leaching of lithium too. These treatments are quite

0304-386X/$ – see front matter © 2013 Published by Elsevier B.V.

http://dx.doi.org/10.1016/j.hydromet.2013.11.006
2 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11

Fig. 1. General flow sheet of hydrometallurgical Li-ion batteries' recycling process.

aggressive, but less harmful leaching solutions exist. The combination of available solid materials. Within this objective, a preliminary research
maleic acid or citric acid with H2O2 can also be used (Li et al., 2010a,b). focused on lithium retention from aqueous solutions was performed
After leaching, the metal recovery process is usually completed by and the results are presented in this paper. Four commercially available
separation and purification unit operation. Several mass transfer pro- solid samples were selected and tested as Li adsorbents, namely activat-
cesses are classically used to separate metals from solutions, namely: pre- ed carbons, Amberlite IR 120 exchange resin, mesostructured alumino-
cipitation (Provazi et al., 2011), liquid–liquid extraction (Flett, 2005; silicate MCM 41 and molecular sieve 13X adsorbent. The present paper
Mantuano et al., 2006; Provazi et al., 2011), sorption/desorption and ion is focused on a detailed study of the equilibrium and the kinetics of lith-
exchange (Li et al., 2012), electrowinning (Bertuol et al., 2012) or mem- ium sorption and desorption. The uptake mechanism is also investigat-
brane processes (Hajdu et al., 2012). ed and a model based on this mechanism is developed.
The state of the art dealing with lithium separation from solution is
not very vast; nevertheless some attention was already given to this 2. Materials and methods
question. The use of sorption (Chung et al., 2004, 2008; Navarrete-
Guijosa et al., 2003; Seron et al., 1996), nanofiltration (Wen et al., 2.1. Chemicals
2006) and electrolysis (Kunugi et al., 1999) were mentioned. Biological
recovery of Li using various microorganisms is also possible according to Three lithium salts – LiCl, Li2CO3 and Li2SO4 – were supplied by
Tsuruta (Tsuruta, 2005). Sigma Aldrich. Four solid materials (Sigma Aldrich) were selected
With regard to the sorption technique, both organic and inorganic according to the literature and were used as received:
materials can be used for lithium separation from different media.
– Amberlite IR 120 exchange resin, both sodium (reference Sigma
Indeed, several types of activated carbons, both chemically or physically
Aldrich 224359) and hydrogen (reference Sigma Aldrich 06428)
activated samples, were used for the uptake of Li from highly concen-
forms were used,
trated boron media (Alfarra et al., 2002; Frackowiak, 1998; Seron
– hexagonal mesostructured aluminosilicate MCM 41 (reference
et al., 1996). Navarrete-Guijosa et al. (Navarrete-Guijosa et al., 2003)
Sigma Aldrich 643 653),
studied the lithium uptake by sodium and hydrogen type of Amberlite
– molecular sieve 13X by Supelco (reference Sigma Aldrich 2-0304),
IR 120 ion exchange resin as the starting material to prepare
– activated carbons: steam activated Norit CAC 1240 W obtained from
controlled-release oral pharmaceutical formulations of this cation.
coal (reference Sigma Aldrich 37771) and chemically activated Norit
Later, the same research team (Navarrete-Casas et al., 2007) used two
CA1 obtained from wood (reference Sigma Aldrich 97876).
synthetic zeolites with the same objective. In the other domain, lithium
ion-sieves of spinel-type manganese oxide (Chung et al., 2004, 2008; Amberlite IR 120 resin is a gel type strongly acidic cation exchange
L. Wang et al., 2009; R.-C. Wang et al., 2009; Wang et al., 2006, 2008) resin. The resin is based on styrene divinylbenzene copolymer spheres
were tested in order to remove lithium from sea water. functionalized with sulfonic acid. The mean size distribution ranges
As the leaching step seems to be controlled, the present paper was from 620 to 830 μm for the hydrogen form. The advantage of this
focused on the recovery and the concentration of lithium from a resin is the range of operating pH going from 0 to 14. The moisture
leaching solution. This preliminary study constitutes an integral part content is comprised between 53 and 58%. The exchange capacity of
of a larger project dealing with hydrometallurgical recycling of used this resin is at least 1.8 eq L−1 given for wetted bed volume, which
Li-ion batteries with a special focus on lithium recovery and reuse. corresponds to 2.25 mol kg−1.
The scope of the present article is to separate lithium from its aqueous The hexagonal mesostructured aluminosilicale MCM 41 is charac-
solution. This solution is used as a model of a battery leachate, whereas terized by a linear formula (SiO2)x(Al2O3)y. It is a powder presenting
the lithium recovery is operated by a classical mass transfer method. 4.6–4.8 μm particles. The specific surface area ranges from 940 to
Therefore, the main objective of our work is to recover lithium from 1000 m2 g− 1. The pore size and volume are respectively 2.5–3 μm
an aqueous solution in order to synthesis a high purity product able and 1 cm3 g−1.
to be re-introduced in the materials manufacturing process for new The molecular sieve 13X is a synthetically produced zeolite. Its
batteries (closed-loop recycling). In this way, the selective recovery of particle size is comprised between 45–60 mesh and the pore size is
lithium from electrode active material leachate is aimed. 13 Å. The bulk density is 0.65 g cm−3.
In the scope of this study, the sorption–desorption technique was The steam activated carbon sample presented much larger particles
selected as the most promising. The first objective was to achieve a sep- (ca 1 mm) than the chemically activated sample (ca 50 μm). The steam
aration of Li and a higher Li concentration. As the state of the art is not activated carbon particles presented an important porosity, whereas in-
rich in the field of lithium sorption, it was necessary to evaluate at terestingly no porosity is observed for chemically activated carbons
first Li uptake from synthetic solutions by different commercially (under SEM observations).
 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11 3

Besides, sulfur acid, hydrochloric acid, sodium hydroxide and potas- was calculated according to Eq. (1). Comparison of experimental data
sium hydroxide were used for pH adjustments. with a sorption model (detailed in the next section) allowed an estima-
tion of the equilibrium data such as maximum sorption capacity of
2.2. Analytical method sorbents.

Lithium concentration in solutions was determined by ion chroma- 2.3.2. Kinetics study of lithium sorption and desorption
tography technique. Metrohm 881 Compact IC Pro device equipped This set of experiments was performed at three initial lithium con-
with conductivity detector was used. C4 250/4.0 column (by Metrohm) centrations: 0.01 M, 0.1 M and 1 M using LiCl as lithium source. Only
enabled the detection of lithium, sodium and four other cations (Ca2+, the molecular sieve 13X and Amberlite resin IR 120 (hydrogen from)
Mg2+, NH+ +
4 , K ). were used. Solid/liquid ratio was increased to 100 g L−1. The suspen-
sions were again stirred at 200 rpm for 24 h at ambient temperature
2.3. Experimental procedure and small volumes of liquid phase were regularly withdrawn to monitor
the decrease of lithium concentration in the aqueous phase.
Prior to the sorption experiments, it was verified that lithium salts The lithium enriched sorbents were then separated by filtration
are soluble and do not precipitate at selected pH and concentration and the desorption kinetics using diluted hydrochloric acid was stud-
range. All concentration values cited in this article are always expressed ied afterwards. The acid concentration was fixed as to obtain H+
in mg or mol of Li L−1. amount 5-fold the final amount of Li+ sorbed. The solid/liquid ratio
was fixed at 100 g L− 1 and the suspensions were again stirred at
2.3.1. Equilibrium experiments 200 rpm for 24 h. As in the previous section, small volumes were regu-
Sorption experiments were performed in duplicate with three larly withdrawn in order to monitor the increase of lithium concentra-
lithium salts and all sorbents at ambient temperature. The influence of tion in liquid phase.
initial pH on lithium uptake efficiency was studied in the pH range
from 3 up to 11. Solid amount was fixed at 20 g L−1 and the initial lith- 3. Results and discussion
ium concentration was 0.2 M (corresponding to ca 1400 mg Li·L−1).
Additional experiments were performed at initial lithium concentration 3.1. Equilibrium results
of 0.02 M with LiCl and both Amberlite forms. PH adjustments were
done with NaOH or KOH and H2SO4 or HCl solutions. The suspensions Fig. 2 summarizes the lithium uptake from 0.2 M LiCl solution
were stirred at 200 rpm for 24 h, and then final pH and lithium concen- obtained with all sorbents. No uptake was observed when activated car-
tration were measured. Lithium uptake (q) was determined from mass bons were used and so these solids are not included in Fig. 2. Alumino-
balance as the amount of lithium retained per g of solid: silicate MCM 41 showed equally poor results. The absence of uptake in
the case of activated carbons is surprising, but can be explained by a

V
q ¼ C i −C f ð1Þ high initial lithium concentration used in our study compared to a rath-
m er low sorption capacity as observed in the literature (Alfarra et al.,
2002; Frackowiak, 1998; Seron et al., 1996). In addition we have often
where Ci is the initial lithium concentration [mg L−1, mM], Cf is the final observed that the activated carbons released non-negligible amount of
(equilibrium) lithium concentration [mg L−1, mM], m is the mass of lithium during the sorption. The traces of lithium in their samples
sorbent [g] and V is the volume of solution [L]. (0.015%) have already been observed by Seron et al. (Seron et al.,
The lithium sorption yield was calculated as follows: 1996). Thus aluminosilicate and activated carbons were not used fur-

 ther in this study. For Amberlite IR 120 (both hydrogen and sodium
τsorption ð% Þ ¼ 1−C i =C f :100: ð2Þ form) and molecular sieve 13X, the lithium uptake seemed not to be in-
fluenced by initial pH and ranged between 11 and 14 mg Li g−1, which
Next to the sorption experiments, lithium enriched sorbents were corresponds to a sorption yield of 20–25%. Furthermore, no difference
separated by filtration. Diluted hydrochloric acid solutions were pre- between both Amberlite IR 120 forms was noticed.
pared for each sorbent sample so as to obtain H+ amount 1- or 5-fold Fig. 3 presents sorption yield obtained with Amberlite IR 120 resins
the amount of Li+ sorbed. Desorption experiments were then carried (both sodium and hydrogen form) and molecular sieve for the three
out at 200 rpm for 24 h, too. Lithium concentration in solution was lithium salts at 0.2 mol Li L−1. The influence of counter-anions is not
determined afterwards in order to estimate lithium desorption yield, significant for Amberlite IR 120 sodium form and the sorption yield
calculated as follows: remains ca 20%, corresponding to 14 mg g− 1 lithium uptake. In the
case of molecular sieve the results corresponding to sulfate and carbon-
C:Vacid ate are slightly higher than for chloride. For Amberlite IR 120 hydrogen
τdesorption ð% Þ ¼ :100 ð3Þ
q:m form, the sorption yields in the case of carbonate and sulfate are signif-
icantly higher in comparison to chloride. This can be explained by the
where C is the final (equilibrium) lithium concentration in HCl solution H+ consumption by carbonate or sulfate anion, that shifted the equilib-
[mg L−1, mM], q is the lithium uptake estimated previously for each rium toward the lithium sorption according to Eq. (5).
sorbent sample [mg g−1, mmol g−1], m is the mass of sorbent [g] and
þ þ þ þ
Vacid is the volume of HCl solution [L]. Liaqueous þ X sorbed ↔Lisorbed þ X aqueous : ð5Þ
Finally, the global lithium recovery yield was calculated as follows:
This phenomenon was particularly pronounced with carbonate.
C:Vacid Indeed, H+ consumption by sulfate anion (pKa ≈ 1.9) is much lower
τre covery ð% Þ ¼ :100: ð4Þ
C i :V compared to hydrogen carbonate (pKa ≈ 6.3) and carbonate
(pKa ≈ 10.3) anions. As demonstrated in Fig. 3, the sorption of lithium
Afterwards, the maximum uptake capacity of the tested materials in the presence of chloride was independent of initial pH and solid
was determined by varying the initial lithium concentration from 0.02 material used. Therefore, the following experiments were performed
to 1 mol Li L− 1 for Li2SO4, from 0.01 to 0.5 mol Li L−1 for LiCl and only with lithium chloride.
from 0.02 to 0.2 mol Li L− 1 for Li2CO3 at the same solid/liquid ratio, The desorption experiments were thus also performed only with
same contact time and stirring speed as used previously. Lithium uptake solid materials previously saturated with LiCl solution and the sorption
4 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11

Fig. 2. Influence of initial pH on lithium uptake by different sorbent materials using LiCl as lithium source (Ci = 0.2 mol Li L−1, 20 g solid L−1, 24 h).

and the desorption results as well as the recovery yields are summa- the initial concentration. This is very interesting because in spite of the
rized in Table 1. Hydrochloric acid was used as a desorption agent in low sorption yield the concentration effect can be obtained. This effect
order not to introduce other anions in the system. The desorption ex- is more pronounced at lower initial concentration values, where three
periments were performed at two different the solid/liquid ratios: times more concentrated solution can be obtained (Table 1).
25 g L− 1 in the case of molecular sieve 13X and 150 g L− 1 for the The uptake mechanism was verified by mass balance taking into
Amberlite IR 120 resin samples saturated with lithium. At the initial account both initial and final concentrations of Na+, Li+ and H+. In
concentration of 0.2 M the recovery yield was around 15%, which is the case of Amberlite IR 120 exchange resin the quantitative ion
primarily due to a low sorption yield. Indeed, desorption seemed not exchange was confirmed. In the case of molecular sieve 13X, it
to be the limiting step. The higher H+aq/Li+sorbed ratio was, the better was shown that each sorbed Li+ cation replaces a Na+ cation which
desorption yield was obtained. The desorption yield of ca 80% obtained is released into the solution. Thus the lithium uptake is due to an ion
with H+aq/Li+sorbed ratio = 5 could probably still be improved by exchange in the case of both solids. Based on the previous observations,
increasing further the H+ amount. An additional experiment was a single-site ion exchange model was developed in order to describe the
performed at 0.02 M LiCl only with both Amberlite IR 120 forms, experimental data as used in (Phongikaroon and Simpson, 2006;
while keeping the other experimental parameters fixed, and similar Shaltry et al., 2012), contrary to Navarrete-Guijosa et al. (Navarrete-
trends were observed. In this case better sorption yield (ca 60%) with Guijosa et al., 2003) who suggested the use of Langmuir-like adsorption
0.02 M lithium salt solutions led to a higher recovery yield (40%). model to describe the lithium uptake by Amberlite resin and a molecu-
Nevertheless, these results should be correlated to a lower lithium lar sieve. The used nomenclature is detailed in Table 2. The first hypoth-
uptake (ca 4.5 mg Li g−1). Interestingly, when the solid/liquid ratio of esis retained here is that the lithium concentration decrease in a
150 g L−1 was used for Amberlite IR 120 resin samples saturated with solution is only due to the retention by the solid material which releases
lithium, the final concentration of solution (after desorption) exceeded at the same time an equal amount of counter-ion of the same valence as

Fig. 3. Influence of initial pH and counter anions on lithium uptake sorption yield (Ci = 0.2 mol Li L−1, 20 g solid L−1, 24 h).
 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11 5

Table 1
Summary of sorption and desorption results (obtained with LiCl (0.2 M and 0.02 M) for molecular sieve 13X and both Amberlite IR 120 forms).

[LiCl]0 Sorbent pH τsorption qLiCl H+/Li+ ratio = 1 H+/Li+ ratio = 5

mM % mg g−1
τdesorption % τrecovery % [LiCl]f mM τdesorption % τrecovery % [LiCl]f mM
a
200 ANa 3 18 13.1 50 9 139 82 15 241
sorption: 5 19 14.2 42 8 133 74 14 224
20 g L−1 7 18 13.6 46 9 138 70 13 203
desorption: 9 17 12.9 38 7 116 85 13 216
150 g L−1 11 16 11.6 48 8 134 91 13 208
AHa 3 18 13.2 64 11 175 80 15 244
sorption: 5 18 13.4 47 9 137 73 13 215
20 g L−1 7 19 13.6 43 8 131 74 13 210
desorption: 9 17 12.3 49 8 124 80 14 226
150 g L−1 11 16 12.1 49 7 120 85 15 239
MSa 3 19 13.2 62 12 30 77 15 37
sorption: 5 19 13.0 69 13 32 80 15 38
20 g L−1 7 19 12.9 66 12 31 82 15 39
desorption: 9 20 13.8 61 12 31 79 15 39
25 g L−1 11 20 14.0 59 12 30 70 14 36

20 ANaa 3 60 4.1 27 16 24 64 38 56
sorption: 5 60 4.1 26 16 23 60 36 54
20 g L−1 7 61 4.1 27 16 24 61 38 57
desorption: 9 61 4.2 26 16 24 57 36 52
150 g L−1 11 62 4.2 26 16 24 60 36 54
AHa 3 62 4.2 25 17 25 65 43 64
sorption: 5 60 4.1 24 16 23 61 40 59
20 g L−1 7 66 4.5 25 17 25 64 43 64
desorption: 9 66 4.5 24 16 24 62 42 62
150 g L−1 11 66 4.4 24 16 24 61 41 62
a
ANa: Amberlite sodium form; AH: Amberlite hydrogen form; MS: molecular sieve.

shown in Eq. (5). In addition, the reaction order was fixed to one for all At equilibrium, the sorption rate is equal to the desorption rate.
components and their activities were considered equal to their concen- Thereby, the equilibrium concentrations in solution (C+ +
Li eq and CX eq)
tration in solution or sorbent as in (Navarrete-Guijosa et al., 2003; and in sorbents (q+Li eq and q +
X eq ) can be linked by coupling Eqs. (6)
Shaltry et al., 2012). This hypothesis could be questionable at a high ini- and (7):
tial lithium concentration. The proposed ion exchange model was so
based on Eqs. (6) and (7) right below: ksor :CLiþ eq :qX þ eq ¼ kdes :CX þ eq :qLiþ eq : ð8Þ

r sor ¼ ksor :CLiþ :qX þ ð6Þ Finally, Eq. (8) leads to the following equation for equilibrium
constant Keq:
r des ¼ kdes :CX þ :qLiþ : ð7Þ ksor CXþ eq :qLiþ eq
Keq ¼ ¼ ð9Þ
kdes CLiþ eq :qX þ eq

Table 2 with:
Nomenclature used in ion exchange model.
CLiþ eq ¼CLiþ 0 −ξeq =V ð10Þ
Variables used

rsor Sorption rate [mmol L−1 s−1]

CX þ eq ¼ξeq =V ð11Þ
ksor Sorption kinetic constant [g mmol−1 s−1],
rdes Desorption rate [mmol L−1 s−1]
kdes Desorption kinetic constant [g mmol−1 s−1] qLiþ eq ¼ξeq =m ð12Þ
rnet Net sorption rate [mmol L−1 s−1]
C+Li Lithium concentration in solution [mmol L−1] qX þ eq ¼qX þ 0 −ξeq =m: ð13Þ
C+X Concentration of exchanged cation X+ in solution [mmol L−1]
+
qLi Lithium concentration in sorbent [mmol g−1]
qX+
Concentration of exchanged cation X+ in sorbent [mmol g−1] Next, it is supposed that initially all sorption sites are occupied by the
m Mass of sorbent [g] counter-ion (H+ or Na+) and that their amount (qX+0) corresponds to
V Volume of solution [L]
the maximum sorption capacity of the solid material (qmax), estimated
ξeq Reaction progress at equilibrium [mmol]
Keq Equilibrium constant
previously. As initial lithium concentration (CLi+0) is known, ξeq and
t Time [s] all equilibrium concentrations can be calculated by solving Eq. (9).
Δt Short time interval between first solution sampling and beginning Finally, Keq is estimated by minimizing the difference between model
of experiment [s] and experimental value of equilibrium lithium uptake by means of a
qmax Maximum sorption capacity of sorbent [mmol g−1]
GRG (Generalized Reduced Gradient) nonlinear solver (Excel software).
Meaning of subscripts Afterwards, the maximum uptake capacities of the Amberlite IR 120
resin and molecular sieve 13X were investigated. The experimental
0 Initially
eq At equilibrium points were fitted with the developed ion exchange model and the pro-
t At time t posed model fits satisfactorily the experimental points (as illustrated
Li/H Exchange of Li+ for H+ in Fig. 4 on the example of molecular sieve 13X). The estimated equilib-
Li/Na Exchange of Li+ for Na+ rium data are summarized in Table 3. The equilibrium constants esti-
Na/H Exchange of Na+ for H+
mated for different lithium salts are comparable: 0.5 ± 0.2 for both
6 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11

Amberlite IR 120 forms and molecular sieve 13X. The maximum sorp- Table 3
tion capacity of sorbents corresponds to the plateau of the equilibrium Estimation of equilibrium data from experimental data using ion exchange model.

curve and the obtained values are respectively: 22 ± 1 mg g− 1 for qmax (mg Li g−1) Keq (adimensional)
Amberlite IR 120 sodium form, 20 ± 1 mg g−1 for Amberlite IR 120
LiCl Li2CO3 Li2SO4
hydrogen form and 25 ± 1 mg g−1 for molecular sieve 13X. These
Amberlite IR 120 Na 22 ± 1 0.4 ± 0.05 0.6 ± 0.05 0.35 ± 0.05
values correspond perfectly to the exchange capacities given by the
Amberlite IR 120 H 20 ± 1 0.65 ± 0.05 0.4 ± 0.05 0.5 ± 0.05
producer of Amberlite IR 120 resins and are slightly higher than those Molecular sieve 13X 25 ± 1 0.3 ± 0.05 0.4 ± 0.05 0.45 ± 0.05
obtained by Navarrete-Guijosa et al. (Navarrete-Guijosa et al., 2003).
These values are also of the same order of magnitude as those reported
for different lithium ion-sieves of spinel-type manganese oxide (Chung Sorption kinetic constant was estimated from experimental value of
et al., 2004, 2008; L. Wang et al., 2009; R.-C. Wang et al., 2009; Wang initial sorption rate r0 as follows:
et al., 2006, 2008), but these materials are not commercially available
r≈ðCLiþ 0 −CLiþ Δt Þ=Δt ð18Þ
and were synthesized in a laboratory. Their maximum sorption capaci-
ties vary from 10 to 25 mg Li g−1 at ambient temperature.
ksor ¼ rnet 0 =ðC Liþ 0 :q max Þ: ð19Þ
3.2. Kinetic study of lithium sorption and desorption
Finally, desorption kinetic constant (kdes) was estimated from an
experimental value of equilibrium constant Keq (Eq. (20) which was
Finally, the kinetics of lithium uptake and desorption were studied
calculated from a reaction progress at equilibrium ξeq (estimated with
and the experimental data were fitted with the ion exchange model.
Eq. (21)) and Eqs. (9)–(13).
In this model the concentrations in solution (C+ +
Li t and CX t) and in solids
+ +
(qLi t and qX t) were calculated as a function of time (t) using the follow- kdes ¼ ksor =Keq ð20Þ
ing equations (Eqs. (14)–(17)). The integral term was calculated with a
modified rectangle method (using variable interval).

ξ¼ CLiþ 0 −CLiþ eq :V: ð21Þ
Zt
CLiþ t ¼CLiþ 0 − r:dt with CLiþ 0 ¼CLiCl0 ð14Þ The lithium concentration decrease is plotted in Fig. 5 as a function
0 of time for molecular sieve and Amberlite IR 120 hydrogen form for
different LiCl solution (0.01 M, 0.1 M and 1 M, respectively). The sorp-
Zt tion process is rather rapid; the equilibrium was always reached
CX þ t ¼CX þ 0 þ r net :dt with CX þ 0 ¼0 ð15Þ before 100 min. As expected the higher was the initial concentration,
the higher was the reaction rate. Interestingly, the reaction rate was
0
significantly higher in the case of molecular sieve 13X. Furthermore,
the experimental data were correctly fitted by the model even at high
Zt
V Li initial concentration (1 M), although the hypothesis of concentration
qLiþ t ¼qLiþ 0 þ : r net :dt with qLiþ 0 ¼0 ð16Þ
m equal to activity was retained in the model. In the case of molecular
0
sieve13X, sodium ions were desorbed during the lithium uptake (as ob-
served previously) and this desorption could also be described by the
Zt
V proposed model satisfactorily (Eq. (15)).
qX þ t ¼qX þ 0 − : r net :dt with qX þ 0 ¼q max ð17Þ Kinetic parameters were estimated for each sorbent and each initial
m
0 concentration tested and are summarized in Table 4. Surprisingly, the
kinetic constant varied as function of the initial concentration, especially
with r net ¼ rsor −r des : in the case of molecular sieve. This variation could be explained by

Fig. 4. Determination of maximum uptake capacity of molecular sieve X13 for three lithium salts — comparison of experimental data and model (Ci = 0.02–1 mol Li L−1, 20 g solid L−1, pH
not adjusted).
 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11 7

Amberlite IR 120H+ Molecular sieve 13X

LiCl
0.01M

LiCl
0.1M

LiCl
1M

Fig. 5. Kinetics of lithium sorption on Amberlite IR 120 hydrogen form and molecular sieve 13X — comparison of experimental data and model (Ci = 0.01 to 1 mol Li L−1, 100 g solid L−1,
pH not adjusted).

several approximations of the model. Indeed, the following phenomena taken into account and ion exchange at the liquid–solid interface
were not taken into account: was considered to be instantaneous.
• Possible parasite reactions which were not comprised in the
• The activity of ions was supposed to be equal to their concentration, model.
which is only true for low concentrations. That is why the apparent ki-
netic constants decrease with increasing concentration. Nevertheless, the estimated equilibrium constants are coherent with
• Diffusion phenomena in micropores which can become limiting at the previous values in Table 3.
high concentrations, especially for molecular sieve. Indeed, the influ- For desorption of Li+ from Amberlite hydrogen form, the same kind
ence of mass transfer in the film and inside the particle was not of kinetic model was used to describe the experimental data. Sorption
8 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11

Table 4
Estimation of kinetic parameters from experimental data using ion exchange model.

ksor kdes Keq

(10−5 g mmol−1 s−1) (10−5 g mmol−1 s−1)

C+
Li AH MS AH MS AH MS

0.01 M 2.5 2.5 4.3 8.9 0.64 0.28

0.1 M 2.9 2.0 4.5 5.4 0.54 0.32
1M 1.2 0.8 4.0 2.1 0.26 0.45

AH: Amberlite hydrogen form; MS: molecular sieve.

and desorption rates of Li+ by H+ (rsor Li/H and rdes Li/H) are given right Desorption kinetic constants of Li+ with H+ and Na + with H+
below: (kdes Li/H and kdes Na/H) were estimated respectively from experimen-
tal value of initial desorption rates of Li+ and Na+ (r+ +
0 Li and r 0 Na )
r sor Li=H ¼ ksor Li=H :C Liþ :qHþ ð22Þ as follows:

r des Li=H ¼ kdes Li=H :C H þ :qLiþ ð23Þ rnet 0 Liþ ≈ðCLiþ Δt −CLiþ 0 Þ=Δt ð34Þ

For molecular sieve, the kinetic model is more complex because of rnet 0 Liþ ≈ðCLiþ Δt −CLiþ 0 Þ=Δt ð35Þ
sorption and desorption rates of Li+ by Na+ (rsor Li/Na and rdes Li/Na)
kdes ¼ rnet 0 Liþ =ðCHþ 0 :qLiþ 0 Þ ð36Þ
and Na+ by H+ (rsor Li/H and rdes Li/H): Li=H

rsor Li=Na ¼ ksor Li=Na :CLiþ :qNaþ ð24Þ kdes Na=H ¼ rnet 0 Naþ =ðCHþ 0 :qNaþ 0 Þ: ð37Þ

rdes Li=Na ¼ kdes Li=Na :CNaþ :qLiþ ð25Þ For molecular sieve, sorption and desorption kinetic constants of Li+
with Na+ (ksor Li/Na and kdes Li/Na) estimated previously in Table 4 were
rsor Na=H ¼ ksor Na=H :CNaþ :qHþ ð26Þ used in the desorption model.
In order to calculate the last unknown kinetic constants (ksor Li/H
rdes Na=H ¼ kdes Na=H :C:q: ð27Þ and ksor Na/H), the fact that, at equilibrium, net desorption rates rnet Li
and rnet Na (detailed here below) are equal to 0 was used.For Amberlite
Concentrations in solution (CLi+t, CH+t and CNa+t) and in sorbents hydrogen form
(qLi+t, qH+t and qNa+t) were calculated as a function of time (t) using
rnet Liþ ¼ rdes Li=H −rsor Li=H ð38Þ
Eqs. (28)–(31) for Amberlite hydrogen form and Eqs. (28)–(33) for mo-
lecular sieve. The integral term was calculated with the modified rectan-
rnet Hþ ¼ −rnet Liþ ð39Þ
gle method as previously.

Zt qHþ 0 ¼qmax −qLiþ 0 ð40Þ

CLiþ t ¼CLiþ 0 þ rnet Li :dt with CLiþ 0 ¼0 ð28Þ
for molecular sieve
0

rnet Liþ ¼ rdes Li=H þ rdes Li=Na −rsor Li=H −rsor Li=Na ð41Þ
Zt
V qLiþ 0 estimated from
qLiþ t ¼qLiþ 0 − : rnet Li :dt with ð29Þ ¼ rdes þ rsor
m sorption experiments rnet Naþ Na=H Li=Na −rdes Li=Na −rsor Na=H ð42Þ
0

rnet Hþ ¼ −ðrnet Liþ þ rnet Naþ Þ ð43Þ

Zt
m
CHþ t ¼CHþ 0 − rnet H :dt with CHþ 0 ¼5:qLiþ 0 : ð30Þ qHþ 0 ¼0: ð44Þ
V
0
Consequently, for Amberlite hydrogen form Eq. (38) leads to
Zt Eq. (45), and for molecular sieve Eqs. (41)–(42) lead to Eqs. (46)–(47)
V qHþ 0 defined in Eq: 40
qHþ t ¼qHþ 0 þ : rnet H :dt with ð31Þ respectively. Finally, sorption kinetic constants were estimated, accord-
m and Eq: 44
0 ing to Eqs. (45)–(47), from experimental reaction progress at equilibri-
um of Li+ and Na+ desorption (ξ+ +
eq Li and ξeq Na) calculated from
Zt Eqs. (48) and (49) respectively.
CNaþ t ¼CNaþ 0 þ rnet Na :dt with CNaþ 0 ¼0 ð32Þ
CHþ eq qLiþ eq
0 ksor Li=H ¼ kdes Li=H : : ð45Þ
CLiþ eq qHþ eq
Zt
V
qNaþ t ¼qNaþ 0 − : rnet Na :dt with qNaþ 0 ¼q max −qLiþ 0 : ð33Þ with:
m
0
CLiþ eq ¼CLiþ 0 þξeq Liþ =V
Net desorption rates (rnet) and initial H+ concentration in sorbent CHþ eq ¼CHþ 0 −ξeq Liþ =V
(qH+0) are defined in Eqs. (38)–(44) for Amberlite hydrogen form and qLiþ eq ¼qLiþ 0 −ξeq Liþ =m
molecular sieve. qHþ eq ¼qHþ 0 þξeq Liþ =m
 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11 9

Table 5
Estimation of kinetic parameters for lithium desorption from molecular sieve and Amberlite H into HCl solution.

C+
Li ksor (10−5 g mmol−1 s−1) kdes (10−5 g mmol−1 s−1) Keq

AH MS AH MS AH MS
+ + + + + + + + + + + + + +
Li /H Li /H Na /H Li /H Li /H Na /H Li /H Li+/H+ Na+/H+

0.01 M 6.7 4.2 8.5 5.8 4.6 4.0 0.9 0.9 2.1
0.1 M 3.8 1.5 1.1 5.7 3.5 2.0 0.65 0.45 0.55
1M 1.2 3.1 1.1 8.8 3.1 3.9 0.15 1.0 0.3

Amberlite IR 120 H+ Molecular sieve 13X

LiCl
0.01M

LiCl
0.1M

LiCl
1M

Fig. 6. Kinetics of lithium desorption from Amberlite IR 120 and molecular sieve 13X into HCl solution — comparison of experimental data and model (H+aq/Li+sorbed ratio = 5, 100 g solid
L−1, pH not adjusted).
10 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11

!
CHþ eq CNaþ eq qLiþ eq qNaþ eq can be obtained (concentration factor ≈ 3 at a low initial concentra-
ksor Li=H ¼ kdes Li=H : þ kdes Li=Na : : −ksor Li=Na : tion). Even if the final lithium concentration after desorption was not
CLiþ eq CLiþ eq qHþ eq qHþ eq
high enough in all studied conditions to produce only by precipitation
ð46Þ
a typical lithium salt such as Li2CO3 used of the synthesis of new lithium
mixed oxide dedicated to a new Li-ion battery manufacturing, the pre-
! cipitation of Li3PO4 less soluble than Li2CO3 could be an opportunity in
CHþ eq CLiþ eq qNaþ eq qLiþ eq the perspective of Li reuse.
ksor Na=H ¼ kdes Na=H : þ ksor Li=Na : : −kdes Li=Na :
CNaþ eq CNaþ eq qHþ eq qHþ eq The study currently continues with some selectivity experi-
ð47Þ ments and a dynamic scale-up of the experiments is in progress.
The separation of lithium from real Li-ion battery leachate present-
ing more metallic elements (Co, Ni, Mn, Na …) will also be studied.
with: The economic feasibility of the process will be evaluated at the end
of the project.
CLiþ eq ¼CLiþ 0 þξeq Liþ =V
CNaþ eq ¼CNaþ 0 þξeq Naþ =V


CHþ eq ¼CHþ 0 − ξeq Liþ þ ξeq =V References
Naþ
qLiþ eq ¼qLiþ 0 −ξeq Liþ =m Alfarra, A., Frackowiak, E., Béguin, F., 2002. Mechanism of lithium electrosorption by
qNaþ eq ¼qNaþ 0 −ξeq Naþ =m activated carbons. Electrochim. Acta 47, 1545–1553.

 Bertuol, D.A., Amado, F.D.R., Veit, H., Ferreira, J.Z., Bernardes, A.M., 2012. Recovery of
qHþ eq ¼qHþ 0 þ ξeq Liþ þ ξeq Naþ =m nickel and cobalt from spent NiMH batteries by electrowinning. Chem. Eng. Technol.
35, 2084–2092.
Castillo, S., Ansart, F., Laberty-Robert, C., Portal, J., 2002. Advances in the recovering of


spent lithium battery compounds. J. Power Sources 112, 247–254.
ξ ¼ CLiþ eq −CLiþ 0 :V ð48Þ Chagnes, A., Pospiech, B., 2013. A brief review on hydrometallurgical technologies
for recycling spent lithium-ion batteries. J. Chem. Technol. Biotechnol. 88,

 1191–1199.
Chung, K.-S., Lee, J.-C., Kim, E.-J., Lee, K.-C., Kim, Y.-S., Ooi, K., 2004. Recovery of lithium
ξ ¼ CNaþ eq −CNaþ 0 :V: ð49Þ
from seawater using nano-manganese oxide adsorbents prepared by gel process.
Mater. Sci. Forum 449–452, 277–280.
Chung, K.-S., Lee, J.-C., Kim, W.-K., Kim, S.B., Cho, K.Y., 2008. Inorganic adsorbent contain-
The kinetic parameters estimated for each sorbent and initial ing polymeric membrane reservoir for the recovery of lithium from seawater.
J. Membr. Sci. 325, 503–508.
lithium concentration are summarized in Table 5 and are compared
Contestabile, M., Panero, S., Scrosati, B., 1999. A laboratory-scale lithium battery recycling
to the experimental data in Fig. 6 where lithium concentration process. J. Power Sources 83, 75–78.
increase is represented as a function of time for molecular sieve Contestabile, M., Panero, S., Scrosati, B., 2001. A laboratory-scale lithium-ion battery
13X and Amberlite IR 120 hydrogen form with different LiCl solution recycling process. J. Power Sources 92, 65–69.
Flett, D.S., 2005. Solvent extraction in hydrometallurgy: the role of organophosphorus
(0.01 M, 0.1 M and 1 M, respectively). Desorption is rapid and extractants. J. Organomet. Chem. 690, 2426–2438.
the equilibrium is attained within 20 min. Interestingly, during the Frackowiak, E., 1998. Electrochemical polarization of activated carbons for the reversible
lithium desorption from molecular sieve 13X, the sodium ions were sorption of lithium ions. Fuel 77, 571–575.
Georgi-Maschler, T., Friedrich, B., Weyhe, R., Heegn, H., Rutz, M., 2012. Development of a
also desorbed by the acid because the solid was not completely satu- recycling process for Li-ion batteries. J. Power Sources 207, 173–182.
rated under the experimental conditions used. As shown in Fig. 6 the Grosjean, C., Miranda, P.H., Perrin, M., Poggi, P., 2012. Assessment of world lithium re-
model describes correctly the experimental data both for lithium and sources and consequences of their geographic distribution on the expected develop-
ment of the electric vehicle industry. Renew. Sust. Energ. Rev. 16, 1735–1744.
sodium desorption. Hajdu, I., Bodnár, M., Csikós, Z., Wei, S., Daróczi, L., Kovács, B., Győri, Z., Tamás, J., Borbély,
In some cases, an important variation of kinetic constants was J., 2012. Combined nano-membrane technology for removal of lead ions. J. Membr.
observed. Possible causes of this difference were already listed in the Sci. 409–410, 44–53.
Kang, J., Senanayake, G., Sohn, J., Shin, S.M., 2010. Recovery of cobalt sulfate from spent
kinetic section. Nevertheless, the constants remain in the same order
lithium ion batteries by reductive leaching and solvent extraction with Cyanex 272.
of magnitude. For the hydrogen form of Amberlite IR 120 the constants Hydrometallurgy 100, 168–171.
are comparable between the sorption and desorption experiments Kunugi, S., Yoshiyuki, I., Itoh, M., 1999. Electrochemical recovery and isotope separation of
lithium ion employing lithium ion conductive perovskite-type oxides. Solid State
(Tables 4–5).
Ionics 122, 35–39.
Li, L., Ge, J., Chen, R., Wu, F., Chen, S., Zhang, X., 2010a. Environmental friendly leaching
4. Conclusion and perspectives reagent for cobalt and lithium recovery from spent lithium-ion batteries. Waste
Manag. 30, 2615–2621.
Li, L., Ge, J., Wu, F., Chen, R., Chen, S., Wu, B., 2010b. Recovery of cobalt and lithium from
The purpose of this study was to develop a sorption/desorption spent lithium ion batteries using organic citric acid as leachant. J. Hazard. Mater. 176,
method to recover the lithium contained in an aqueous solution such 288–293.
as a leachate of used Li-ion batteries obtained during their hydrometal- Li, B., Liu, F., Wang, J., Ling, C., Li, L., Hou, P., Li, A., Bai, Z., 2012. Efficient separation and high
selectivity for nickel from cobalt-solution by a novel chelating resin: batch, column
lurgical recycling. Since limited information is actually available on lith- and competition investigation. Chem. Eng. J. 195–196, 31–39.
ium sorption by commercially available solid materials, a preliminary Mantuano, D.P., Dorella, G., Elias, R.C.A., Mansur, M.B., 2006. Analysis of a hydrometallur-
study was required and its results were presented in this paper. The gical route to recover base metals from spent rechargeable batteries by liquid–liquid
extraction with Cyanex 272. J. Power Sources 159, 1510–1518.
activated carbons and hexagonal mesostructured aluminosilicate Navarrete-Casas, R., Navarrete-Guijosa, A., Valenzuela-Calahorro, C., López-
MCM 41 revealed not to be efficient for Li uptake. Molecular sieve 13X González, J.D., García-Rodríguez, A., 2007. Study of lithium ion exchange
and Amberlite IR 120 performed approximately the same uptake rang- by two synthetic zeolites: kinetics and equilibrium. J. Colloid Interface Sci. 306,
345–353.
ing between 20 and 25 mg Li g− 1. It was further shown that after Navarrete-Guijosa, A., Navarrete-Casas, R., Valenzuela-Calahorro, C., Lopez-Gonzalez, J.D.,
the sorption step a quantitative desorption was possible from both Garcia-Rodriguez, A., 2003. Lithium adsorption by acid and sodium amberlite.
materials (confirmed by dynamic experiments which are currently J. Colloid Interface Sci. 264, 60–66.
Phongikaroon, S., Simpson, M.F., 2006. Equilibrium model for ion exchange between
in progress). The uptake mechanism in both cases was identified to
multivalent cations and zeolite-a in a molten salt. AIChE J. 52, 1736–1744.
be ion exchange. Furthermore, the sorption and desorption kinetics Provazi, K., Campos, B.A., Espinosa, D.C.R., Tenório, J.A.S., 2011. Metal separation from
were rather fast and the developed model fits correctly with the exper- mixed types of batteries using selective precipitation and liquid–liquid extraction
imental data. techniques. Waste Manag. 31, 59–64.
Seron, A., Benaddi, H., Béguin, F., Frackowiak, E., Bretelle, J.L., Thiry, M.C., Bandosz, T.J.,
The recovery of lithium from an aqueous solution is therefore possi- Jagiello, J., Schwarz, J.A., 1996. Sorption and desorption of lithium ions from activated
ble by sorption/desorption method. Moreover, a concentration effect carbons. Carbon 34, 481–487.
 J. Lemaire et al. / Hydrometallurgy 143 (2014) 1–11 11

Shaltry, M., Phongikaroon, S., Simpson, M.F., 2012. Ion exchange kinetics of fission products Wang, R.-C., Lin, Y.-C., Wu, S.-H., 2009b. A novel recovery process of metal values from the
between molten salt and zeolite-A. Microporous Mesoporous Mater. 152, 185–189. cathode active materials of the lithium-ion secondary batteries. Hydrometallurgy 99,
Tarascon, J.-M., 2010. Is lithium the new gold? Nat. Chem. 2, 510. 194–201.
Tsuruta, T., 2005. Removal and recovery of lithium using various microorganisms. J. Biosci. Wen, X., Ma, P., Zhu, C., He, Q., Deng, X., 2006. Preliminary study on recovering lithium
Bioeng. 100, 562–566. chloride from lithium-containing waters by nanofiltration. Sep. Purif. Technol. 49,
Wang, L., Ma, W., Liu, R., Li, H.Y., Meng, C.G., 2006. Correlation between Li + adsorption 230–236.
capacity and the preparation conditions of spinel lithium manganese precursor. Xu, J., Thomas, H.R., Francis, R.W., Lum, K.R., Wang, J., Liang, B., 2008. A review of processes
Solid State Ionics 177, 1421–1428. and technologies for the recycling of lithium-ion secondary batteries. J. Power
Wang, L., Meng, C.G., Han, M., Ma, W., 2008. Lithium uptake in fixed-pH solution by ion Sources 177, 512–527.
sieves. J. Colloid Interface Sci. 325, 31–40. Zhang, P., Yokoyama, T., Osamu, I., Suzuki, T.M., Inoue, K., 1998. Hydrometallurgical
Wang, L., Meng, C.G., Ma, W., 2009a. Study on Li + uptake by lithium ion-sieve via the pH process for recovery of metal values from spent lithium-ion secondary batteries.
technique. Colloids Surf. A 334, 34–39. Hydrometallurgy 47, 259–271.