Desalination 468 (2019) 114073

Contents lists available at ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Lithium recovery from desalination brines using specific ion-exchange resins T

a a a b a,⁎
Fátima Arroyo , José Morillo , José Usero , Daniel Rosado , Hicham El Bakouri
a
Department of Chemical and Environmental Engineering, University of Seville, Camino de los Descubrimientos, s/n. Isla de la Cartuja, 41092 Seville, Spain
b
Department of Chemistry and Exact Sciences, Universidad Técnica Particular de Loja, 11 01 608 Loja, Ecuador

G R A P H I C A L A B S T R A C T

A R T I C LE I N FO A B S T R A C T

Keywords: This study evaluated the possibility of recovering lithium from brines by ion-exchange procedures. Three
Lithium commercial ion-exchange resins were studied: K2629, TP207 and TP208.
Ion exchange Different tests have been carried out with artificial solutions and real brines. In addition, tests were carried
Brines out for Li elution, regeneration of the resins and reuse of the regenerated resins.
Recovery
Sorption kinetics of lithium retention onto the three resins were studied and experimental data fit to the
pseudo-second order kinetics model. Equilibrium sorption data were analysed by the Langmuir, Freundlich,
Temkin and Dubinin-Radushkevic approaches. Langmuir isotherm model best described the process. The order
of retention capacity of the amendments was TP207 > K2629 > TP208.
Recovering Li from brines was possible with ion exchange resins. In solutions containing only Li, the three
resins studied had high retention yields (> 95%). The presence of other ions in solution negatively affects the
behavior of the three resins studied.
Regarding desorption, yields obtained ranged 73.8% - 89.8%, reaching the highest (> 80%) using 4 M HCl as
eluting solution. Regenerated resins showed similar yields to those obtained when the resin is used for the first
time.

1. Introduction Li is obtained mainly from two natural sources: the spodumene

mineral, double silicate of Al and Li (LiAISi2O6) associated with quartz,
The demand of Li has greatly increased in the last decades due to its mica and feldspar, and the natural brines of the salt lakes and geysers,
applications in large-capacity rechargeable batteries, light aircraft al- mainly in the form of double sulphates of Li and K (KLiSO4) and other
loys, cosmetics and pharmaceuticals applications. Furthermore, it is salts. Currently, brines provide the highest production at the lowest cost
expected to continue growing in the future because of new applications, [2–4] by evaporation of water and the subsequent precipitation of some
such as fuel for nuclear fusion [1]. lithium compound.

⁎
Corresponding author.
E-mail addresses: [email protected] (F. Arroyo), [email protected] (J. Morillo), [email protected] (J. Usero), [email protected] (D. Rosado),
[email protected] (H. El Bakouri).

https://doi.org/10.1016/j.desal.2019.114073
Received 18 March 2019; Received in revised form 6 July 2019; Accepted 10 July 2019
0011-9164/ © 2019 Elsevier B.V. All rights reserved.
F. Arroyo, et al. Desalination 468 (2019) 114073

If seawater is not considered, existing reserves of Li will be in- element of great importance such as sodium have been studied. With
sufficient to cover the demand for this metal in the future. So, research these experimental data, It has been determined the load kinetics that is
is beginning to focus on extracting Li directly from seawater, where it is adjusted to the lithium retention with each type of resin, the equili-
estimated at approximately 2.5·1014 kg, or through brines generated in brium isotherms that best define the retention of lithium by the resins
seawater desalination plants with reverse osmosis technology. The and a mathematical model has been developed that allows predict li-
latter is interesting because of the large number of existing plants, their thium retention in brines with different concentrations of Li and Na.
foreseeable growth and the fact that brine is more concentrated than The optimal conditions for elution of lithium and regeneration of resins
seawater. This technology allows obtaining large quantities of Li and have also been studied. Summing up, this work deepens in the knowl-
other valuable metals and, thus, helps the economic profitability of edge about the retention of lithium from cationic and low cost resins
desalination brine management, which generates a significant en- from brines.
vironmental impact in the areas surrounding its discharge.
The incorporation of selective extraction technologies to the current 2. Materials and methods
desalination systems is challenging due to the low concentrations, the
limited selectivity of the extractants and the complexity of the brine 2.1. Introduction to the experimental design
matrix. For this purpose, materials with high operating capacity and
selectivity are required to avoid competition within alkaline and alka- In this research, 5 different types of experiments were performed to
line earth cations maintaining low operating costs. The main existing analyze the capacity of three type of resins to capture Li in synthetic
technologies for this purpose are: liquid-liquid extraction, ion ex- and real brines: kinetic studies, equilibrium studies, the elution and
change, leaching, precipitation, adsorption and electrochemical pro- reutilization, effect of Na content studies and Li retention from real
cesses. At present, the processes for extracting Li from seawater and brines. Experiments, together with their operating conditions are de-
desalination brines are mainly based on membrane technologies, ad- tailed in Table 2.
sorption and ion exchange, although research on liquid-liquid extrac- In each test, 50 mL of synthetic brines (containing Li and/or Na) and
tion was carried out decades ago. Initially, the behavior of zirconium resin were placed in 100 mL flasks and maintained in contact using a
phosphate (Zr3(PO4)4) was investigated and currently it is intended to rotary shaker at 25 °C. Different synthetic brines/resin ratios, and dif-
extract the Li through various compounds of manganese oxide (MnO2), ferent Li and/or Na contents were varied throughout the study. After
a substance with a very high selectivity towards Li. The Lithium re- the contact, resin and solution (raffinate) were separated using a
covery from brines by membrane and/or electrochemical processes are membrane filter (0.45 μm) and the resins were washed twice with
relatively novel technologies becoming very important [5–11]. distilled water. Raffinates were analysed for Li content using Li reten-
The most studied processes to recover Li from seawater require tion efficiency (R) and capacity (Q) that were calculated according to
manufacturing a solid material morphologically modified with a Li Eq. (1) and Eq. (2). The loaded resin and eluting solution were con-
solution that creates the active sites with greater selectivity for the Li. tacted as described in sorption stage.
Research found has been divided according to the starting material with
(C0 − Ce )
which the solid is manufactured: manganese oxides, aluminum com- R (%) = ·100
C0 (1)
pounds, exchange resins and other solids (Table 1).
mg ⎞ (C0 − Ce )·V
- Recovery through contact with modified resins. Several patents Q ⎛⎜ ⎟ =

have been found, dealing with the modification of anion exchange ⎝ g ⎠ W (2)
−1
resins by contact with aluminum compounds. It involves the im- where C0 and Ce (mg L ) are the Li content in the solution initially and
pregnation of an ion exchange resin with AlCl3 and ammonia. After at equilibrium, respectively, V (L) is the volume of solution, and W (g) is
the impregnation, it is contacted with a Li halide (LiX) to form a Li the resin mass. The elements in the solution were analysed using an
aluminate, and subsequently heated until LiX·2 Al(OH)3 is formed. ICP-OES Agilent 5100.
Then, the LiX is removed from the solid and the modified resin al- Synthetic brines were used and prepared by dissolving weighed
lows the selective Li recovery in the presence of other elements such amount of LiCl and NaCl in ultrapure water, i.e. type 1 water (Merck
as Na, Ca, Mg, K and/or B. There are no substantial differences Millipore, Germany). All other used chemicals were of analytical grade.
between the following patents, having even the same name [28–33]. A real sea-water reverse osmosis brine collected in a desalination plant
- Recovery through contact with commercial resins. Few studies located in Almería (Spain) was also used. Its composition is described in
on the recovery of lithium by conventional ion exchange resins have Table 3. The compositions of the different synthetic solutions are shown
been published [37]. in Table 2.
Resins Lewatit K2629, TP 207 and TP 208 were obtained from
Most of these processes have been designed for natural brines, and Lanxess (Germany). Its characteristics are given in Table 4. Lewatit TP
there are not many studies on metal recovery from industrial effluents. 207 and TP 208 are macroporous cation exchange weakly acidic resins
In this work, the recovery of lithium from synthetic solutions and from with chelating groups for the selective removal of metal cations from
real brines the brine of a seawater desalination plant using reverse solutions. Compared to Lewatit TP 207, TP 208 resin has a hetero-
osmosis have been studied. With this aim, three low cost cationic ion dispersed bead size, a modified polymer structure and a modified
exchange resins have been used. The influences of the contact time, the substitution grade of the imino-diacetate groups. Lewatit K2629 is a,
amount of lithium in the brines and the presence of an interfering macroporous, polymer-based strongly acidic resin.

Table 1
Technologies for the recovery of Li by adsorption in morphologically altered solid materials.
Lithium adsorption sorbents References

Sorbents based on manganese oxide ([12]; [13,14]; [15]; [16]; [17]; [18]; [19]; [20])
Sorbents based on activated alumina and aluminum compounds ([21]; [22–24]; [25]; [26]; [27])
Modified resins ([28–32]; [33])
Sorbents based on zirconium phosphate from zirconium oxide ([34]; [35]; [36])
Commercial resins [37]

2
F. Arroyo, et al. Desalination 468 (2019) 114073

Table 2
Experimental design.
Resin mass Brine volume Brine content (mg L−1) Contact time Temperature (°C) Elution
(g) (ml) (min)
Li Na Eluting Time (h) Temperature (°C)
solution

Kinetics studies 1 50 10 0 1–1440 25 – – –

Equilibrium studies 1 50 0.5–50 0 60 25 – – –
Elution and reutilization 1 50 10 0 60 25 HCl 3 M, 4 M 24 25
Effect of Na content 5–100 100–10,000 1440 25 – – –
Retention from real brines 1, 10, 20 50 0.6 15,300 1440 25 – – –

Table 3 (Qeexp − Qecal )2

χ2 =
Composition and properties of sea-water reverse osmosis brine Qecal (6)
collected in Almería (Spain) desalination plant.
Property Value
2.3. Equilibrium studies
Conductivity at 25 °C 73 mS cm−1
pH 7.84 Equilibrium experiments were conducted by the batch method.
Alcalinity 445 mg CaCO3 L−1 Kinetic studies indicated that 30 min of equilibration were sufficient to
Chloride 28.7 g L−1
attain the equilibrium.
Sulfate 4.60 g SO4 L−1
Sodium 15.3 g L−1 The retention yields (R) and the loading capacities (Qe) at equili-
Lithium 0.60 mg L−1 brium were calculated according to Eq. (1) and Eq. (2), respectively.
Results obtained were fitted to Langmuir, Freundlich, Temkin and
Dubinin-Radushkevic models of equilibrium in ion-exchange. Isotherm
2.2. Kinetics studies models represent the equilibrium distribution of Li between the solid
and liquid phases and illustrate the type of interaction between the Li
The kinetics tests were carried out by a batch technique according to ion and the resin in each case. The isotherm model can determine the
the method described in Section 2.1. The obtained results were ana- maximum sorption capacity and several thermodynamic parameters
lysed using the pseudo-first and pseudo-second order kinetic equations. that can be used for a better understanding of the sorption system.
Pseudo-first-order kinetic model fits when sorption is driven by mass Langmuir model describes quantitatively the formation of a mono-
concentration gradient. The pseudo-second-order rate reaction is layer adsorbate on the outer surface of the solid, and after that, no
mostly concerned with the amount of metal on the adsorbent's surface further retention takes place [41] and it describes the adsorption onto a
at equilibrium [38]. The differential and integral forms of both models surface containing a finite number of identical active sites. Langmuir
are shown in Eqs. (3) and (4). represented the Eq. (7).
dQt 1 1 1
= k1 ∙ (Qe − Qt ) log(Qe − Qt ) = log Qe − k1 ∙t = +
dt (3) Qe QM KL QM Ce (7)

dQt t 1 1 where Qe is the equilibrium amount of Li retained per unit resin mass in
= k2 ∙ (Qe − Qt )2 = + ∙t
dt Qt k2 ∙Qe 2 Qe (4) mg g−1, QM (mg g−1) is the maximum capacity and KL (L mg−1),
named Langmuir isotherm constant, and is related to the energy of
where Qt and Qe are the amount of Li retained by each resin at any time adsorption and to the affinity between Li and resin. The equilibrium
t and at the equilibrium time respectively, calculated according to Eq. dimensionless constant, RL (Eq. (8)), is related to separation factor or
(2). k1 (min−1) and k2 (g mg−1 min−1) are the equilibrium rate con- equilibrium parameter [42] and indicates the retention nature (un-
stants of the pseudo-first-order and pseudo-second-order kinetics favorable if RL > 1, linear if RL = 1, or favorable if 0 < RL < 1).
models respectively. If the retention rate is defined as H = Qt/t [39],
1
when t approaches 0, H (meq g−1 h−1) is the initial Li retention yield RL =
(Eq. (5)). 1 + (KL + C0 ) (8)
Freundlich model is commonly used to describe the adsorption onto
H = k2 ∙Qe 2 (5)
heterogeneous surface [43]. These data often fit the empirical equation
The integral forms of both models were used for the analysis of the proposed by Freundlich (Eq. (9)):
kinetics and characteristic parameters determination [40]. Based on the
Qe = KF Ce1/ Fr (9)
results, the orders of reaction were determined. The linear plots of the
−1
models were made and k1 and k2 were obtained from the slopes. The where KF is Freundlich isotherm constant (mg g ) is referred to the
values of k1 and k2 and the experimental and estimated values of Qe are adsorption capacity. The exponent Fr is adsorption intensity, and 1/Fr
given with the determination coefficients of each model. is a function of the strength of adsorption in the adsorption process
The suitability and precision of the kinetic models were tested by [44]. If Fr = 1 then the partition between solid and liquid phases is
the chi-square (χ2) (Eq. (6)). independent of the concentration. If value of 1/Fr < 1, it indicates a

Table 4
Properties of the resins Lewatit K2629, TP207 and TP 208 [data from Lanxess].
Resin Matrix Effective size Active group Ionic form Bulk density (g L−1) Capacity (eq/L)

Lewatit K2629 Crosslinked polyestirene 0.55 mm Sulfonic acid H+ 760 1.7

Lewatit TP 207 Imino-diacetatic acid Na+ 720 2.2
Lewatit TP 208 Na+ 740 1.8

3
F. Arroyo, et al. Desalination 468 (2019) 114073

normal adsorption, and 1/Fr > 1 indicates cooperative adsorption

[45]. If 1/Fr lies between one and ten, this indicates a favorable sorp-
tion process [46].
The Temkin Isotherm (Eq. (10)) contains a factor assessing ad-
sorbent–adsorbate interactions. The model assumes that heat of ad-
sorption of all molecules in the layer would decrease linearly with
coverage [47]. It is characterized by a uniform distribution of binding
energies that was carried out by plotting the quantity sorpted Qe
against lnCe and the constants were determined from the slope and
intercept [48].
RT
Qe = ln(AT Ce )
bt (10)
where AT is the Temkin isotherm equilibrium binding constant
(L g−1), bT is the Temkin constant, R is the universal gas constant
Fig. 1. Central composite rotatable design model.
(8.314 J mol−1 K−1) and T is defined equal to 298 K.
Dubinin–Radushkevich isotherm model (Eq. (11)) is generally ap-
plied to express the adsorption mechanism with a Gaussian energy where Li and Na coded contents are the “normalized” Li and Na con-
distribution onto a heterogeneous surface [49,50]. tents in coded terms (adimensional), respectively. The actual Li and Na
content of solutions are expressed in mg L−1.
Qe = Qs exp(−K ad ε 2) (11)
The experimental data were fitted by a polynomial equation (re-
−1
where Qs is the theoretical isotherm saturation capacity (mg g ), Kad sponse surface) in order to determine the relationships between factors
is the Dubinin–Radushkevich isotherm constant (mol2 kJ−2) and ε is the and the response variables. The predicted responses in the model are
Dubinin–Radushkevich isotherm constant (Eq. (12)). retention and capacity. First or third order polynomial equations are
suggested too. Description is analogous. The quality of fit polynomial
1⎞
ε = RT ln ⎛1 +
⎜ ⎟ model was expressed by R2 and the F-test model examined its statistical
⎝ Ce ⎠ (12) significance. The response surface mathematical models predicted the
optimum Li retention conditions with each resins in solutions con-
2.4. Elution and reutilization studies taining Li and Na.

Loading efficiency was calculated by mass balance (eq. 13), and 3. Results and discussion
eluting efficiency was calculated according the following expression:
3.1. Kinetic studies. Effect of contact time
Cs ∙Vs
SLi = ∙100
(C0 − Ce ) ∙VR (13)
Experimental Li loading capacity of each resin was plotted as a
where cs is the Li content in the stripping solution after elution, Vs is function of contact time and it was observed that the loading capacity
the stripping solution volume (L), VR is the working solution volume increased with time in all cases (Fig. 2). Two loading zones for the three
(L), co and ce are the initial and equilibrium Li content of solutions (in resins are appreciated, a fast one (for the first 5 min), and then a very
loading step), in mg L−1. slow one (up to 20 min). Increasing time from 5 to 20 min, the retention
For reutilization experiments, resins were loaded and eluted under increased only from 82.5% to 97.9% for TP 208, from 96% to 98% for
the experimental conditions showed in Table 2. After the elution, resins TP207 and from 80% to 95% for K2629. The first loading zone could
were carefully washed with water and each type of resin was divided in correspond to the stages of surface retention and diffusion within the
two identical parts, conditioned according to provider: one half to H pores of the resin, and would be very fast thanks to the small size of the
form (using HCl) and the other to Na form (using NaCl). Once re- Li. In the second charging zone, the controlling mechanism could be the
generated, the conditioned-resins were contacted with working solu- diffusion of Li to the resins, due to the low concentration of Li in so-
tions under the conditions described in Table 2. New sample of each lution after the first 10 min of contact.
resin were included in reutilization tests for comparative purposes. The coefficients of determination (R2) obtained for the pseudo-first
order model are lower than those obtained for the second order-models,
2.5. Effect of Na content studies indicating that the pseudo-second order kinetic equation described
better the sorption process (Table 5). The high values of R2 indicated
The response surface methodology (RSM) technique based on a five- that the models described successfully the Li retention onto the studied
level, two-variable central composite rotatable design (CCRD) was resins. Moreover, the experimental capacity values (Qe-exp) are in
employed. The Stat-Ease Design-Expert (version 9.0.6.2) software was agreement with the calculated values of sorption capacities from the
used for experimental design and response surface design and analysis. second-order kinetic equation (Qe-cal).
The space of interest was defined by the following ranges: Na initial Values of χ2 are low for the three resins: 7.8·10−3 for K2629,
concentration: 100–15,000 (mg L−1), and Li initial concentration: 3.5·10−3 for TP207 and 9.5·10−4 for TP208. So the pseudo- second
0.5–100 (mg L−1). The CCRD consisted of a total of 13 experiments for order kinetic model determined in this study can be considerate sui-
each resin (Fig. 1). Table SI-1 shows the design matrix. For statistical table and precise. In all cases equilibrium was achieved after 30 min, so
calculations experimental conditions are expressed in coded terms. The the Li content at 30 min were used for k1 and k2 determination. As it is
coded values of actual parameters could determinate by Eqs. (14) and shown in Table 5, the fastest retention of Li is attributed to TP 208 (the
(15). highest rate constant), and the slowest for K2629. Compared to TP 207
Li actual content − 50 and TP 208, K2629 showed lower initial rate (H) of ion exchange be-
Li coded content = tween resin and aqueous solution, which in agreement with k2 values.
30 (14)
The theoretical capacity at equilibrium sequence is TP
Na actual content − 5000 207 > K2629 > TP 208, but the values are quite similar, which is in
Na coded content =
3500 (15) agreement with the experimental values of capacity. The sequence of

4
F. Arroyo, et al. Desalination 468 (2019) 114073

100 1

90 0.9

80 0.8

Lithium retention yield (%)

70 0.7

Capacity at equilirbium (mg·g-1)

60 0.6

50 0.5

40 0.4

30 0.3

20 R-TP207 R-TP208 R-K2629 0.2

10 Q-TP207 Q-K2629 Q-TP208 0.1

0 0
0 5 10 15 20 25 30
Contact time (min)

Fig. 2. Li retention yields and capacities at equilibrium during the first 30 min of contact for the resins TP 207, TP 208 and K2629.

theoretical capacities is quite similar to the order of capacities reported factor (RL) is > 0 but < 1 in all experiments, indicating that equili-
by the manufacturer. brium sorption is favorable for the 3 resins. For the 3 resins, RL de-
Taking into account both criteria (R2 and χ2), results indicated that creased when Li content increased, so the retention is favored at high Li
pseudo-second order was more suitable for the Li retention onto K2629, contents. If the 3 resins are compared, TP 207 show a less RL. The
TP 207 and TP 208 resins. The better fit of the pseudo-second-order maximum monolayer coverage capacity (QM) from Langmuir model
suggest that resins-Li main interaction is controlled by chemical sorp- was achieved by TP 207 (2.536 mg g−1). The maximum KL is also
tion, and the retention rate depends more on the availability of the achieved with the resin TP 207. Therefore attending the Langmuir
active sites than the concentration of Lithium in solution [51–53]. parameters, TP 207 is the resin showing more affinity with Li ions (the
lowest RL and the highest KL and QM).
Regarding the Freundlich constants (Table SI- 2), as the 3 resins
3.2. Equilibrium studies. Effect of initial Li concentration shown values of 1/Fr < 1 (0.559–0.657), the sorption of Li+ into 3
studied resins are favorable. The smaller value of (1/Fr) for resin K2629
The sorption data obtained were fitted to the different equilibrium indicated a stronger bond between Li and resin, but the values are quite
models and the obtained linearized equations are summarized in Table similar.
SI-2. Due to the determination factors of three resins values and the
If the three resins are compared, for low concentrations of Li characteristics of resins (homogeneous sites), Langmuir model is pre-
(< 5 mg L−1), the distribution coefficient is much higher for the case of ferred for describing the interaction between Li and resins TP 207, TP
resin K2629 than in the case of resins TP 207 and TP 208, which show 208 and K2629. Therefore, the retention correspond to a mono-
values similar to the distribution coefficient. When the concentration of molecular layer of Li+ onto homogeneous sites within resin surfaces.
Li is higher than 10 mg L−1, the distribution coefficient is similar for the The relative position of the lines on the graph (Fig. 3) is an in-
three resins. Therefore, if the solution has a low Li content, the K2629 dication of the retention capacity of Li onto each resin. The lower the
will retain more Li than the TP 207 and TP 208 resins. From 10 mg Li curve on the Y-axis, the more retained Li per unit mass of resin. It can
L−1, the behavior of the three resins will be similar. be seen that the behavior of the TP 207 and TP 208 resins are very
Isotherm constants of Langmuir, Freundlich, Temkin and Dubinin- similar to each other, and slightly worse than that of the K2629 resin.
Radushkevic were obtained from the slopes and intercepts of linearized This difference is accentuated at lower Ce, as was determined by ana-
form of models. Constants and values of determination coefficients are lyzing the value of Kd.
shown in Table SI-2 and Table 6.
Although all models represent satisfactorily the adsorption process
(showing high determination coefficients), it seems that the Langmuir 3.3. Elution and reutilization studies
and Freundlich isotherms (R2 > 0.946) give a better fit than Temkin
and Dubinin-Radushkevic models (R2 < 0.919) for the three resins. So Samples of 50 mL of synthetic solution with 10 mg L−1 Li were
an analysis for both Langmuir and Freundlich models were performed contacted with 1 g of resin, as described in previous sections. After
for comparative reasons. contact, resin and solutions were filtered and resin was contacted with
From the data calculated in Table SI-2, the Langmuir separation 50 mL of HCl, 3 M and 4 M. The elution of lithium was 89.8% for the

Table 5
Model parameters of ion-exchange kinetic models for Li retention onto TP 207, TP 208 and K2629 resins.
Resin Qe-exp (mg/100 g) First order Second order

Qe-cal (mg/100 g) k1 (min−1) R2 Qe-cal (mg/100 g) k2 (g mg−1·min−1) H = k2Qe2 R2

K2629 48.95 83.91 0.0762 0.9596 52.72 0.0600 208 0.9943

TP 207 48.95 31.70 0.1311 0.9826 53.04 0.1249 460 0.9997
TP 208 49.10 7.43 0.0892 0.7768 50.67 0.2168 844 0.9989

R2: coefficients of determination; Qe-exp: experimental capacity values; Qe-cal: calculated capacity values; H: initial rate of ion-exchange; k1 and k2 are the first and
second order kinetics constants.

5
F. Arroyo, et al. Desalination 468 (2019) 114073

Table 6
Parameters from equilibrium models.
Langmuir model Freundlich model Dubinin-Radushkevic model Temkin model

2 2 2
QM KL R KF Fr R Qs KAD R AT bT R2
(mg g−1) (L mg−1) (mg g−1) (mol2 kJ−2) (L m−1 g−1)

K2629 1.84 0.108 0.997 0.936 1.79 0.968 1.40 0.168 0.896 42.5 5714 0.862
TP 208 1.23 0.325 0.990 0.702 1.68 0.946 1.29 0.240 0.849 19.7 4628 0.882
TP 207 2.54 2.434 0.988 0.828 1.52 0.968 1.17 0.232 0.854 21.2 5692 0.919

RL: Langmuir separation factor; QM: maximum monolayer coverage capacity from Langmuir model; Fr: Freundlich isotherm constant; Qs: theoretical isotherm
saturation capacity; KAD: Dubinin–Radushkevich isotherm constant; AT: Temkin isotherm equilibrium binding constant; bT: Temkin constant.

resin K2629, 80.4% for TP 207 resin and over 84.9% for TP 208 resin Table 7
(Table 7). The best option to recuperate the Li retained in the resins is to Elution tests experimental results.
employ 4 M HCl for TP 207 and K2629, but for TP 208 the elution yield Retention yield K2629 TP 207 TP 208
is slightly higher using HCl 3 M than 4 M. Both molarities are higher
than indicated by the manufacturer. After elution, resins were condi- 98.0% 98.0% 98.4% 98.4% 98.8% 98.8%
tioned according to the manufacturer to H form and Na form. Once
Eluting solution HCl 3 M HCl 4 M HCl 3 M HCl 4 M HCl 3 M HCl 4 M
regenerated they were contacted with synthetic solutions under the Eluting yield 83.6% 89.8% 73.8% 80.4% 87.1% 84.9%
same conditions as in the previous case.
After regeneration, the retention yields of K2629 are quite similar to
obtained with fresh resins (Table 8). It has been proven that the K2629 Table 8
resin can be regenerated to H-form or Na-form, and in both cases it still Reutilization tests experimental results with fresh and regenerated resins.
allows the recovery of Li with high capture yields (Table 8). In the case
K2629 TP 207 TP 208
of TP 207 and TP 208, the regenerated resin behavior is similar to
obtained with fresh resins when they are regenerated to Na-form, but Lithium retention yield with fresh resins 98.0% 98.4% 98.8%
not in H-form (Table 8). In this case, the performance of Li capture Lithium retention yield with regenerated Resin (in H 97.9% 52.2% 40.3%
form)
decreases drastically. This is in accordance with the recommendations
Lithium retention yield with regenerated Resin (in Na 97.8% 98.1% 98.2%
of the manufacturer (R = 40.3%). form)

3.4. Effect of Na content studies by response surface methodology

Compared with the results obtained with Li-bearing solutions without
Tests were performed with Li and Na bearing aqueous solutions Na, experimental yields and capacities decreased significantly, in-
(Table 9). In the multivariable design of experiments, a set of experi- dicating that although Li can be recovered with K2629 resin, even in
ments were performed with each resin and the relationship between the presence of Na, larger scale operation will be difficult, since the capa-
variables (Li and Na contents in coded terms) and responses (R and Q) city of the resin is very low in the presence of large amounts of sodium.
was studied using regression models 10. Statistical parameters and The stat-Ease Design expert was used for ANOVA analysis of the R
analysis of regression parameters for the predicted response quadratic (for K2629 resin) model (Fig. 4). The statistical significance of the
model are shown in Tables SI-3, SI-4 and SI-5. R models for the studied model was checked by an F-test. The model-F value of 6.31 and the
resins and Q for TP 208 were significant. Mathematical equations for probability value (Prob - F < 0.0136) implies that the model is sig-
significant models are shown in Table 10. nificant (significant models has prob-F < 0.05) (Table SI-3). Adequate
The average Li retention yield by the resin K2629 is 62.7%. precision compares the range of the predicted values to the mean
Experimental capacities varied from 0.16 mg g−1 to 3.24 mg g−1.

25

20
1/Qe (g·mg-1)

15

10

5
K2629 TP207 TP208

0
0 20 40 60 80 100 120

1/Ce (L·mg-1)

Fig. 3. Linearized Langmuir isotherms for K2629, TP 207 and TP 208 resins (Qe capacity at equilibrium; ce: lithium content at equilibrium).

6
F. Arroyo, et al. Desalination 468 (2019) 114073

Table 9
Experimental lithium retention yields (R) and capacity (Q) obtained for K2629, TP 208 and TP 207 in the CCRD experiments.
Run Coded variables K2629 TP 208 TP 207

−1 −1
Li Na R (%) Q (mg g ) R (%) Q (mg g ) R (%) Q (mg g−1)

1 1 1 64.5 0.65 89.5 0.90 90.5 0.91

2 1.414 0 62.8 1.57 86.8 2.17 88.0 2.20
3 -1 1 49.2 1.23 50.4 1.26 68.0 1.70
4 0 0 64.0 1.60 87.6 2.19 88.2 2.21
5 0 0 64.8 3.24 86.7 4.34 86.0 4.30
6 0 0 61.4 2.45 80.4 3.22 80.5 3.22
7 0 0 62.2 1.56 88.0 2.2 89.0 2.23
8 0 0 66.5 2.66 89.6 3.59 89.5 3.58
9 1 -1 62.6 1.57 86.6 2.17 88.2 2.21
10 -1 -1 65.2 1.63 87.4 2.19 88.8 2.22
11 0 −1.414 67.6 1.69 89.6 2.24 90 2.25
12 0 1.414 59.0 0.59 84.0 0.84 85 0.85
13 −1.0.414 0 65.2 0.16 80.0 0.20 87.2 0.22

Table 10 results, the model was considered adequate.

Predicted models for lithium retention yields (R) and capacity (Q) for each resin Final mathematical equation is a linear model (Table SI-3), in which
(RSM). the most significant factor is the Na content (linear coefficient: −3.28),
Resin Response Predicted model R2 much more influencing than the Li content (linear coefficient: +1),
confirming that the performance is very sensitive to the concentration
K2629 R (%) 62.71–3.28 Na + 1.00 Li + 4.48·Li·Na 0.678 of Na. Na content affects negatively to Li retention, probably because
TP 208 R (%) 83.69 + 5.59 Li – 5.25 Na + 9.98·Li·Na 0.670
Na moves the equilibrium to the opposite direction to the desired for
Q (mg·g−1) 3.13–0.52 Na +0.20 Li - 0.086 Li Na – 0.742
0.81·Li2–0.76 Na2 the retention. In the middle of the range of study, the Li yield is 62.71
TP 207 R (%) 86.12 + 2.61 Li - 3.20 Na + 5.17 Li Na 0.630 (independent term). In the range of study, Na is an important para-
meter, and retention could vary from 58% when Na content is the
Na: Na content in coded terms (adimensional); Li: Lithium content in coded highest, to 70%, when Na content is the lowest (Fig. 4). Regarding to
terms (adimensional). the Li loading capacity of K2629, the software suggest a not significant
model. For this reason, the model is not shown.
prediction error. The signal to noise ratio of 9.202 indicated an ade- In the case of TP 208 resin (Fig. 5, Table SI-4), experimental values
quate signal. A value > 4 is desirable for a good model and indicates of Li retention are in the range: 50.4%–89.6% (mean Li retention is
the predicted model could be used to navigate the design space. The 83.6%). The mathematical models of the retention is significant (Table
correlation coefficient (R2) shows how the model accounts for the re- SI-4) and there is a 1.32% chance that an F-value this large could occur
sponse. A value of 0.8 is desired for the good fit of the model. In this due to noise. This model can be used to navigate the design space. In
model, R2 is 0.678 for Lithium retention. The “Lack of fit tests” com- coded model, the Na content affected adversely to retention. It is a
pares the residual error to pure error from replicated design points. A linear model, and the influence of Li and Na is almost similar (coeffi-
significant lack of fit suggests there may be some systematic variation cients +5.59 and −5.25), so Li affects positively and Na negatively.
unaccounted for the hypothesized model. In this model a “lack of fit” The performance decreased in the case of solutions containing Na, the
not significant was determined by ANOVA analysis. Based on the

70

65
Lithium retention (%)

60

55

50

45
90
80
70
60
8500 50
7500
6500 Li content (mg/L)
5500 40
4500 30
3500
2500
1500 20
Na content (mg/L)
Fig. 4. RSM for lithium retention yield onto K2629 resin.

7
F. Arroyo, et al. Desalination 468 (2019) 114073

100

Lithium retention (%)

90

80

70

90
70
50
Li content (mg/L)
9000 7000 30
5000 3000 1000 10
Na content (mg/L)
Fig. 5. Response Surface for lithium retention yield onto TP 208 resin.

higher Na, and the lesser retention yield. Comparing the results ob- the software suggest a model not significant (F-value = 3.46), and it
tained in solutions without Na, both performance and capacity have cannot be used to navigate the design space (adequate preci-
lower values. sion = 4.525).
Regarding the Li loading capacity, the Model F-value of 4.02 implies If results are compared with the experimental results with solutions
there is 4.87% chance that an F-value this large could occur due to without Na, both performance and capacity vary within the same
noise. So, the model is considered as significant. Model can be used to ranges. The lithium retention yield decrease from 83% to around 68%.
predict the TP 208 Li retention capacity. Experimental capacities varied According to the experimental results and to the RSM model, TP 207
from 0.20 to 4.34 mg g-1 (mean capacity is 0.48 mg g−1). and TP 208 show a better Li sorption behavior than K2629 in the
Mean lithium retention yield onto TP 207 resin is 86.25% and the presence of Na. If the performance and capacity values are compared
experimental capacities varied from 0.22 to 4.30 mg g−1. The experi- for the same concentrations of Li with and without the presence of Na in
mental lithium retention yields are in the range: 68.0%–90.5%. The the solution, the low yield can be seen for the three resins due to the
Model F-value of 5.11 implies the model is significant and it showed an presence of Na in solution: the decrease is more important in the case of
adequate signal precision (7.646), so the model can be used in the range resin K2629 than in the case of resins TP 207 and TP 208. The ex-
of study. In the case of TP 207 resin, according to the mathematical perimental capacities also decrease for the three resins studied due to
model, the positive influence of lithium concentration is similar to the the presence of Na in solution. This means that although the resins
negative Na influence (−2.61 and −3.20, respectively), in the range allow to recover Li from this type of solutions, the operation with real
studied (as it was said for TP 208). brines, in which the concentration of Na will be much higher than that
Regarding to the Li loading capacity of TP 207 (Fig. 6, Table SI-5), of Li will be difficult, because the need for resin will be very high, which

100
Lithium retention (%)

90

80

70

60

50

10

30

50
Li content (mg/L)
70 9000
7000
5000
3000
90 1000
Na content (mg/L)
Fig. 6. Response Surface for lithium retention yield onto TP 207 resin.

8
F. Arroyo, et al. Desalination 468 (2019) 114073

100 4.5
90 K2629 TP207 TP208 K2629 TP207 TP208
4
80 3.5

Q (mg·g-1) · 10-3
70
3
60
2.5
R (%)

50
2
40
1.5
30
1
20
0.5
10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
W resin (g) W resin (g)

Fig. 7. Lithium retention (R, %) and capacity (Q, mg g-1) for each resin with real brines.

hinder the viability of scaling this process. predict the Li retention yields. While the predicted retention values are
higher than the experimental ones, the difference is much smaller when
3.5. Li retention from real brines the Mg and Ca ions have been removed. Therefore, it can be concluded
that by eliminating Ca and Mg, the Li retention is possible from the
The tests were performed with samples of 50 mL of real brines and generated brines in desalination facilities; although it is necessary to
different amounts of resin (1 g, 10 g and 20 g). Li retention yield is in take into account the partial abatement of Li due to the method of
the range 30%–55% and 52%–80%, depending on the type of resin and elimination of Mg and Ca.
the amount(Fig. 7).
In these tests TP 208 shown the maximum Li retention efficiency 4. Conclusions
and loading capacity, and the lowest Li retention and loading capacity
are shown for K2629. The capacities are below 10−3 mg g−1. The ca- This study shows that it is possible to recover Li from brines with ion
pacity obtained in studies with TP 208 resin is slightly higher than with exchange resins. For solutions in which there is only Li, the TP 207, TP
TP 207 and K2629. Both the yield and the capacity suffer a very im- 208 and K2629 resins have high retention yields (> 95%). When, in
portant decrease with respect to the tests with synthetic solutions, due addition to Li, there is Na in solution, the retention performance is low,
to the presence of other ions in the real brine (besides Li or Na). That is, especially in the case of resin K2629. This tendency is much more
there are other elements competing with the Li for the active sites of the marked in the case of using brines with other elements in solution. The
three resins. It is necessary to add a lot of resin to keep the yield above a resin most affected by the competitive effect of other ions in solution is
certain value; this causes the capacity to decrease drastically. K2629. The Lewatit K2629 resin is therefore very effective when used in
The Li retention yields are lower than predicted by the mathema- solutions where there is no Na, or the concentration of this element is
tical models developed in Section 3.4, i.e. the predicted performance is low. In other cases, it would be preferable to employ the TP207 or
much higher than the experimental performance in the case of real TP208 resins to retain Li from brines. It is also possible recovering Li
brines due to the presence of high concentrations of interfering ele- from the brines generated in desalination facilities using the three
ments. Taking into account the brine composition, Mg and Ca are studied resins if Ca and Mg are previously eliminated. in this case, the Li
considered the main interfering ions. To evaluate this issue, some ex- lost due to the method of elimination of Mg and Ca should be con-
periments have been carried out with a modified brine to which 99% of sidered.
Ca and Mg have previously been eliminated. To do that, the necessary Regarding kinetics, the fastest retention of Li is attributed to TP208
amount of sodium carbonate to precipitate Ca was added and pH was and the slowest to K2629. The theoretical capacity sequence is found to
increased to 11.5 with NaOH, removing Ca as CaCO3 and Mg as Mg be TP207 > K2629 > TP208, but the three values are quite similar.
(OH)2. The concentration of Li after Mg and Ca precipitation is de- Results indicated that the models obtained for pseudo-second order
termined to be 0.52 mg·L−1. were more suitable for the Li retention onto K2629, TP207 and TP208
The experiments were carried out by contacting 1 g of each type of resins. The better fit of the pseudo-second-order suggest that resins-Li
resin with 50 mL of the modified brine. Then they have been main- main interaction is controlled by chemical sorption, and the retention
tained in contact for 24 h to ensure equilibrium. After the contact, resin rate depends more on the availability of the active sites than the con-
and liquid solution were separated and the Li concentration not re- centration of Li in solution.
tained by resins was measured (Table 11). Langmuir separation factor (RL) varied from 0 to 1 in all experi-
The Na and Li contents of modified brines were replaced in the ments, indicating that lithium sorption is favorable for the 3 resins. Due
mathematical models calculated in Section 3.4 (Table 11) in order to to the determination factors of three resins values and the

Table 11
Results obtained in experiments with modified brines.
Modeled Li retention yield Initial Li content Li content at equilibrium Experimental Li retention yield Experimental/modeled Li retention
(%) (mg L−1) (mg L−1) (%) yield

K2629 30.5 0.52 0.39 25,0 0.82

TP 207 26.1 0.52 0.40 23.1 0.88
TP 208 44.9 0.52 0.31 40.4 0.90

9
F. Arroyo, et al. Desalination 468 (2019) 114073

characteristics of resins (homogeneous sites), Langmuir model is pre- 8309043 B2.

ferred for describing the interaction between Li and resins TP207, [22] Bauman, W.C., Burba, J.L., 2001. Composition for the recovery of lithium values
from brine and process of making/using said composition. US 6280693 B1.
TP208 and K2629. Therefore, the retention correspond to a mono- [23] Bauman, W.C., Burba, J.L., 1986. Intercalations of crystalline lithium aluminates.
molecular layer of Li onto homogeneous sites within resin surfaces. US 4727167.
Regarding desorption, yields ranged 73.8%–89.8%, achieving the [24] Bauman, W.C., Burba, J.L., 1965. Recovery of lithium from brines. US 5389349.
[25] A.H. Hamzaoui, B. Jamoussi, A. M’nif, Lithium recovery from highly concentrated
highest (> 80%) when using 4 M HCl as eluting solution. It is possible solutions: response surface methodology (RSM) process parameters optimization,
to reuse regenerated resins with high yields, similar to those obtained Hydrometallurgy 90 (2008) 1–7, https://doi.org/10.1016/j.hydromet.2007.09.
when the resin is used for the first time. 005.
[26] Harrison, S., Sharma, C.V.K., Viani, B.E., Peykova, D., 2014. Lithium extraction
composition and method of preparation thereof. US 8637428 B1.
Appendix A. Supplementary data [27] S. Hawash, E. Abd, E. Kader, G. El Diwani, Methodology for selective adsorption of
lithium ions onto polymeric aluminium (III) hydroxide, J. Am. Sci. Am. Sci. 66
(2010) 301–309.
Supplementary data to this article can be found online at https://
[28] J.M. Lee, W.C. Bauman, Recovery of lithium from brines. (1978) 4116858.
doi.org/10.1016/j.desal.2019.114073. [29] J.M. Lee, W.C. Bauman, Recovery of lithium from brines. (1979) 4159311.
[30] J.M. Lee, W.C. Bauman, Recovery of lithium from brines. (1980) 4221767.
References [31] J.M. Lee, W.C. Bauman, Recovery of lithium from brines. (1982) 4347327.
[32] J.M. Lee, W.C. Bauman, Lithium halide brine purification. (1983) 4376100.
[33] W.J. Repsher, K.T. Rapstein, Recovery of Lithium From Brine, (1981), p. 4291001.
[1] S. Nishihama, K. Onishi, K. Yoshizuka, Selective recovery process of lithium from [34] C.A. Borgo, A.M. Lazarin, Y.V. Kholin, R. Landers, Y. Gushikem, The ion exchange
seawater using integrated ion exchange methods, Solvent Extr. Ion Exch. (2011), properties and equilibrium constants of li+, Na+ and K+ on zirconium phosphate
https://doi.org/10.1080/07366299.2011.573435. highly dispersed on a cellulose acetate fibers surface, J. Braz. Chem. Soc. (2004),
[2] A. Lauren, P. Oppenheimer, World Lithium Resource Impact on Electric Vehicles, https://doi.org/10.1590/S0103-50532004000100010.
(2008). [35] C.A. Borgo, Y. Gushikem, Zirconium phosphate dispersed on a cellulose fiber sur-
[3] E. Siame, R.D. Pascoe, Extraction of lithium from micaceous waste from China clay face: preparation, characterization, and selective adsorption of Li+, Na+, and K+
production, Miner. Eng. 24 (2011) 1595–1602, https://doi.org/10.1016/j.mineng. from aqueous solution, J. Colloid Interface Sci. (2002), https://doi.org/10.1006/
2011.08.013. jcis.2001.8045.
[4] A. Yaksic, J.E. Tilton, Using the cumulative availability curve to assess the threat of [36] M. Petersková, C. Valderrama, O. Gibert, J.L. Cortina, Extraction of valuable metal
mineral depletion: the case of lithium, Resour. Policy 34 (2009) 185–194, https:// ions (Cs, Rb, Li, U) from reverse osmosis concentrate using selective sorbents,
doi.org/10.1016/j.resourpol.2009.05.002. Desalination 286 (2012) 316–323, https://doi.org/10.1016/j.desal.2011.11.042.
[5] S. Bunani, K. Yoshizuka, S. Nishihama, N. Kabay, Application of Bipolar Membrane [37] H. Bukowsky, E. Uhlemann, D. Steinborn, The recovery of pure lithium chloride
Electrodialysis (BMED) for Simultaneous Separation and Recovery of Boron and from “brines” containing higher contents of calcium chloride and magnesium
Lithium From Aqueous Solutions, 424 (2017), pp. 37–44, https://doi.org/10.1016/ chloride, Hydrometallurgy 27 (1991) 317–325, https://doi.org/10.1016/0304-
j.desal.2017.09.029. 386X(91)90056-R.
[6] C. Jiang, Y. Wang, Q. Wang, H. Xu T. Feng, Production of lithium hydroxide from [38] Y. Khambhaty, K. Mody, S. Basha, B. Jha, Kinetics, equilibrium and thermodynamic
lake brines through electro–electrodialysis with bipolar membranes (EEDBM), Ind. studies on biosorption of hexavalent chromium by dead fungal biomass of marine
Eng. Chem. Res. 53 (2014) 6103–6112, https://doi.org/10.1021/ie404334s. Aspergillus niger, Chem. Eng. J. 145 (2009) 489–495, https://doi.org/10.1016/j.
[7] S. Kim, H. Joo, T. Moon, S.H. Kim, J. Yoon, Rapid and Selective Lithium Recovery cej.2008.05.002.
from Desalination Brine Using an Electrochemical System, Processes & Impacts, [39] D. Wen, Y.S. Ho, X. Tang, Comparative sorption kinetic studies of ammonium onto
Environmental Science, 2019. zeolite, J. Hazard. Mater. (2006), https://doi.org/10.1016/j.jhazmat.2005.10.020.
[8] X. Liu, X. Chen, L. He, Z. Zhao, Study on extraction of lithium from salt lake brine by [40] I.-H. Lee, Y.-C. Kuan, J.-M. Chern, Equilibrium and kinetics of heavy metal ion
membrane electrolysis, Desalination 376 (2015) 35–40. exchange, J. Chinese Inst. Chem. Eng. (2007), https://doi.org/10.1016/j.jcice.
[9] X. Nie, S. Sun, X. Song, J. Yu, Further investigation into lithium recovery from salt 2006.11.001.
lake brines with different feed characteristics by electrodialysis, J. Memb. Sci. 530 [41] K.R. Hall, L.C. Eagleton, A. Acrivos, T. Vermeulen, Pore- and solid-diffusion kinetics
(2017) 185–191, https://doi.org/10.1016/j.memsci.2017.02.020. in fixed-bed adsorption under constant-pattern conditions, Ind. Eng. Chem.
[10] S.-Y. Sun, L.-J. Cai, X.-Y. Nie, X. Song, J.-G. Yu, Separation of magnesium and li- Fundam. 5 (1966) 212–223, https://doi.org/10.1021/i160018a011.
thium from brine using a Desal nanofiltration membrane, J. Water Proc. Eng. 7 [42] T.W. Weber, R.K. Chakravorti, Pore and solid diffusion models for fixed-bed
(2015) 210–217, https://doi.org/10.1016/j.jwpe.2015.06.012. Adsorbers, AICHE J. 20 (1974) 228–238, https://doi.org/10.1002/aic.690200204.
[11] B. Swain, Recovery and recycling of lithium: a review, Sep. Purif. Technol. 172 [43] N.D. Hutson, S.C. Zajic, R.T. Yang, Influence of residual water on the adsorption of
(2017) 388–403, https://doi.org/10.1016/j.seppur.2016.08.031. atmospheric gases in Li-X zeolite: experiment and simulation, Ind. Eng. Chem. Res.
[12] Chang, I.-L., Jiang, Y.-L., Shiu, J.-Y., Lin, J.-R., 2012. Process for Producing Lithium 39 (2000) 1775–1780, https://doi.org/10.1021/ie990763z.
Concentrate From Brine or Seawater. US 6764584 B2. doi:https://doi.org/10.1126/ [44] E. Voudrias, F. Fytianos, E. Bozani, Sorption description isotherms of dyes form
science.Liquids. aqueous solutions and waste waters with different sorbent materials. Glob. Nest,
[13] R. Chitrakar, H. Kanoh, Y. Miyai, K. Ooi, A new type of manganese oxide Int. J. 4 (2002) 75–83.
(MnO2·0.5H2O) derived from Li1.6Mn1.6O4 and its lithium ion-sieve properties, [45] S.V. Mohan, J. Karthikeyan, Removal of lignin and tannin colour from aqueous
Chem. Mater. (2000), https://doi.org/10.1021/cm0000191. solution by adsorption onto activated charcoal, Environ. Pollut. 97 (1997) 183–187.
[14] R. Chitrakar, H. Kanoh, Y. Miyai, K. Ooi, Recovery of lithium from seawater using [46] Tabatabai, M.A., Sparks, D.L., Goldberg, S., 2005. Equations and Models Describing
manganese oxide adsorbent (H 1.6 Mn 1.6 O 4) derived from li 1.6 Mn 1.6 O 4, Ind. Adsorption Processes in Soils. doi:https://doi.org/10.2136/sssabookser8.c10.
Eng. Chem. Res. 40 (2001) 2054–2058, https://doi.org/10.1021/ie000911h. [47] A.O. Dada, Langmuir, Freundlich, Temkin and Dubinin–Radushkevich isotherms
[15] K.-S. Chung, J.-C. Lee, W.-K. Kim, S.B. Kim, K.Y. Cho, Inorganic adsorbent con- studies of equilibrium sorption of Zn 2+ unto phosphoric acid modified Rice husk,
taining polymeric membrane reservoir for the recovery of lithium from seawater, J. IOSR J. Appl. Chem. 3 (2012) 38–45, https://doi.org/10.9790/5736-0313845.
Membr. Sci. (2008), https://doi.org/10.1016/j.memsci.2008.09.041. [48] M. Temkin, V. Pyzhev, Kinetics of ammonia synthesis on promoted iron catalysts,
[16] Kim, J.-S., Chung, K.-W., Lee, J.-Y., Kim, S.-D., 2012. Method for Preparing High- Acta Physicochim. 12 (1940) 327–356.
purity Lithium Carbonate From Brine. 2012/0328498 A1. [49] A. Da̧browski, Adsorption - from theory to practice. Adv. Colloid Interface Sci,
[17] L.W. Ma, B.Z. Chen, X.C. Shi, K. Zhang, Li+ Extraction/Adsorption Properties of Li- (2001), https://doi.org/10.1016/S0001-8686(00)00082-8.
Sb-Mn Composite Oxides in Aqueous Medium, Trans. Nonferrous Met. Soc, China [50] A. Günay, E. Arslankaya, I. Tosun, Lead removal from aqueous solution by natural
(English Ed, 2011, https://doi.org/10.1016/S1003-6326(11)60911-4. and pretreated clinoptilolite: adsorption equilibrium and kinetics, J. Hazard. Mater.
[18] K. Ooi, Y. Miyai, S. Katoh, Recovery of lithium from seawater by manganese oxide (2007), https://doi.org/10.1016/j.jhazmat.2006.12.034.
adsorbent, Sep. Sci. Technol. 21 (1986) 755–766, https://doi.org/10.1080/ [51] M.L.H. Figueroa, B.L. Rodríguez, V.M. Miranda, Cinética e isotermas de adsorción
01496398608056148. de Pb(II) en suelo de Monterrey XI, (2008).
[19] J. Park, H. Sato, S. Nishihama, K. Yoshizuka, Separation and recovery of lithium [52] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process
from geothermal water by sequential adsorption process with λ-MnO 2 and TiO2, Biochem. (1999), https://doi.org/10.1016/S0032-9592(98)00112-5.
Ion Exch. Lett. 5 (2012) 1–5, https://doi.org/10.3260/iel.2012.08.001. [53] J.M. Salman, V.O. Njoku, B.H. Hameed, Bentazon and carbofuran adsorption onto
[20] T. Wajima, K. Munakata, T. Uda, Adsorption behavior of lithium from seawater date seed activated carbon: kinetics and equilibrium, Chem. Eng. J. 173 (2011)
using manganese oxide adsorbent, Plasma Fusion Res. (2012), https://doi.org/10. 361–368, https://doi.org/10.1016/j.cej.2011.07.066.
1585/pfr.7.2405021.
[21] Alurralde, P., Mehta, V., 2012. Recovery of li values from sodium saturate brine. US

10