You are on page 1of 8

Separation and Purification Technology 276 (2021) 119307

Contents lists available at ScienceDirect

Separation and Purification Technology


journal homepage: www.elsevier.com/locate/seppur

Development of heterogeneous equilibrium model for lithium solvent


extraction using organophosphinic acid
Junnan Lu , Geoff W. Stevens , Kathryn A. Mumford *
Department of Chemical Engineering, The University of Melbourne, Parkville, VIC 3010, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: End-of-life lithium-ion batteries can be recycled to manufacture new batteries. The focus of this study is the
Circular Economy application of solvent extraction technology for lithium ion recovery from synthetic leach solutions. This study
Solvent Extraction demonstrates that an organic solvent containing Cyanex 272, tributyl phosphate and kerosene can be used for
Heterogeneous Equilibrium Model
this purpose. Additionally, a new heterogeneous equilibrium model for extraction is proposed and validated
Spent Lithium-ion Batteries recycling
using experimental data. Both the model, and the isotherm test results obtained indicate that pH, extractant
concentration and multivalent metal concentration (Co(II)) significantly impact lithium extraction, while the
impact of temperature, ionic strength, monovalent metals (Na(I)) and modifier concentration is minimal.

1. Introduction popular in the future [8].


Solvent extraction is a mature hydrometallurgical separation tech­
Driven by the global desire to reduce carbon dioxide emissions, the nology, typically implemented after leaching. Many studies have shown
revolution of energy storage and electric vehicles’ emergence has that organophosphinic acids are effective extractants to reclaim cobalt,
boosted lithium-ion batteries consumption [1,2]. Lithium-ion batteries nickel, and manganese from the spent lithium-ion batteries [9–11].
are currently the most common rechargeable battery due to their high Among these, Cyanex 272 has exhibited better performance for sepa­
energy density and stability [3]. Lithium-ion batteries’ core is the rating nickel and cobalt [12]. Typically, lithium is left in the raffinate for
cathodic material, containing transition metals oxides such as LiNi1/ later precipitation [13]. Despite many studies applying solvent extrac­
3Mn1/3Co1/3O2, LiCoO2 or LiFePO4. The high use of these transition tion to retrieve lithium from primary resources (brine, seawater, min­
metals in batteries has raised concerns regarding future supply chains eral), studies of extracting lithium from spent lithium-ion batteries
recently [4]. leachate is rare. The performance of extracting lithium with organo­
After thousands of recharging cycles, the growth of lithium dendrites phosphinic acid has not been fully investigated. In this paper, the same
reduce the capacity of batteries and as a result they need replacement. organophosphinic acid – Cyanex 272 for cobalt and nickel was applied
Rather than environmental-unfriendly landfilling, the recently proposed to lithium extraction. With the same solvent, lithium can be enriched
circular economy proposes manufacturing of new batteries from the without contamination for using multiple solvents.
spent ones [5]. The thermodynamic equilibrium model for metal extraction can aid
Among all recycling technologies, hydrometallurgy is so far the only in the screening and optimisation of extractants and operating condi­
approach to recover lithium from black mass (a mixture of cathode and tions, which can also facilitate the process development [14]. Chagnes
anode powders) [4]. Hydrometallurgy is also a preferred separation described a model to simulate cobalt extraction by D2EHPA (Di-(2-
technology due to its lower energy intensity and minimal gas emission Ethylhexyl) phosphoric acid), which can predict the extraction in
[6]. In a typical hydrometallurgical process, after pre-treatment various flowsheets [15]. In Vasilyev’s work, a model of extracting cobalt
including discharging, dismantling, and shredding; sulphuric or hydro­ and nickel with Cyanex 272 was developed [16]. Nevertheless, there is
chloric acid are often used to leach lithium-ion batteries’ cathodic ma­ still lack of model describing the extraction behaviour of lithium.
terial [5,7]. To date, the application of sulphuric acid has typically been In the present paper, the feasibility of extracting lithium with Cyanex
used in industry due to its stability and lower price, while hydrochloric 272 (an organophosphinic acid based extractant developed by Solvay,
acid is reported to have a higher leaching rate and so may become more also referred as ‘phosphinic acids’ in this paper for simplicity) was

* Corresponding author.
E-mail address: mumfordk@unimelb.edu.au (K.A. Mumford).

https://doi.org/10.1016/j.seppur.2021.119307
Received 25 April 2021; Received in revised form 12 July 2021; Accepted 12 July 2021
Available online 15 July 2021
1383-5866/© 2021 Elsevier B.V. All rights reserved.
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

studied with model prediction and experimental validation. The impact min is sufficient to ensure equilibrium. The initial concentration of Li(I)
of various operating conditions, regarding different pH, temperature, is 0.06 mol L-1, of Co(II) is 0.02 mol L-1, the pH range of extraction was
extractant and modifier concentrations, and other metals, was also controlled below 8.5 to avoid emulsion formation, in which point, the
examined. organic loss could be minimised, all experiments were conducted at
ambient temperature as 21 ± 2 ◦ C.
2. Experimental Chemicals, apparatus and procedures

2.1. The aqueous solution and organic solvent 2.4. Isotherm tests at different O/A ratios

The aqueous solution was prepared by dissolving metal salts (LR Isotherm experiments were conducted at constant pH and various O/
grade, >98%, purchased from Chem-Supply) into Reverse Osmosis A ratios to validate the model prediction. The solvent was pre-
water without further purification. The initial concentration of Li(I) was neutralised by 10 mol L-1 NaOH proportionally (88.47 mL per mol
0.6 mol L-1. 1 mol L-1 NaCl was chosen to mimic the salt product of acidic Cyanex 272). 20 mL aqueous solution was pipetted to the container,
leachate neutralisation from upstream. In addition, NaCl was proved to organic solution was then added quantitively to perform extraction at
create enough ionic strength buffer to prevent the emulsion formation various O/A ratios as 0.40, 0.50, 0.63, 0.80, 1.00, 1.26, 1.58, 2.00, 2.50.
during the extraction. Cyanex 272 (Bis (2,4,4-trimethylpentyl) phos­ Operating conditions were adjusted to perform the isotherm test at
phinic acid) with a density of 0.92 g/mL and Molecular Weight of 290 g/ various extractant concentrations (16 v%/24 v% Cyanex 272), tem­
mol, was supplied by Solvay [17] as extractant. The purity was titrated peratures (21 ◦ C/40 ◦ C), concentration of Co(II) (0.0 mol L-1/0.1 mol L-
1
as 90.0%. TBP (Tri-butyl phosphate) and VivaSol 2046 (A kerosene ) and NaCl concentrations (1 mol L-1/3 mol L-1).
provided by Viva Energy) were used as modifier and diluent respec­
tively. Cyanex 272 and TBP were pipetted quantitively and diluted with
2.5. Spectral analysis
VivaSol 2046 at the ambient temperature (21 ± 2 ◦ C). 10 mol L-1 NaOH
was used as pH modifier in the experiments to neutralise the solvent.
At each steady state, the aqueous phase was sampled and diluted for
analysis by Inductively Coupled Plasma Atomic Emission Spectroscopy
2.2. Experimental setup
(ICP-OES, Agilent ICP-720). Excess sodium chloride was added to the
standards to suppress the ionisation interference. Wavelengths at 610.4
A semicontinuous stirred cell was used, which is shown in Fig. 1
nm and 238.9 nm were applied to calibrate the aqueous Li(I) and Co(II)
[14]. A refrigerated/Heating circulator (Julabo F12) was used to ensure
concentrations, respectively. The metal’s distribution ratio was calcu­
an isothermal environment during extraction. Aqueous and organic
lated based upon mass balance. Once extraction was completed, the
liquids were contained within the jacketed cylinder (Schott Duran, in­
organic solvent was stripped by 1.0 mol L-1 hydrochloric acid solution to
ternal volume 250 mL, external volume 600 mL). The magnetic mixer
validate the mass balance.
(U-lab) with a criss-cross stirrer was used to disturb the interphase and
accelerate the mass transfer. pH meter (Mettler Toledo Seven compact
3. Thermodynamic equilibrium model
duo S213) equipped with pH probe (Mettler Toledo inlab® Ma Pro-ISM)
were applied to monitor the pH during the experiment.
3.1. Mechanism description for the solvent extraction
2.3. Equilibrium experiments at different pH
The mechanism of extraction is interpreted as a series of heteroge­
neous complexations. Monovalent metals (M+ ) diffuse from aqueous
The equilibrium experiments were conducted at various equilibrium
phase to organic phase due to the complexation with hydrophobic
pH conditions. 50 mL aqueous solution was first pipetted into the
extractant (phosphinic acid). The heterogeneous extraction reactions
container and then contacted with 50 mL organic solvent. The organic
happen across organic bulk, aqueous bulk, and interphase. The inter­
volume over aqueous volume ratio (O/A ratio) for all the experiments
phase is considered as the major site of these reactions [18,19].
was as constant at 1.0. Then, 10 mol L-1 sodium hydroxide aqueous
Consider the protonated form of phosphinic acid as L , and depro­
solution was added to the system gradually to represent different equi­
tonated form as L− , the reactions are as follows:
librium pH values. The total volume of sodium hydroxide solution added
was less than 2.5 mL; with this, the volume effect from adding sodium
1) Protonated form (either monomer HL(org) or dimer (HL)2(org) ) of the
hydroxide solution was negligible. Preliminary tests had shown that 10
extractant migrates from the organic bulk to the interphase, where it
is deprotonated by hydroxyl from the aqueous phase
HL(org) ⇌HL(inte) (1)

(HL)2(org) ⇌(HL)2(inte)

HL(inte) + OH −(aqu) ⇌L−(inte) + H2 O(aq) (2)

(HL)2(inte) + 2OH −(aq) ⇌2L−(inte) + 2H2 O(aq)

Here i(aq) , i(org) and i(inte) denote the species i at aqueous bulk, organic
bulk, and interphase.

2) Metals are attracted to the interphase due to affinity for the


extractant
+
M(aq) +
⇌M(inte) (3)
Fig. 1. Experimental setup of a semicontinuous stirred cell: 1 - jacketed cyl­
inder, 2 - pH probe, 3 – pH meter, 4 – refrigerated/heating circulator, 5 –
magnetic mixer. 3) Stepwise complexation reactions at the interphase are

2
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

+
M(inte) + L−(inte) ⇌ML(inte) (4) The heterogeneous complexation reaction can also be expressed in
the form of deprotonated extractant as:
ML(inte) + L−(inte) ⇌ML−2(inte) (5) +
M(aq) + aNaL(org) + Cl−(aq) ⇌MLa Cl1− a(org) + aNaCl(aq) (12)
+
M(inte) + aL−(inte) ⇌ML1−a(inte)
a
(6) The corresponding interphase equilibrium constant K1 is:

a is the stoichiometric ratio of extractant over metals. [MLa Cl1− a(org) ]


K1 = (13)
+
[M(aq) ]∙[NaL(org) ]a
4) As complex’ hydrophobicity increases, it transforms into the organic
bulk with counter ions for charge neutrality. If a is larger than 1, the Define the interphase dissociation constant of phosphinic acid as K’a :
[ ] [ ]
negative charge of complex can be neutralised by sodium ions from +
c NaL−(org) ∙c H(aq)
aqueous phase as: Ka’ = [ ] (14)
c HL(org)
ML1−a(inte)
a
+ (a − 1)Na+
(aq) ⇌MLa Naa− 1(org) (7)
The mechanism is also shown in Fig. 2 schematically.
When a is smaller than 1, the positive charge needs to be neutralised Define the ratio of metal M in different phases as distribution ratio
by anions from the aqueous phase. As chloride is the main anion present DM [20] as:
in the aqueous solution, the extraction can be expressed as:
totalamoutofMinorganic [MLa Cl1− a(org) ]
DM = = (15)
ML1−a(inte)
a
+ (1 − a)Cl−(aq) ⇌MLa Cl1− a(org) (8) totalamoutofMinaqueous +
[M(aq) ]

As a is smaller than 1 for lithium extraction in this study. The het­ K1 in Eq. (13) then equals to:
erogeneous complexation reaction is expressed in form of protonated
1
extractant as: K1 = DM ∙ (16)
[NaL(org) ]a
+
M(aq) + aHL(org) + (1 − a)Cl−(aq) ⇌MLa Cl1− a(org) + aH +
(aq) (9)

Because there is excess chloride existing in the aqueous phase. Re­ 3.2. Characteristics of leachate
action is simplified with assumption that chloride concentration is
constant. The corresponding interphase equilibrium constant K2 is: The Activity model can represent the characteristics of the aqueous
solution and organic solvent [21]. The equilibrium constant in Eq. (16)
+ a
[MLa Cl1− a(org) ]∙[H(aq) ] is rewritten in terms of activity coefficient γ as:
K2 = (10)
[M(aq) ]∙[HL(org) ]a
+
γ(MLa Cl1− a(org) ) 1
K1 = ∙DM ∙ (17)
+
γ(M(aq) )∙y(NaL(org) )a [NaL(org) ]a
5) Remaining deprotonated extractant can transform back to the
organic bulk with the cation (Na+) from pH adjuster (NaOH was used Define Y(org) and Y(aq) , which represent the nonideality of each phase.
in this study as pH adjuster) for charge neutrality. γ(MLa Cl1− a(org) )
Y(org) = (18)
L−(inte) + Na+ (11) γ(NaL(org) )a
(aq) ⇌NaL(org)

Fig. 2. Interphase complexation reactions during extracting (counter ions (such as Na+ and Cl-) are not shown in the scheme).

3
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

1 Table 1
Y(aq) = + (19) Pitzer Model Parameters [23]
γ(M(aq) )
cd (0)
βcd
(1)
βcd Cϕcd
Express the equilibrium constant in logarithm form.
LiCl 0.1494 0.3074 0.00359
logK1 = logY(org) + logY(aq) + logDM − a∙log[NaL(org) ] (20) CoCl2 0.4857 1.9670 − 0.02869
NaCl 0.0765 0.2664 0.00127
In this study, only variations to aqueous non-ideality are included by
Pitzer model, while organic non-ideality is regarded as constant [22].
Y(aq) is determined by Pitzer equations [23], and Y(org) is assumed to be extractant log[HL(init) ]. However, the reliability of this approach is
constant. The result of model prediction indicated that the assumption of debatable. The relation between logDM and log[HL(init) ] is not essentially
constant organic non-ideality is appropriate. linear if metals are not highly diluted. It has been elaborated by Baird
A summary of Pitzer equations is provided in Eqs. (21)–(29). The [20] that simply determining the structure of extracted species by slope
molality-based activity coefficient γM of cation M in a multi-component analysis is less rigorous and requires other methods to validate the ra­
aqueous electrolyte (with cation c and anions d, d’ ) can be expressed by tionality. Whereas this study proposed a new approach, which can
the Pitzer equation [24] as: determine the value of a and logC with equation:
( )
∑ ∑ ∑ logC = logDM − a∙log[NaL(org) ] (30)
2
lnγM =zM F+ md (2BMd +ZCMd )+ mc 2ΦMc + md ΨMcd
d c d where logC is the constant, and contains equilibrium constant and
∑∑ ∑∑
+ md md’ Ψdd’M +|ZM | mc md Ccd (21) activity coefficients from Eq. (20). Experimental data, at different
d d’ c d operating conditions are used to regress logC and a as per Fig. 3. With the
correct assumption of stoichiometric ratio, the equilibrium constant
where zM is the charge of the cation M, mi is the molality of the
should only change with temperature.
speciation i, subscripts a, a’ and d represents the other speciation of
The results for Li(I) and Co(II) are listed in Table 2. The stoichio­
cations and anion in the electrolyte. Z is given by:
metric ratios are regressed as 0.675 and 1.620 for lithium(I) and cobalt

Z= mi |zi | (22)
i

Ccd can be calculated with the parameter Cϕcd


ϕ
Ccd
Ccd = (23)
2(|zc zd |)0.5
F in equation (27) can be calculated as:
[ √̅̅ ] ∑∑
I 2 √̅̅
F = − AΦ √̅̅ + ln(1 + b I ) + mc md B’cd (24)
1+b I b c d

where I is the overall ionic strength of the electrolyte:


1∑
I= mi ∙z2i (25)
2 i

Also BMd in Eq. (21) and B’cd in Eq. (24) define the interaction of
oppositely charged ions as:
√̅̅ √̅̅
BMd = β(0) (1) (2)
Md + βMd g(d1 I ) + βMd g(d2 I ) (26)

∂Bcd 1 (1) ’ √̅̅ √̅̅


B’cd = = ∙(βcd g (d1 I) + β(2) ’
cd g (d2 I )) (27)
∂I I
where
2
g(x) = (1 − (1 + x)∙e− x ) (28)
x2

− 2 x2
g’ (x) = (1 − (1 + x + ))∙e− x
(29)
x2 2
Aϕ is the Debye-Hückel parameter, which equals 0.3881 for water at
20 ◦ C, b is an empirical parameter, which equals 1.2.
The interaction parameters (βcd , βcd and Cϕcd ) can be experimentally
(0) (1)

determined. These values are cited from Pitzer’s work in Table 1 [23],
and were pre-validated to be applicable over the range of concentration
used in this study.

3.3. Stoichiometric ratio

The stoichiometric ratio a of the complexation is vital when under­


standing the solvent extraction mechanism. The most used standard Fig. 3. Diagram to determine the correct value of stoichiometric ratio and
slope analysis utilises linear fitting of logDM and initial concentration of extraction constant.

4
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

(II). The extraction constant logC of lithium at 21 ◦ C is 0.1289, which is


smaller than cobalt’s (5.1363). The smaller constant indicates the
weaker affinity between the deprotonated phosphinic acid and the
lithium ion, which a typical monovalent metal [25]. It’s worth reiter­
ating that the approach was only validated in the leachate concentration
range where constant stoichiometric number assumption is valid (0.0 –
0.6 mol L-1).

3.4. Incorporation of temperature influence by Van’t Hoff equation

Temperature can affect the equilibrium in both ways of solution non-


ideality and the equilibrium constant. Pitzer model has already intro­
duced the impact upon aqueous non-ideality. The remaining context
relates the change of the equilibrium constant due to temperature effect,
which is described by the Van’t Hoff equation [26].
Given the Gibbs Equation:
ΔG = ΔH − TΔS (31)
Fig. 4. Natural logarithm of the partition ratio versus the reciprocal absolute
Combine it with the definition of Gibbs free energy: temperature to determine the extraction reaction enthalpy.

ΔG = − RT∙lnK (32)
4.1. Impact of pH
The relation of equilibrium constant and temperature is obtained as:
ΔH ΔS The deprotonated form of phosphinic acid was regarded as the major
lnK = − + (33) extractant to transform metals from the aqueous bulk to the organic
RT R
solvent. The proportion of deprotonated form L−(org) and the protonated
Provided the assumption that ΔH and ΔS are constant, differentiate
form HL(org) of the phosphinis acid is related to the aqueous proton
the equation (33) in respect of T, and obtains
concentration (pH). The relation with pH was deviated from Eq. (11)
d
lnK =
ΔH
(34) and Eq. (14) as
dT RT 2
[HL(org) ]
In equilibrium experiments, the changes of[NaL(org) ], Y(org) and Y(aq) in pH = − logKa’ − log (37)
[L−(org) ]
Eq. (20) are regarded as constant according to temperate T. Thus:
pH can represent the proportion of deprotonated phosphinic acid.
d d ΔH
lnK1 = lnDM = (35) The value of K’a of Cyanex 272 was determined by the experimental data.
dT dT RT 2
0.457 mol L-1 Cyanex 272 with different amounts of TBP was titrated by
And: adding sodium hydroxide quantitively as per Fig. 5. The influence of TBP
dlnDM ΔH on pH is negligible. The value of the K’a in Eq. (37) was regressed as
( )= − (36) 7.674 • 10-8 mol L-1.
1 R
d /T The relation between the deprotonated phosphinic acid and the
metal distribution ratio was described as per Eqs. (27) and (30). The
Therefore, the complexation reaction enthalpy can be obtained from single-stage extraction efficiency of lithium (I) and cobalt (II) at initial
the slope of the natural logarithm distribution ratio changing versus the concentrations of 0.06 mol L-1 and 0.02 mol L-1 are predicted and vali­
reciprocal absolute temperature. The result of temperature tests is dated by experimental data in Fig. 6. Regarding the extraction behaviour
shown in Fig. 4, the complexation enthalpy of Cyanex 272 with Li(I) is Co(II), Cyanex 272 extracts almost all cobalt at a pH value of 5.5. In
8.5 ± 0.5 kJ mol− 1, indicating that lithium extraction is an endothermic comparison, lithium is not extracted until pH reaches 5.5, indicating that
reaction. The extraction of lithium can benefit from a higher operating
temperature.

4. Results and discussion

The current model involves a series of nonlinear equations. The


calculations were completed via MATLAB and PYTHON. The least-
square method was validated to be sufficient to solve the heteroge­
neous equilibrium [22]. Equilibrium model was solved numerically by
an open-source Python library: Scipy, using ‘scipy. optimise.least_s­
quares’ function. The model parameters used are listed in Tables 1 and
2.

Table 2
Stoichiometric ratio, extraction constant and enthalpy of the extraction
equilibrium.
a logC(21 ◦ C) ΔH[kJ/mol] Range

Li (I) 0.675 0.1289 8.5 ± 0.5 0.0–0.6 10 mol L-1


Fig. 5. Comparison of the fitted and the experimental results when titrating
Co (II) 1.620 5.1363 – 0.0–0.6 10 mol L-1
organic solvent (16 v% Cyanex 272 = 0.457 mol L-1) with sodium hydroxide.

5
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

the affinity of the phosphinic acid with Li(I) is not as strong as it with Co
(II). Besides, the maximum extraction efficiency of Li(I) is smaller than
Co(II) as well. Extraction efficiency of lithium does not increase after pH
reaches 8.0. The extraction behaviour seems to be ‘capped’. This is also
due to the significantly weaker affinity between the lithium and the
extractant. Despite the excess organic extractant and presence of lithium
in the aqueous phase at equilibrium, the small equilibrium constant only
allows partial lithium to be transported to the organic phase. As per
Fig. 7, lithium’s single-stage extraction efficiency increases due to the
titration leap of deprotonated phosphinic acids. As the titration leap
slows down at pH of ~ 8.0, lithium’s single-stage extraction efficiency
cease to increase at pH of ~ 8.0.
It is worth noting that, at pH = 5.5, most cobalt is extracted into the
organic phase while most lithium stays in the aqueous leachate. The
model prediction indicates that the separation factor of these two metals
is high (the separator factor greater than 10,000). As a result, adjusting
the equilibrium pH at 5.5 is a sufficient approach to separate cobalt(II)
from lithium(I).

4.2. Impact of organic solvent composition

TBP has a vital role in the solvent, such as improving the physical
properties or preventing third phase formation [22]. Because TBP can
also be used as an extractant with FeCl3 to extract lithium [27], it is
essential to ascertain the role of TBP in this study. The organic solvent
(diluted with kerosene) containing both Cyanex 272 and TBP can extract
Li(I) and Co(II) as Fig. 6. In contrast, TBP alone does not exhibit any
noticeable extraction ability in preliminary tests over this pH range. In Fig. 7. Proportion of deprotonated phosphinic acid, single-stage extraction
addition, TBP has a minor influence of aqueous pH as per Fig. 5, and thus efficiency of Li (I) and Co (II) at various equilibrium pH.
neglected in this study. Consequently, TBP added to the organic solvent
is simply treated as a modifier, which can improve the physical prop­ 4.3. Impact of temperature
erties of solvent and avoid gel formation.
Cyanex 272 - a phosphinic acid, is regarded as the major extractant in The impact of temperature is interpreted by deriving Van’t Hoff
the solvent. Experiments show that different concentrations of Cyanex equation (Eq. (36)). The reaction enthalpy was determined as 8.5 ± 0.5
272 can impact the extraction significantly. Isothermal prediction and kJ mol− 1. The positive value indicates that the overall heterogeneous
validation are shown in Fig. 8(a). Experimental data obtained agrees extraction of lithium is endothermic. Model prediction and experimental
with the prediction. By increasing phosphinic acid from 16 v% to 24 v% validation were conducted at 21 and 40 ◦ C according to Fig. 8(b). Data
(0.4751 mol L-1 to 0.6856 mol L-1), the average distribution ratio of indicates that the organic loading is generally higher at a higher tem­
lithium is increased by 38%. perature range. In consequence, the extraction can benefit from a higher
Despite the improvement of isotherm behaviour, Cyanex 272 is a operating temperature.
viscous fluid. Adding more Cyanex 272 will exacerbate solvent fluidity
for increasing viscosity. In practice, the amount of Cyanex 272 added 4.4. Impact of multivalent metals
needs to compromise with the fluidity behaviour in units.
In practice, there are also other multivalent metals (for example, Co
(II), Ni(II) and Mn(II)) existing in the leachate of lithium-ion batteries.
Co(II), as a typical lithium-ion battery leachate component, are chosen
to represent the impact of other multivalent metals in this study. The
prediction and validation data in Fig. 8(c) indicate the competitive in­
fluence of cobalt. As lithium has much weaker affinity with the extrac­
tant, when solvent is not abundant at low O/A ratio, lithium gets
‘crowded out’. Hence there is more lithium residing in the aqueous
phase. As per Fig. 8(c) the isotherm of lithium does not increase
monotonically; instead, it bends downward.
The influence of other metals needs to be fully identified in process
development. Abundant O/A ratio and solvent need to be assured for
maximum lithium extraction yield. It is also important to identify the
competition from other multivalent metals, as they might hinder the
lithium extraction.

4.5. Leachate’s ionic strength

Fig. 6. Comparison of the prediction and the experimental validation of the Leaching is an essential upstream before solvent extraction [28] and
single-stage extraction efficiency of Li (I) and Co (II) at various equilibrium pH can impact the composition of the leachate. The most common used
(initial condition: Li(I) = 0.06 mol L-1, Co(II) = 0.02 mol L-1 ; O/A ratio = 1.0; lixiviants are conventional mineral acids like hydrochloric and sulphuric
temperature = 21 ± 2 ◦ C, with 16 v% Cyanex 272 = 0.457 mol L-1). acids due to their stability and easy availability [29]. In this study,

6
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

Fig. 8. Isothermal results of extracting lithium with Cyanex/272/TBP/kerosene at various operating conditions. (For all figures, curves represent the model pre­
diction, while scatters show the experimental validation). Error bars were analysed via measuring accuracies. All isotherm tests were controlled at the same pH of 8.0
± 0.1, and different O/A ratios. Results of 16v% Cyanex 272, 21 ◦ C, with 0.0 mol L-1 Co(II) and 1 mol L-1 NaCl was set as the benchmark. Operating conditions are
also adjusted at different (a) Cyanex 272 concentrations (16 v%/24 v%), (b) temperatures (21 ◦ C/40 ◦ C), (c) Co(II) concentrations (0.0 mol L-1/0.1 mol L-1), (d) Ionic
strengths (1 mol L-1 NaCl/3 mol L-1 NaCl)).

sodium chloride at different concentrations is chosen to represent the slope analysis, the current study uses a new approach to determine the
leachate’s characteristics, indicating the impact of hydrochloric acid stoichiometric ratio. The stoichiometric ratio of extraction is not
and pH modifier NaOH. necessarily an integer. Secondly, this model can predict the isotherm
As shown in Fig. 8(d), both predictive and experimental data show performance regarding different operating conditions satisfactorily by
that the influence of higher sodium chloride content on lithium further integrating heterogeneous reaction constant, aqueous activity
extraction equilibrium is not obvious. The result implies that influence coefficient, and reaction enthalpy. In the future, methodologies devel­
of concertation of lixiviant and pH modifier on extraction performance is oped in this study can be further integrated with industrial solutions.
minimal.
Worth of iterating that, NaOH is a preferred pH modifier. Because CRediT authorship contribution statement
sodium(I)’s affinity with phosphinic acid is weaker than lithium(I),
hence, will residues in the raffinate. On the other hand, some pH mod­ Junnan Lu: Conceptualization, Data curation, Formal analysis,
ifier (such as lime), may introduce multivalent cation into the system Investigation, Methodology, Visualization, Writing - original draft,
(Ca2+), which might hinder the lithium extraction. Writing - review & editing. Geoff Stevens: Methodology, Supervision,
Writing - review & editing. Kathryn Mumford: Funding acquisition,
5. Conclusion Project administration, Resources, Methodology, Supervision, Writing -
review & editing.
In conclusion, this study has investigated the feasibility of extracting
lithium with a conventional solvent - Cyanex272/TBP/kerosene. The Declaration of Competing Interest
results indicate that lithium can be extracted at high pH range (5.5–8.0),
providing a way to recycle, purify, and concentrate lithium-ion batte­ The authors declare that they have no known competing financial
ries’ cathodic metals in one step with the same solvent as cobalt and interests or personal relationships that could have appeared to influence
nickel. Hence, the process could prevent the contamination of different the work reported in this paper.
solvents. However, due to the weak affinity between lithium and
phosphinic acid, influence of other multivalent metals (such as Co(II)) Acknowledgement
needs to be fully identified in preliminary tests.
A model, which probes into the mechanism of solvent extraction, has The authors would like to acknowledge the University of Melbourne
been proposed in this study as well. Firstly, rather than using standard for the Melbourne Research Scholarship. The help and support from the

7
J. Lu et al. Separation and Purification Technology 276 (2021) 119307

families and friends are also gratefully acknowledged. [13] D. Dutta, A. Kumari, R. Panda, S. Jha, D. Gupta, S. Goel, M.K. Jha, Close loop
separation process for the recovery of Co, Cu, Mn, Fe and Li from spent lithium-ion
batteries, Sep Purif Technol 200 (2018) 327–334.
References [14] J. Rydberg, Solvent extraction principles and practice, New York : M. Dekker,
c2004. 2nd ed., rev. and expanded.2004.
[1] D. Bank, F.I.T.T for investors Welcome to the Lithium-ion Age, Deutsche Bank [15] A. Chagnes, Simulation of Solvent Extraction Flowsheets by a Global Model
(2016). Combining Physicochemical and Engineering Approaches—Application to Cobalt
[2] T.P. Narins, The battery business: Lithium availability and the growth of the global (II) Extraction by D2EHPA, Solvent Extr Ion Exc 38 (1) (2020) 3–13.
electric car industry, The Extractive Industries and Society 4 (2017) 321–328. [16] F. Vasilyev, S. Virolainen, T. Sainio, Numerical simulation of counter-current
[3] P. Greim, A. Solomon, C. Breyer, Assessment of lithium criticality in the global liquid–liquid extraction for recovering Co, Ni and Li from lithium-ion battery
energy transition and addressing policy gaps in transportation, Nature Commun 11 leachates of varying composition, Sep Purif Technol 210 (2019) 530–540.
(2020) 1–11. [17] Solvay, Cyanex® 272 Extractant.
[4] J. Piątek, S. Afyon, T.M. Budnyak, S. Budnyk, M.H. Sipponen, A. Slabon, [18] G.W. Stevens, J.M. Perera, F. Grieser, Interfacial aspects of metal ion extraction in
Sustainable Li-Ion Batteries: Chemistry and Recycling, Adv Energy Mater 2003456. liquid-liquid systems, Rev Chem Eng 17 (2001) 87–110.
[5] G. Harper, R. Sommerville, E. Kendrick, L. Driscoll, P. Slater, R. Stolkin, A. Walton, [19] Y. Marcus, A.K. SenGupta, Ion Exchange and Solvent Extraction: A Series of
P. Christensen, O. Heidrich, S. Lambert, A. Abbott, K. Ryder, L. Gaines, Advances, Volume 17, CRC Press, 2004.
P. Anderson, Recycling lithium-ion batteries from electric vehicles, Nature 575 [20] M.H.I. Baird, C. Hanson, T.C. Lo, Handbook of solvent extraction, New York :
(7781) (2019) 75–86. Wiley, c1983.1983.
[6] J.B. Dunn, L. Gaines, J. Sullivan, M.Q. Wang, Impact of recycling on cradle-to-gate [21] K.H. Lum, G.W. Stevens, J.M. Perera, S.E. Kentish, The modelling of ZnCl2
energy consumption and greenhouse gas emissions of automotive lithium-ion extraction and HCl co-extraction by TBP diluted in ShellSol 2046, Hydrometallurgy
batteries, Environ Sci Technol 46 (22) (2012) 12704–12710. 133 (2013) 64–74.
[7] Y. Yao, M. Zhu, Z. Zhao, B. Tong, Y. Fan, Z. Hua, Hydrometallurgical Processes for [22] B. Tan, C. Chang, D. Xu, Y. Wang, T. Qi, Modeling of the Competition between
Recycling Spent Lithium-Ion Batteries: A Critical Review, ACS Sustain Chem Eng 6 Uranyl Nitrate and Nitric Acid upon Extraction with Tri-n-butyl Phosphate, ACS
(11) (2018) 13611–13627. Omega 5 (21) (2020) 12174–12183.
[8] Z. Takacova, T. Havlik, F. Kukurugya, D. Orac, Cobalt and lithium recovery from [23] K.S. Pitzer, Activity Coefficients in Electrolyte Solutions, CRC Press, 2018.
active mass of spent Li-ion batteries: Theoretical and experimental approach, [24] K.S. Pitzer, Thermodynamics of electrolytes. I. Theoretical basis and general
Hydrometallurgy 163 (2016) 9–17. equations, The Journal of Physical Chemistry 77 (2) (1973) 268–277.
[9] Y. Yang, S. Xu, Y. He, Lithium recycling and cathode material regeneration from [25] J. Ribas Gispert, Coordination chemistry, Wiley-VCH, 2008.
acid leach liquor of spent lithium-ion battery via facile co-extraction and co- [26] R.J. Silbey, R.A. Alberty, M.G. Bawendi, Physical chemistry, 4th ed. / Robert J.
precipitation processes, Waste Manage 64 (2017) 219–227. Silbey, Robert A. Alberty, Moungi G. Bawendi. ed., Wiley, 2005.
[10] G. Dorella, M.B. Mansur, A study of the separation of cobalt from spent Li-ion [27] H. Su, Z. Li, J. Zhang, W. Liu, Z. Zhu, L. Wang, T. Qi, Combining Selective
battery residues, J Power Sources 170 (1) (2007) 210–215. Extraction and Easy Stripping of Lithium Using a Ternary Synergistic Solvent
[11] P. Zhang, T. Yokoyama, O. Itabashi, T.M. Suzuki, K. Inoue, Hydrometallurgical Extraction System through Regulation of Fe3+ Coordination, ACS Sustain Chem
process for recovery of metal values from spent lithium-ion secondary batteries, Eng 8 (4) (2020) 1971–1979.
Hydrometallurgy 47 (2-3) (1998) 259–271. [28] J. Zhang, J. Hu, W. Zhang, Y. Chen, C. Wang, Efficient and economical recovery of
[12] K. Omelchuk, P. Szczepański, A. Shrotre, M. Haddad, A. Chagnes, Effects of lithium, cobalt, nickel, manganese from cathode scrap of spent lithium-ion
structural changes of new organophosphorus cationic exchangers on a solvent batteries, J Clean Prod 204 (2018) 437–446.
extraction of cobalt, nickel and manganese from acidic chloride media, Rsc Adv 7 [29] C. Peng, J. Hamuyuni, B.P. Wilson, M. Lundström, Selective reductive leaching of
(10) (2017) 5660–5668. cobalt and lithium from industrially crushed waste Li-ion batteries in sulfuric acid
system, Waste Manage 76 (2018) 582–590.

You might also like