Yong Xu, Qingbai Chen, Yang Gao, Jianyou Wang,2,*, Huiqing Fan, Fei Zhao

1 Tianjin Key Laboratory of Environmental Technology for Complex Trans-Media Pollution,College of Environmental Science and Engineering,Nankai University,Tianjin 300350,China

2 College of Chemical Engineering and Material, Quanzhou Normal University, Quanzhou 362000, China

3 Center for Water and Ecology, State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China

Keywords:

ABSTRACT

1. Introduction

Lithium (Li) consumption soars exponentially due to the massive usage of portable electronics and electric vehicles in recent years [1–3]. It has been estimated that Li demand as Li2CO3will increase from 1.8 × 105t in 2015 to 1.6 × 106t in 2030 and most will be used in the manufacture of lithium ion batteries [4,5].Although Li reserves in seawater is about 5000 times higher than that in land, Li abundance in the former, only 0.1–0.2 mg?L-1, is extremely low compared with that in the latter, which makes reserves in the oceans hard to be extracted and makes land the only currently viable Li source of mining value [1,4]. According to statistics, the total Li reserves in land is ～53 Mt and mainly exists in the form of ores (～22%) and salt lakes (～78%) [6]. In addition,considering the higher energy consumption and more severe environmental issues in Li ore mining, extracting Li from salt lakes is becoming more industrially preferable [2,7]. However, the high mass ratio of Mg/Li in most of salt lakes hinders the application of traditional Li extraction methods, including solar evaporation,chemical precipitation, and solvent extraction, which is due to the similar chemical properties of Mg2+and Li+[7–9]. Therefore,before Li extraction using the above conventional methods, it has become a consensus to reduce the Mg/Li ratio of salt lake brine.

At present, the effective methods for reducing the Mg/Li ratio mainly include membrane[2,10–12]and electrochemical methods[4,13,14]. As typical membrane methods, selective nanofiltration(SNF)and selective electrodialysis(SED)are the hotspots of current research,attracting extensive attention in the field of fractionating Li+from Mg2+[15–22]. As described in Text S1 in Supplementary Material, SNF and SED are completely different in their mechanisms, involving pressure and electro-driven processes, respectively. The feasibility of SNF in recovering Li from diluted bittern has been first proposed by Wen et al.[15].They found that satisfying rejection ratios of Mg2+(～64%) and(>96%) are achieved with the Desal-5 DL membrane.Subsequently, Sun et al. [16]have studied the influence of operating pressure, pH, temperature, and Mg/Li ratio on Mg/Li separation performance. They found that the competitive coefficient of Mg2+and Li+is sensitive to the Mg/Li ratio when the Mg/Li ratio is <20, but tends to be a constant value when the Mg/Li ratio is >20. Meanwhile, Li et al. [17] have conducted SNF separation experiments to fractionate Li+from brine with multi-cation presence, and found that the co-existence of divalent ions significantly decreases the rejection ratio of Li+and increases that of Mg2+, having a larger influence on Mg/Li separation performance than that observed for Na+and K+. To improve the Mg/Li separation of SNF membranes, Xu et al.[18] have designed and successfully fabricated a positivelycharged SNF membrane via interfacial polymerization between polyethyleneimine and trimesoyl chloride on polyethersulfone ultrafiltration membranes. The prepared membranes exhibit high positive charge at pH < 9.3 due to a large amount of unreacted—NH3+and —NH2+from PEI. In terms of Li fractionation using SED,Nie et al. [19] have adopted SED stacks assembled with CSO and ASA membranes to fractionate Li+from high Mg/Li ratio brine.With optimized operating parameters, the Mg/Li ratio in the product stream was reduced from 150 in the feed to 8.0. Chen et al.[20] have further investigated in detail the influence of coexisting cations on Li recovery during SED.They found that Li+migration is negatively affected by coexisting cations and the influence order of coexisting cations is contrary to their hydration radius, i.e., K+-> Na+> Ca2+> Mg2+, which is reasonably explained by their conceptual model of partial dehydration. Zhao et al. [21] have systematically studied the sensitivity of Li+fractionation using SED to temperature.The results indicated that,although increased temperature is favorable for Li recovery due to the enhancement of ion exchange membrane (IEM) permeability and ion mobility, the separation performance of Li+and coexisting cations is only slightly affected by temperature change.Zhao et al. [22] have proposed a sandwiched liquid-membrane electrodialysis system,composed of two conventional cation exchange membranes and a Li-loaded organic liquid film, to selectively recover Li+from high Mg/Li ratio brine. According to experimental results, the Mg/Li ratios in the product stream are reduced from 25, 50, and 100 in brine to 1.73,2.06,and 1.79,respectively.From the above analysis,it has been found that the two processes have their own characteristics and related existing studies have mostly focused on their own technology development or relative membrane material fabrication strategies. This provides an important technical support for effectively reducing the Mg/Li ratio in salt lake brine. However, in the field of treating high Mg/Li ratio brine, few studies have provided a fair and systematic comparison of the two processes,which is crucial for the rational selection of the two processes in practical application.

In the current study,the differences of SNF and SED processes in continuous mode for treating the same high Mg/Li ratio brine were analyzed, mainly focusing on the objectives below:

(1) to study the influence of typical operating parameters on SNF and SED processes and optimize relative operating parameters,

(2) to determine the Li+fractionation performance of SNF and SED at optimized operating parameters, and

(3) to analyze systematically the differences of the two processes during the Mg/Li separation process.

2. Materials and Methods

2.1. Brine preparation and membranes

Brine,with a high Mg/Li ratio around 60, was prepared according to the composition of West Taijinar salt lake of Qinghai province in China (Table 1) [10]. NaCl, Na2SO4, LiCl, and MgCl2?6H2O(AR) were purchased from Tianjin Bohua Chemical Reagent Co.,Ltd. (Tianjin, China). Pure water with resistivity higher than 15 MΩ?cm was produced using a RO/EDI-500 pure water equipment(Tianjin Zhongling Water System Technology Co., Ltd., Tianjin,China).

Table 1 Composition of high Mg/Li ratio brine

In this study, SNF membranes were 4040-SR200 selective nanofiltration membranes purchased from Koch Membrane Solutions, Inc. (Wilmington, DE, USA). Anion exchange membrane(AEM) and Monovalent selective cation exchange membrane(MSCEM) were AMX and CIMS type, respectively, which were purchased from ASTOM Corp. (Tokyo, Japan). The relevant performance parameters of SNF membranes and IEMs are listed in Table S1.

2.2. Experimental set-ups

The schematic diagram of the SNF device, with one 4-inch SNF membrane, is shown in Fig. 1(a). High Mg/Li ratio brine was pumped into the SNF membrane with pressurization by a booster and a high-pressure pump. The operating pressure (OP) and feed flow rate (FFR) were controlled by adjusting the frequency of the high-pressure pump and valve opening. To maintain the stability of brine qualities, the reject and permeate streams of SNF process were returned and well mixed in the feed tank. In addition, the brine temperature was maintained around 25 °C using a coiled condensing tube connected to a thermostat.

The SED stack contained 20 cell-pairs, in which the IEM dimensions were 20 cm × 40 cm with an effective area of 390 cm2(Fig. 1(b)). Specifically, 21 sheets of CIMS membrane and 20 sheets of AMX membrane were alternated in the SED stack. During the SED process, high Mg/Li ratio brine was fed to the dilute circulating tank at a constant flow rate, while the dilute and concentrate of the SED were continuously circulated between the stack and respective circulating tank. Meanwhile, dilute and concentrate overflow from the respective circulating tank were produced. In addition,the electrolyte solution was 5% (mass) Na2SO4which circulated between the SED electrode compartments and electrolyte tank.To maintain the temperature around 25 °C, three coiled condensing tubes connected to a thermostat were put in the circulating tanks of dilute, concentrate and electrolyte, respectively.

2.3. Experimental design and procedure

For the SNF process, OP and FFR play an important role in the Mg2+/Li+separation performance. The detailed conditions of SNF experiments are described in Table S2. First, single-factor experiments of OP and FFR were respectively conducted to obtain relatively-optimized operating parameters. To ensure reliability of experimental results, the permeate and reject were sampled after 30 min running time when operating conditions were changed.

For the SED process, single-factor experiments were conducted with the experimental conditions listed in Table S2. The influence of cell-pair voltage (CPV) and replenishment flow rate (RFR) were investigated. At the beginning of each experiment, the concentrate and dilute circulating tanks contained with 3 g?L-1NaCl solution and brine, respectively. The circulating flow rate of dilute and concentrate compartments were 300 L?h-1and the flow rate of electrolyte solution 100 L?h-1. When the system was stable, the flow rates of the dilute/concentrate overflow were measured at least three times and the results averaged.

Fig. 1. Schematic diagrams of the SNF (a) and SED (b) device.

2.4. Analytical methods

Concentrations of Cl-,Mg2+,Li+,and Na+in water samples were determined by the methods listed in Table S3. The silver nitrate titration method was adopted to determine Cl-concentrations,while the EDTA titration method was used to determine Mg2+concentrations[20,23].Li+concentrations were analyzed by an atomic absorption spectroscopy(TAS-990,Beijing Purkinje General Instrument Co.,Ltd.,Beijing,China)[20].Na+concentrations were determined by electrically balancing the ions in solution according to Eq. (1) [20]

where C(Cl-), C(Li+), and C(Mg2+) are the concentrations of Cl-, Li+,and Mg2+, respectively, g?L-1.

2.5. Data analysis

2.5.1. Ionic flux

The ionic fluxes of SNF and SED processes were calculated using Eqs. (2) and (3), respectively [23–25]. They could be used to indicate the ionic migration amount through the membrane per unit time over the unit membrane area.

In Eqs. (2) and (3), J (mol?h-1?m-2) is the ionic flux, Qp(L?h-1)the flow rate of permeate solution,Cp(g?L-1)the ion concentration in the permeate solution, Am(m2) the effective membrane area, M(g?mol-1) the ion molar mass, CC(g?L-1) the ionic concentration of concentrate overflow,QC(L?h-1)the flow rate of concentrate overflow,n the number of cell-pairs in SED stack,M(g?mol-1)the ionic molar mass, and Se(m2) the effective membrane area.

2.5.2. Selectivity coefficient

To make a fair and valid comparison of Mg/Li permselectivity between SNF and SED, selectivity coefficient between Li+and Mg2+, i.e. S(Li+/Mg2+), was defined as Eq. (4) [23]. If S(Li+/Mg2+) > 1, it can be inferred that the transmembrane migration of Li+is prior to that of Mg2+. Meanwhile, the bigger S(Li+/Mg2+),the higher permselectivity of Li+than that of Mg2+.

where JLiand JMg(mol?h-1?m-2) are the fluxes of Li+and Mg2+,respectively, and CLiand CMg(mol?L-1) the molar concentrations of Li+and Mg2+in the feed brine, respectively.

2.5.3. Recovery ratio

The recovery ratio of Li+of SNF and SED were defined by Eqs.(5)and (6), respectively [7,17].

where R(Li) (%) is the Li+recovery ratio of the two membrane processes,Cp,Li,Cc,Liand Cf,Li(g?L-1)the Li+concentration of SNF permeate,SED concentrate and feed,respectively,Qp,Qcand Qf(L?h-1)the flow rate of SNF permeate, SED concentrate and feed, respectively.

2.5.4. Specific energy consumption

The specific energy consumption is the energy consumption per unit mass of Li in the purified solution. It was calculated using[7,25]

where ESEC(kW?h?kg-1) is the specific energy consumption, P (W)the total power of the SNF device or SED stack (calculated by Eqs.(S1) and(S2), respectively), and Qp(L?h-1) and Cp,Li(g?L-1)the flow rate and Li concentration of purified solution, respectively.

2.5.5. Li+fractionation capacity

The Li+fractionation capacity meant the Li+production per unit time and effective membrane area, which was defined using

where P(Li) (g?h-1?m-2) is the specific energy consumption, Qp(L?h-1)and Cp,Li(g?L-1)the flow rate and Li concentration of purified solution, respectively, and Se(m2) the effective membrane area.

2.5.6. Concentration factor

The concentration factor of Li+indicated the fold of Li+concentration in the purified solution compared with that in the feed,which was calculated using

where ε(Li) is the concentration factor; Cp,Liand Cf,Li(g?L-1) the Li+concentration in purified solution and feed, respectively.

3. Results and Discussion

3.1. Li+ fractionation from Mg2+ via SNF process

As a membrane process driven by a pressure difference across the membrane, SNF is mainly influenced by OP and FFR. Thus,the influence of these operating parameters on the performance of Li+fractionation from Mg2+was first investigated.Then,Li+fractionation performance of SNF was analyzed based on optimization of these parameters.

3.1.1. Influence of operating pressure

The influence of OP, which ranged from 1.8 to 3.2 MPa, was studied under a constant FFR of 200 L?h-1. The rejection ratios to different ions,calculated using Eq.(S3),were significantly different(Fig. S2(a)). Specifically, in the OP range, the rejection ratios to Li+and Na+were –48.8% to –36.0% and –14.5% to –5.1%, respectively,and the negative rejection of monovalent ions(Table S4)had been reported in many studies [14,17,18]. Meanwhile, the rejection ratios to Cl-and Mg2+were 36.2%–44.5%and 52.7%–62.4%,respectively.This indicated that Li+and Na+ions were concentrated in the SNF permeate while more than one half of Mg2+ions were rejected by the SNF membrane.This phenomenon resulted from the higher diffusion coefficient,hydrated ion radius and hydration free energy of Mg2+compared to those of Li+and Na+[17,18]. It caused Mg2+ions to be much more influenced by the effects of steric hindrance,Donnan exclusion, and dielectric exclusion from the SNF membrane [12,15,17]. Moreover, a considerable part of Mg2+ions exist in the form of ion pairs, i.e. MgCl+, which further enhanced the steric hindrance of Mg2+[15].Therefore,considering that the rejection of Mg2+ions was relatively higher than that of Cl-ions, more Li+and Na+ions were ‘‘pulled” through the SNF membrane to maintain the electro neutrality in the SNF permeate.It finally made that the concentrations of monovalent ions in SNF permeate were higher than those in feed brine.

Ion fluxes increased with the increased OP due to the enhanced driving force for ion transmembrane migration(Fig.2(a)).However,the increasing trend of Li+flux was different from that of Mg2+flux.Specifically,Li+flux increased significantly with OP increasing from 1.8 to 2.4 MPa and tended to be stable at OPs higher than 2.8 MPa,while the Mg2+flux increased almost linearly with increased OP.This was mainly because the Mg2+concentration in the feed was much higher than the Li+concentration while the Li+concentration in the SNF reject was lower than that in the SNF permeate.In addition, S(Li+/Mg2+) increased from 3.54 to 4.42 with OP from 1.8 to 2.4 MPa and then decreased from 4.42 to 3.94 with OP from 2.4 to 3.2 MPa, respectively. Corresponding to the variation trend of S(Li+/Mg2+) with OP, the mass ratio of Mg/Li in SNF permeate was lowest when the OP was 2.4 MPa. On one hand, with increased OP, the SNF membrane structure could have become more compacted, which was conducive to steric hindrance effect enhancement, thereby improving Mg2+rejection. On the other hand, the concentration polarization phenomenon was intensified with increased OP, which was not conducive to Mg2+/Li+separation.Clearly, with increased OP, especially when higher than 2.4 MPa,the negative effect of concentration polarization gradually exceeded the positive effect of membrane compaction, thus weakening the selective separation performance of Mg2+and Li+. Therefore,although Li+recovery ratio increased from 23.9% to 52.6% with increased OP, the optimal OP was chosen to be ～2.4 MPa.

Fig. 2. Effects of operating pressure on ion fluxes (a), selectivity coefficient, Mg/Li ratio, and Li+ recovery ratio (b) during SNF process.

3.1.2. Influence of feed flow rate

Based on the result of OP variation, the influence of FFR on Mg2+/Li+separation performance was investigated with a constant 2.4 MPa OP while the FFR ranged from 100 to 300 L?h-1.The order of the ion rejection ratios was the same as mentioned above, i.e.,Mg2+> Cl-> Na+> Li+(Fig. S2(b)). Different from the effect of OP,with increased FFR,rejection ratios for Mg2+and Cl-increased significantly while the rejection ratios of Na+and Li+declined by varying degrees. When the FFR increased from 100 to 300 L?h-1, the rejection ratios of Mg2+and Cl-increased from 46.3% and 30.4%to 69.8% and 45.7%, respectively, while the rejection ratios of Li+and Na+decreased from –31.8% and –16.6% to –52.6% and –25.3%, respectively. In addition, with FFR increased from 100 to 300 L?h-1, the fluxes of Li+, Na+, and Cl-fluxes increased by～65.6%, 53.0%, and 11.1%, respectively, while the Mg2+flux decreased ～20.1%, indicating that increased FFR significantly improved the separation performance of Mg2+/Li+(Fig. 3(a)).

Specifically, with increased FFR, the linear flow velocity (calculated using Eq. (S4)) increased from 0.47 to 1.42 cm?s-1, which could have alleviated the concentration polarization of Mg2+on the SNF membrane surface and facilitated the mass transfer of Li+, Na+, and Cl-[26,27]. Therefore, the separation performance of Mg2+/Li+with a high FFR was superior to that with a low high FFR. When the FFR increased from 100 to 300 L?h-1, S(Li+/Mg2+)increased from 2.61 to 5.41 and the mass ratio of Mg/Li was further reduced from 24.2 to 11.7. However, increased FFR was not conducive to improving the Li+recovery ratio. Compared with 100 L?h-1, the Li+recovery ratio with an FFR of 300 L?h-1decreased from 56.2%to 32.1%.Thus,to ensure the efficiency of SNF,the optimal FFR was selected at 140 L?h-1.

3.1.3. Li+fractionation performance of SNF process

As discussed above,OP is the source of driving force of SNF process while FFR has a direct influence on the concentration phenomenon of the process. Therefore, to further determine the Li+fractionation performance,SNF was conducted with the optimized OP of 2.4 MPa and optimized FFR of 140 L?h-1. The concentrations of Li+and Na+in the SNF permeate were 38.7% and 17.3% higher than those in the feed, respectively, which was mainly due to the negative rejection of Li+and Na+(Fig. 4(a)). Meanwhile, the Mg2+concentration in SNF permeate was 54.8% lower than that in feed,which meant that an excellent Mg2+/Li+separation performance was obtained through SNF. Although the water recovery of the SNF process was only about ～35.7%, the Li+recovery ratio could reach 49.5% due to the negative rejection of Li+of the SNF membrane (Fig. 4(b)). In addition, the mass ratio of Mg/Li was reduced by 67.2% with the ESECof 185.2 kW?h?kg-1.

3.2. Li+ fractionation from Mg2+ via the SED process

Fig. 3. Effects of feed flow rate on ion fluxes (a), selectivity coefficient, Mg/Li ratio, and Li+ recovery ratio (b) during SNF process.

Fig. 4. Composition of feed and SNF permeate (a) and Li+ fractionation performance of SNF process at optimal operating parameters (b).

Although the separation of Mg2+/Li+using the SED process has been studied by many researchers, most SED operational modes have been in batch mode, which cannot achieve continuous Li+fractionation from Mg2+[10,28]. However, SNF, as introduced earlier can treat high Mg/Li ratio brine continuously.Therefore,to better illustrate the difference between the two processes,SED in this study adopted a feed-and-bleed mode to realize continuous treatment of high Mg/Li ratio brine.For SED with feed-and-bleed mode,CPV, RFR and circulating flow rate were typical influencing factors[7]. In this section, the influence of CPV and RFR on the performance of Mg2+/Li+separation were investigated in detail with a constant circulating flow rate of 300 L?h-1(corresponding to the linear flow velocity of 3.27 cm?s-1calculated by Eq. (S6)), based on results from preliminary experiments.Subsequently,with optimized operating parameters, Li+fractionation performance of SED was discussed.

3.2.1. Influence of cell-pair voltage

The influence of CPV on the performance of Mg2+/Li+separation was examined by conducting SED with the CPV ranging from 0.8 to 1.2 V and a constant RFR of 8 L?h-1.The current density(CD,calculated by Eq.(S7))in the steady state with high CPV was higher than that with low CPV (Fig. S3). Meanwhile, the conductivity of dilute overflow in steady state with high CPV was lower than that with low CPV, while the conductivity of concentrate overflow in steady state with high CPV was higher than that with low CPV. This indicated that increased CPV significantly improved SED concentration performance.

Although increased CPV meant increased driving force for ion transmembrane migration, this was not necessarily beneficial for Mg2+/Li+separation performance. Compared with a CPV at 0.8 V,Li+fluxes with a CPV at 1.0 and 1.2 V increased 15.8% and 17.4%respectively, while Mg2+fluxes with the same CPVs increased 39.0% and 82.9%, respectively (Fig. 5(a)). The significant difference in the increased rate of the two ion fluxes was mainly caused by the Mg2+concentration being much higher than the Li+concentration in the feed and the driving force of Mg2+enhanced with increased CPV, thus impairing Mg2+rejection. In addition, a clear decline of the S(Li+/Mg2+) and remarkable increase in the Mg/Li ratio in the concentrate overflow were observed with increased CPV (Fig. 5(b)). Specifically, relative to the condition with 0.8 V CPV, the S(Li+/Mg2+) with 1.0 and 1.2 V CPV decreased 20.6% and 38.6%, respectively, while the Mg/Li ratio in the concentrate overflow with the same CPVs increased 25.6% and 65.4%, respectively.Although the separation performance of Mg2+/Li+was weakened by increased CPV, the recovery ratio of Li+with higher CPV was much higher than that with 0.8 V CPV. In summary, to balance the Mg2+/Li+separation performance and Li+fractionation efficiency, the optimal CPV was determined to be 1.0 V.

3.2.2. Influence of replenishment flow rate

The influence of RFR was further examined with RFR ranging from 8 to 20 L?h-1and a constant CPV of 1.0 V. The increase of RFR made the CD and conductivities of the dilute overflow and concentrate overflows with RFR at 14 and 20 L?h-1higher than those with RFR at 8 L?h-1(Fig. S4). Clearly, increased RFR meant that more ions in the dilute compartments were able to migrate through the IEMs in the steady state,which was not only beneficial for SED concentration performance, but also conducive for the Mg2+/Li+separation process.

Compared with an RFR at 8 L?h-1,the Li+flux with RFR at 14 and 20 L?h-1increased 32.0%and 68.5%,respectively,while the Na+flux at the same RFRs increased 50.1% and 69.0%, respectively (Fig. 6(a)). However, different from the flux variation of monovalent cations with increased RFR,the Mg2+flux at 14 L?h-1RFR was only 19.8% higher than that at 8 L?h-1, while the Mg2+flux of 20 L?h-1was closed to that at 8 L?h-1. Meanwhile, to maintain the electro neutrality, the decrease of the Mg2+flux of 20 L?h-1resulted in the obvious decrease of Cl-flux,and finally caused that the Cl-flux of 20 L?h-1was a bit lower than that of 14 L?h-1.These phenomena indicated that Mg2+transmembrane migration was significantly weakened by increased RFR, which was mainly caused by the formation of an electric double layer (EDL) at the solution-MSCEM interface [19,23]. During SED, positively-charged EDLs were formed in dilute compartments due to Mg2+accumulation at solution-MSCEM interfaces. Considering that the charge density of Mg2+was much higher than that of Li+,Mg2+was more repulsed by positively-charged EDLs than Li+. Meanwhile, positivelycharged EDLs were intensified by increased RFR from increased Mg2+concentrations in dilute compartments. Therefore, the separation performance of Mg2+/Li+at higher RFR was superior to that in lower RFR. Specifically, when RFR increased from 8 L?h-1to 14 and 20 L?h-1, the S(Li+/Mg2+) increased 24.2% and 74.4%, respectively, while the Mg2+/Li+in concentrate overflow decreased 17.7%and 40.8%,respectively(Fig.6(b)).However,it was observed that the Li+recovery ratio dramatically declined with increased RFR.Compared with 8 L?h-1RFR,the Li+recovery ratio with RFR at 14 and 20 L?h-1decreased from 80.9% to 69.1% and 55.8%, respectively. Therefore, the optimal RFR was selected at 14 L?h-1in this study.

Fig. 5. Effects of cell-pair voltage on ion fluxes (a), selectivity coefficient, Mg/Li ratio, and Li+ recovery ratio (b) during SED process.

Fig. 6. Effects of replenishment flow rate on ion fluxes (a), selectivity coefficient, Mg/Li ratio, and Li+ recovery ratio (b) during SED process.

3.2.3. Li+fractionation performance of SED process

For the SED process,increased CPV not only facilitated Li+transmembrane migration but also weakened the Mg2+rejection effect of MSCEMs. In addition, although increased RFR intensified the CD of EDLs adjacent to MSCEMs and thus improved Mg2+/Li+separation performance, the Li+fractionation efficiency significantly declined with increased RFR. Therefore, based on optimization of CPV and RFR, SED was further examined to determine its Li+fractionation performance. In steady state SED, although the concentrations of all ions in concentrate overflow were much higher than those in the feed, the fold of concentration of these ions in the concentrate overflows were different (Fig. 7(a)). Specifically,the concentrations of Li+and Na+in the concentrate overflows were 6.05 and 6.07-fold of those in the feed, while the Mg2+concentration in concentrate overflows was only 2.30-fold that in the feed. This indicated a clear permselectivity of MSCEMs between mono and divalent cations. In terms of SED Li+fractionation performance, the Li+flux and S(Li+/Mg2+) were 0.2020 mol?h-1?m-2and 2.63, respectively (Fig. 7(b)). Meanwhile,the mass ratio of Mg2+/Li+in concentrate overflows was reduced by 67.2% compared with that in the feed and the recovery ratio of Li+was 69.1% with the ESECat 154.2 kW?h?kg-1.

3.3. Differences of fractionating Li+ from Mg2+ with SNF and SED processes

Regardless of the SNF or SED process,the separation and purification of monovalent ions is always accompanied by divalent ion leakage, while the difference between the two processes in their driving forces causes differences in the migration characteristics of mono/divalent ions. For the Li+fractionation process from high Mg/Li ratio brine with the separation and purification of Li as a primary goal, the macroscopic lithium extraction performance and energy consumption need to be considered. Thus, the technical indicators concerned should also clarify the selectivity difference and relative reasons of the two processes. Therefore, five primary performance parameters, including S(Li+/Mg2+), P(Li), R(Li), ε(Li),and ESEC, were selected to systematically illustrate the differences of the two processes.These parameters,under relatively optimized operating conditions, vividly reflected the differences of the two processes (Fig. 8). First, S(Li+/Mg2+) of SNF was 24.7% higher than that of SED and,consequently,the Mg/Li ratio in the purified solution of SNF was 19.0% lower than that of SED. This indicated that the Mg2+/Li+permselectivity of SNF was superior to that of SED.However, the P(Li) of SNF was only a little higher than that of SED, meaning that the Li+fractionation capacities of the two processes were very similar. Although the difference in P(Li) of the two processes was not clear, SED was much superior to SNF in terms of R(Li), at 69.1% and 49.5%, respectively. Meanwhile, compared with SNF, SED had a distinct advantage in Li+concentration effects. Specifically, the ε(Li) of SED and SNF were 6.05 and 1.39,respectively. As mentioned before, the driving force for SNF is the transmembrane pressure difference and exerted directly on the brine solution, while that for SED is an electric field force and exerted on ions in brine solution [23]. This made the SED driving force for Li+migration much higher than that in SNF and, thus,SED not only had a clear advantage in Li+extraction efficiency,but also had a dominant position in Li+concentration effects. In addition,in terms of energy consumption,SED was also better than SNF and the ESECof the former 20.1% lower than that of the latter.Therefore, considering the S(Li+/Mg2+) of the two processes, SNF was the first choice when a better purified performance is preferred. However, SED was a more suitable option if R(Li), ε(Li),and ESECwere considered.

Fig. 7. Composition of feed and SED concentrate overflow (a) and Li+ fractionation performance of SED process at optimal operating parameters (b).

Fig. 8. Differences of fractionating Li+ from Mg2+ with SNF and SED.

4. Conclusions

The separation of Mg2+/Li+in brine with a high Mg/Li ratio has gradually attracted attention while SNF as well as SED,as two typical membrane separation processes for monovalent/divalent ion separation, have been extensively studied. In this investigation,for the first time,the differences of SNF and SED processes in fractionating Li+from a high Mg/Li ratio brine were systematically illustrated. First, the influence of typical operating parameters for the two processes were studied and optimized. Specifically, the optimal OP and FFR for SNF were 2.4 MPa and 140 L?h-1, respectively, while the optimal CPV and RFR for SED were 1.0 V and 14 L?h-1, respectively. Subsequently, based on these optimized operating parameters,the Li+fractionation performance of the two processes were determined.In terms of purified solution composition,the Mg/Li ratio in purified solution of SNF was 19.0% lower than that of SED due to the higher S(Li+/Mg2+) of the former. Although the P(Li) of the two processes were similar, the R(Li) of SED was 39.6% higher than that of SNF. Moreover, ε(Li) of SED was 4.35-fold that of SNF,while ESECof the former was 20.1%lower than that of the latter. Therefore, SNF was found more suitable for Li+fractionation due to its higher S(Li+/Mg2+).It also should be noted that SED was recommended as best when Li+extraction efficiency was considered. This study provided a more intuitive understanding and guidance for the application and improvement of the two technologies in the field of Mg2+/Li+separation.

CRediT Authorship Contribution Statement

Yong Xu: Investigation, Methodology, Data curation, Writing –original draft. Qingbai Chen: Methodology, Investigation,Resources, Funding acquisition, Writing - review & editing. Yang Gao: Methodology, Data curation. Jianyou Wang: Conceptualization, Project administration, Writing – review & editing. Huiqing Fan: Methodology, Data curation, Resources. Fei Zhao: Resources.

Data Availability

Data will be made available on request.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors are grateful for research financial support by the National Key Research and Development Program of China(2017YFC0404003), the Tianjin Natural Science Foundation(21JCZDJC00270), the China Postdoctoral Science Foundation(2021M701875), the Tianjin Special Project of Ecological Environment Management Science and Technology (18ZXSZSF00050),the Tianjin Science and Technology Support Project(19YFZCSF00760) and the Fundamental Research Funds for the Central Universities (63221312).

Supplementary Material

Supplementary material to this article can be found online at https://doi.org/10.1016/j.cjche.2022.11.013.