Accepted Manuscript

Lithium ion-selective membrane with 2D subnanometer channels

Amir Razmjou, Ghazaleh Eshaghi, Yasin Orooji, Ehsan Hosseini, Asghar HabibnejadKorayem, Fereshteh Mohagheghian, Yasaman Boroumand, Abdollah Noorbakhsh,Mohsen Asadnia, Vicki Chen

PII: S0043-1354(19)30398-7

DOI: https://doi.org/10.1016/j.watres.2019.05.018

Reference: WR 14667

To appear in: Water Research

Received Date: 27 February 2019

Revised Date: 30 April 2019

Accepted Date: 5 May 2019

Please cite this article as: Razmjou, A., Eshaghi, G., Orooji, Y., Hosseini, E., Korayem, A.H.,Mohagheghian, F., Boroumand, Y., Noorbakhsh, A., Asadnia, M., Chen, V., Lithium ion-selectivemembrane with 2D subnanometer channels, Water Research (2019), doi: https://doi.org/10.1016/j.watres.2019.05.018.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

https://doi.org/10.1016/j.watres.2019.05.018https://doi.org/10.1016/j.watres.2019.05.018https://doi.org/10.1016/j.watres.2019.05.018

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

1

Lithium ion-selective membrane with 2D subnanometer channels 1

Amir Razmjoua,b*,1, Ghazaleh Eshaghia,1, Yasin Oroojic, Ehsan Hosseinid, Asghar Habibnejad 2

Korayemd, Fereshteh Mohagheghiana, Yasaman Boroumanda, Abdollah Noorbakhsha, 3

Mohsen Asadniae, Vicki Chenf,b 4

aDepartment of Biotechnology, Faculty of Advanced Sciences and Technologies, University 5

of Isfahan, Isfahan 73441-81746, Iran 6

bUNESCO Centre for Membrane Science and Technology, School of Chemical Science and 7

Engineering, University of New South Wales, Sydney, 2052, Australia 8

cCollege of Materials Science and Engineering, Nanjing Forestry University No. 158, 9

Longpan Road, Nanjing, 210037 Jiangsu, People's Republic of China 10

dSchool of Civil Engineering, Iran University of Science and Technology, Tehran, Iran 11

eDepartment of Engineering, Macquarie University, Sydney, New South Wales 2109, 12

Australia 13

fSchool of Chemical Engineering, University of Queensland, St. Lucia, 4072, Australia 14

Abstract 15

In the last two years, the rapidly rising demand for lithium has exceeded supply, 16

resulting in a sharp increase in the price of the metal. Conventional electric driven 17

membrane processes can separate Li+ from divalent cations, but there is virtually no 18

commercial membrane that can efficiently and selectively extract Li+ from a solution 19

containing chemically similar ions such as Na+ and K+. Here, we show that the different 20

movement behavior of Li+ ion within the sub-nanometre channel leads to Li+ ion-21

selectivity and high transport rate. Using inexpensive negatively charged 2D 22

subnanometer hydrous phyllosilicate channels with interlayer space of 0.43 nm in a 23

membrane-like morphology, we observed that for an interlayer spacing of below 1 nm, 24

1 These authors contributed equally to this work.

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

2

Li+ ions move along the length of the channel by jumping between its two walls. 25

However, for above 1 nm spacing, the ions used only one channel wall to jump and 26

travel. Molecular dynamic (MD) simulation also revealed that ions within the 27

nanochannel exhibit acceleration-deceleration behaviour. Experimental results showed 28

that the nanochannels could selectively transport monovalent ions of Li+> Na+> and K+ 29

while excluding other ions such as Cl- and Ca2+ , with the selectivity ratios of 1.26, 1.59 30

and 1.36 for Li+/Na+, Li+/K+, and Na+/K+ respectively, which far exceed the mobility 31

ratios in traditional porous ion exchange membranes. The findings of this work provide 32

researchers with not only a new understanding of ions movement behavior within 33

subnanometer confined areas but also make a platform for the future design of ion-34

selective membranes. 35

Keywords: Subnanometer channels; Li ion selective membrane; two-dimensional materials; 36

lithium extraction; vermiculite 37

1 Introduction 38 It is predicted that the supply of the energy-critical element of lithium will soon fall below its 39

continuously increasing demand, which will render it a strategically influential 40

element(Choubey et al. 2017). The amount of Li+ in seawater is estimated to be nearly 41

230,000 million tons−almost 57000 times higher than its abundance on land(Yoshizuka et al. 42

2006). However, the low concentration of Li+ and its coexistence with chemically similar 43

ions such as Na+ and K+ in seawater makes the extraction of Li+ from this source very 44

challenging. 45

Extraction of Li+ using membrane technology from seawater, lakes or geothermal brine as 46

secondary Li resources is highly sought-after, particularly for fast-growing industries such as 47

electric vehicles and lithium-ion batteries. Due to the poor monovalent selectivity of 48

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

3

conventional membrane technologies such as nanofiltration or electrodialysis using cation 49

exchange membranes, they can only be used to concentrate Li+ by removing all the divalent 50

ions. A membrane that could transport certain ions such as monovalent cations (Li+, Na+, and 51

K+) while excluding other ions would be highly valued. 52

For the design of a Li+ selective nanochannel/membrane, the two parameters of nanochannel 53

structure and chemistry must be adjusted carefully (Razmjou 2019). Nanochannel 54

dimensions, i.e., interlayer spacing, length, and symmetric or asymmetric channel 55

morphology, along with the inner surface charge, play a vital role in the selective 56

transportation of Li+ ions. Many comprehensive reviews have discussed the ion transport 57

mechanisms in nanopores and nanochannels (Daiguji 2010, Horike et al. 2013, Oener et al. 2017, 58

Tagliazucchi and Szleifer 2015). Membrane driving force (pressure or potential difference) is 59

also another key parameter that should be taken into account. 60

For a neutral nanochannel, the channel dimension must be smaller than the Li+ hydration 61

diameter (~0.76 nm). Otherwise, the channel will not exhibit Li+ selectivity (>1) among alkali 62

metal ions (Guo et al. 2016, Zhang et al. 2018b). It is reported that for neutral sub-nanometre 63

channels, the main mechanism for small monovalent ions to move is based on the partial 64

dehydration and loss of their hydration layers (Razmjou 2019). When the nanochannel 65

surfaces hold negative charges, the partial dehydration is not the only ion conductivity 66

mechanism, and the ions’ affinity to the functional groups also comes into play. The channel 67

dimension then can be higher than (~0.76 nm). 68

Playing with the design above parameters, recently researchers examined a range of building 69

blocks to prepare Li+ ion-selective membranes. Metal/organic frameworks such as UIO-66 70

(Zhang et al. 2018a), MOP-18(Jung et al. 2008), ZIF-8 (Zhang et al. 2018a), ZIF-7 (Zhang et al. 71

2018a), and sulfonated HKUST-1(Guo et al. 2016) have recently exhibited Li+ ion selectivity. 72

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

4

MOFs, particularly when they are threaded with negative ligands, revealed high Li+/Na+ or 73

Li+/K+ selectivities and great potential. However, stable defect free flexible MOF thin film is 74

a significant ongoing issue. Researchers have tried to oxidize graphene sheets (GO) partially 75

and stack them to create membranes with 2D nanochannels for precise ionic and molecular 76

sieving in aqueous solution (Han et al. 2013, Joshi et al. 2014, Liu et al. 2015). Recently, 77

Cseri et al.(Cseri et al. 2018) prepared a mechanically robust and highly permselective anion 78

exchange membranes based on GO and polybenzimidazolium nanocomposite. Their 79

membrane exhibited high ion exchange capacity (1.7–2.1 mmol g−1), with exceptional 80

permselectivity (up to 0.99). GO membranes also showed Li+ ion selectivity (Abraham et al. 81

2017, Joshi et al. 2014, Zhao et al. 2018). However, difficulty in reducing the interlayer spacing 82

of graphene oxide membranes, and maintaining this space during swelling when immersed in 83

the aqueous solution, have been considered the main challenges hindering potential ion 84

filtration applications of graphene oxide membranes (Goh et al. 2015, Huang et al. 2013, 85

Hung et al. 2014, Su et al. 2014, Sun et al. 2016). Creating Li+ ion selective nanochannels on 86

polymeric membranes such as PET (Wang et al. 2018, Wen et al. 2016, Zhang et al. 2018a) using 87

ion irradiation and UV radiation is currently a lab scale process and requires expensive 88

infrastructure. 89

Although the proposed Li+ ion-selective membranes are promising, there are several 90

challenges that keep them from meeting the industry requirements. The three major 91

challenges are 1) scalability, 2) high cost of the building blocks, and 3) a lack of fundamental 92

understanding Li+ ion behaviour in nanoconfined areas. 93

As mentioned ion transportation mechanisms inside nanochannels have been well-studied. 94

However, the mechanisms by which Li+ ions show higher transportation rates and thus 95

selectivity in some designed nanochannels than over Na+ or K+ ions is not yet clear. This 96

uncertainty and poor understanding, particularly with regards to dehydration energy barrier, 97

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

5

has resulted in contradictory conclusions and discussions. Therefore, more experimental and 98

theoretical explorations are needed to understand what dictates Li+ ion to transport within the 99

nanochannel at a higher rate than Na+ or K+. 100

The concept of this work is to exploit 2D nanofluidic vermiculite (VCT) channels into a 101

morphology which can be used to control ion transportation across the membrane selectively. 102

Here, VCT is selected because it has significant advantages compared with GO and other 103

two-dimensional structures. VCT with its extraordinary chemical and thermal stability is 104

much more readily available and much less costly (Orooji et al. 2017). It can be exfoliated 105

easily in water via thermal heat treatment and ionic exchange, as opposed to the exfoliation 106

procedure of graphene and other 2D materials. Our novel and strategically-designed 107

membranes were produced from vastly available inexpensive VCT minerals (~20 USD/kg 108

from Sigma Aldrich and ~20 CNY/kg from Chinese suppliers).To prepare VCT membranes 109

we introduced pressure-assisted vacuum deposition techniques that can be easily scaled up. 110

Recently, Shao et al. (Shao et al. 2015) demonstrated that such 2D nanofluidic channels could 111

show superionic proton conductivity. Here, we show for the first time another fascinating 112

property of 2D nanofluidic VCT channels: selective transportation of Li+. We also used 113

molecular dynamic (MD) simulation to understand the mechanism by which Li+ ions 114

transport through VCT nanochannels. 115

The outcomes of this work provide researchers with a new understanding of Li+ions 116

movement behavior within subnanometer confined areas, and also make a platform for the 117

future design of ion-selective membranes 118

119

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

6

2 Experimental 120

2.1 Materials 121

Vermiculite (VCT) (CAS No.: 0001318009) Calcium chloride (CaCl2, 97%), Potassium 122

chloride (KCl, 99%) Sodium chloride (NaCl, 99.5%), and Lithium chloride (LiCl, 99%) were 123

purchased from Sigma-Aldrich. Hydrochloric acid (HCl, 37%), Hydrogen peroxide (H2O2 124

35%) and ethanol (96%) were sourced from a local provider. 125

2.2 Methods 126

Details of VCT membrane preparation, nanofluidic device fabrication, membrane 127

monovalent cations conductivity measurement and analysis of desalination performance of 128

VCT membranes are presented in “Methods” in Supporting Information. A variety of 129

techniques, i.e. AFM, SEM, TEM, BET, and XRD were used to characterize the membranes 130

(see Supporting Information for more details). 131

3 Results and discussions 132

3.1 Preparation and characterization of Li+ ion selective VCT membrane 133 In this work, the exfoliation of VCTs followed the approaches used by Walker and Garrett 134

(Walker and Garrett 1967) and Obut and Girgin (Obut and Girgin 2002). Briefly, the 135

exfoliation was performed based on a rapid thermal shock and subsequent several ion 136

exchange steps and a final hydrogen peroxide treatment (See Figure S1a). As shown in 137

Figure 1a, VCT is known as the layered magnesium aluminosilicate compound. Each layer 138

has a negative charge (1.4 mCm-2) and consists of one Mg-based octahedral sheet that is 139

sandwiched between two tetrahedral silicate sheets (Valaskova and Martynkova 2013). The 140

substitutional Al+3 cations in the tetrahedral sheets give rise to the negative charges on the 141

layers. These negative charges are balanced by cations between the layers. During the 142

thermal shock, the interlayer water rapidly vaporizes and expands the crystals into an 143

accordion-like structure (see Figure 1b and c). XRD and BET results exhibited in Figure 1d 144

and 1e revealed that the thermal shock at elevated temperature could destroy the VCT 145

structure. The disappearance of XRD peaks in the 2θ range of 5-15o which correspond to K 146

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

7

and Mg-VCT and hydration layers(Marcos et al. 2009) alongside with a significant reduction 147

in the BET surface area and BJH pore volume (from 44.59 m2/g and 0.104 cm3/g at 900 °C to 148

15.03 m2/g and 0.0108 cm3/g at 1000 °C) confirm the possibility of structural VCT unit 149

collapse (Orooji et al. 2017). During the ionic exchange process, the interlayer distances 150

increase by replacing the interlayer cations such as Na+, K+, and Mg2+ with Li +, which has a 151

larger hydration diameter. During the final hydrogen peroxide treatment, the oxygen 152

evolution from its decomposition helps to exfoliate the VCT layers further. As can be seen in 153

Figure 1f, the thickness of the structural unit (two magnesium aluminosilicate layers plus an 154

interlayer space) is approximately 0.83±0.06 nm with interlayer space of 0.43±0.05 nm.” The 155

values were obtained using an image processor software provided by the TEM instrument. 156

The d-spacing was calculated using XRD data and the Bragg equation. The inset image in 157

Figure 1f shows that the thermal and chemical exfoliation processing led to a light brown 158

colloidal suspension of VCT layers, as has also been observed elsewhere(Shao et al. 2015). 159

The nanofluidic membrane can be readily prepared by reassembly of the exfoliated VCT 160

layers using different approaches, including filtration, dip-coating, spraying and solvent 161

evaporation (Ballard and Rideal 1983, Shao et al. 2015). In this work, the filtration technique 162

was used to stack the exfoliated 2D sheets (see Methods). Figure 2a shows the flexible free-163

standing VCT membrane (1 mm thickness) obtained by the vacuum filtration process. SEM 164

and AFM images in Figure 2b show the cross-section of our membrane with lamellar 165

structure. It appears that the membrane consists of a stack of 40 nm layers (flakes) such that 166

each layer consists of many 0.83 nm structural units (see Figure 2c). XRD results in Figure 167

2d are attributed to the raw VCT particles and our membrane. As can be seen, the absence of 168

two strong peaks below 2θ of 10° and the appearance of one peak around 6° (which is 169

attributed to the lithium-intercalated VCT) indicate that the original interlayer cations (Mg2+, 170

K+) have been successfully exchanged by Li+. 171

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

8

Following Sessile drop technique used in our previous works (Orooji et al. 2018, Shirani et 172

al. 2017), the water contact angles of the VCT membranes were measured. It was observed 173

that the water droplet disappeared immediately within a second after touching the 174

membranes. A broad definition of superhydrophilicity which is mostly referred to in the 175

literature is achieving zero contact angles within the first 5 sec (Razmjou et al. 2011). 176

Therefore, our membrane is considered as the superhydrophilic membrane. 177

The water uptake and swelling ratio were measured following our previous works with some 178

amendments ((Razmjou et al. 2013a, Razmjou et al. 2013b, Razmjou et al. 2013c)). Briefly, a 179

piece of the membrane (1 by 2 cm) without PDMS sandwiching was placed on a wet tissue to 180

gradually absorb water molecules. The sample was weighed every few mins until it reached 181

saturation and thus its weight was unchanged. The water uptake Q of the membrane was 182

calculated as Q = (Wt-Wd)/Wd×100, where Wt is the weight (g) of the wet membrane at the 183

end of the experiment and Wd is the weight (0.04g) of the initial dry sample before the test. 184

The sample did uptake water 50% of its dry weight. The swelling ratio of the membrane was 185

also calculated based on the volumetric changes of the sample after water uptake, and it was 186

found 30%. It should point out here the PDMS sandwiched sample showed insignificant 187

swelling ratio as expected. 188

189

3.2 Li+ ion selectivity 190 The electric-field-driven ion transport in VCT nanochannel was analyzed by linear sweep 191

voltammetry (L.S.V) using LiCl, NaCl, KCl, CaCl2 solutions, and an in-house cell (see 192

Figure S1b). The ionic conductivity of the fabricated membrane can be estimated from the 193

slope of the I-V curves in the range of potentials within which no redox reactions can occur 194

(Shao et al. 2015, Zhang et al. 2018b) . As can be seen in Figure 3a, the slope of I-V curves 195

remained zero for dry VCT membrane, and also when deionized water was used as the 196

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

9

electrolytic. The slope values reduced for HCl, LiCl, NaCl, and KCl and approached zero 197

when CaCl2 solution was used. An insignificant number of Ca2+ divalent or Cl- anions were 198

able to pass through the VCT nanochannel as the independent movement of ion species 199

allows the passage of current through the solution. 200

It should be noted that in the applied potential range (-0.1 to 0.3 V), electrolysis of water 201

molecules cannot occur. So any change in the I-V curve is due to the transportation of the 202

corresponding cations. The figure also shows that the ion conductivity increases as the ion 203

sizes reduce, such that H+ and K+ showed the highest (0.1 S.cm-1) and lowest (0.04 S.cm-1) 204

conductivity at 0.0001 M related electrolyte solutions respectively. 205

Electrochemical impedance spectroscopy (EIS) is the most common AC technique used to 206

characterize electrochemical properties (ion conductivity) of the membrane [2, 3]. In this 207

technique, the ohmic resistance of the cell can be extracted by applying an alternating current 208

(or voltage) driving force and measuring the magnitude and phase of the cell voltage (or 209

current) to determine the complex impedance of the system. As can be seen in Figure 3b and 210

Figure S2a, Nyquist plots of proposed membrane versus Li-ion concentration (ranging from 211

0.0001 to 1 M) using EIS revealed that the ionic conductivity increases by increasing the 212

electrolyte solutions. 213

Currently, there is no commercial synthetic membrane with atomic-sized pores that can 214

efficiently separate monatomic ions (for instance Li+, Na+, and K+) with the same valence and 215

similar sizes. Having such a membrane is highly demanded, particularly for mineral 216

extraction and ion batteries. To examine the Li+ ion selectivity of VCT membranes, the ion 217

selectivity ratio of Mi/Mj was calculated from the equation(Zhang et al. 2018b) below, 218

������������ =�������.��

�������.�� (1) 219

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

10

where M is a cation, and i and j stand for Li, Na, and K at +0.2 V. Using the I-V curves, the 220

selectivity ratios of Li+/Na+, Li+/K+ and Na+/K+ were calculated as 1.26, 1.59 and 1.36, 221

respectively. These ratios are much greater than the corresponding mobility ratios in water. 222

For example, the mobility ratio of Na+ /K+ in water is around 0.7, which is significantly 223

smaller than the value for our VCT membranes (1.36). The selectivity ratios of the binary 224

ions were evaluated using ICP-MS. As can be seen in Table S1, the selectivity ratios are close 225

to those calculated by I-V curves and Equation 1.Figure S2b and Table S2 show a 226

comparison between the selectivity performance of VCT membranes and similar techniques 227

described in the literature. As can be seen, the VCT membranes with 2D nanochannels 228

showed either higher ion selectivity performance or a much simpler synthesis route. Table S3 229

also compares the alkali ion selectivity performance of VCT membrane with other synthetic 230

nanochannels. Apart from ZIF8/GO/AAO membranes introduced in Ref. (Zhang et al. 2018b), 231

which has similar ion selectivity performance but a complex, expensive synthetic route, our 232

VCT membranes showed significantly superior performance. It should be noted that the 233

ZIF8/GO/AAO showed anion conductivity (calculated ion mobility of ~2.5 for Cl-(Zhang et 234

al. 2018b)) whereas VCT membranes prevent the transport of anions. 235

It is well-proven that ion conductivity normally increases with temperature. The effect of 236

electrolyte temperature (up to 90 oC) on the VCT membranes’ performance was studied in 237

Figure S3. As can be seen, the performance of the membrane increased slightly without 238

destroying the VCT unit structure, indicating high thermal stability. Shao et al. also studied 239

the effect of high temperature (500 oC) treatment on VCT membrane(Shao et al. 2015). They 240

showed that the nanochannels of VCT exhibit extraordinary thermal stability such that they 241

can maintain their proton conduction function even after annealing at 500 °C in air. Lack of 242

thermal stability of available commercial polymeric membranes has limited their 243

applications. The effect of membrane orientation was examined to see if there is any ion 244

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

11

transport perpendicular to the direction of reconstructed layers (horizontal vs. vertical ion 245

transport, see inset in Figure 3c). As shown in Figure 3c, ion flow is mainly along the 246

membrane, but it can also flow across it through defects in its structure. Although horizontal 247

transportation of ions along the membrane is dominant, vertical transport may be desired for 248

applications such as water purification, when a high effective membrane surface area is 249

needed to increase footprint. As shown in the inset image in Figure 2a, the VCT membrane 250

was bent for 24 hr then tested to examine any mechanical tension induced structural damage. 251

No significant difference in the I-V curves of the VCT membrane before and after bending 252

was observed (refer to Figure S4a). The effect of membrane thickness on the ion 253

transportation was also investigated in Figure S4b. Results indicated that reducing the 254

membrane thickness by half led to a marginal reduction in ion conductivity. This indicates 255

the VCT membrane has good ion transportability. 256

During the selective transportation of ions, there is a chance that some Li+, Na+ or K+ remain 257

in the interlayer space, and may result in a change in the degree of hydration for the cation. 258

This, in turn, could lead to a change in the interlayer spacing, thus decreasing the 259

performance of the membrane. In order to determine if the membrane performance remained 260

stable, a test of transportation of Na+ (NaCl 1 M) over a number of hours was conducted. 261

This experiment was repeated four times. Samples were also immersed in a solution of NaCl 262

1 M for six months and then tested. As can be seen in Figure S5a, the membranes exhibited 263

long-term stability. Energy Dispersive X-ray (EDX) microanalysis (Bruker Nano Berlin, 264

Germany) was performed on the 6-month VCT membrane to examine Na ion trace. As can be 265

seen in Figure S5b, the associated K peak was not observing, confirming the pervious 266

result.The performance of VCT membrane was also compared with its two commercial 267

counterparts of Nafion and CMI-7000S. As shown in Figure 3d, at 1 M HCl, the VCT 268

membrane portion conductivity is similar to that for a Nafion membrane. Nafion membranes 269

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

12

have exhibited ion mobility of divalent cations inside themselves.(Okada et al. 1998), and 270

show very poor Li+ ion selectivity (see Table S2). The ion conductivity of our 2D nanofluidic 271

VCT membranes at 0.1 M HCl is about 0.165 S.cm-1 compared with that of graphene oxide 272

membrane which is 0.075 at 0.1 M HCl. The proton conductivities of commercial PEMs of 273

Nafion and CMI-7000S are also 0.18 and 0.075 at 0.1 M HCl. The VCT membrane has ease 274

of fabrication and cost advantages over Nafion, and shows great potential for commercial 275

implementation. 276

Li concentration in different sources varies from 0.1-0.2 ppm for seawater (Yoshizuka et al. 277

2006) to 18-2000 ppm for lakes, geothermal and oilfield brines. There are some reports of 278

lithium concentration values of >4000 ppm for the brines (Choubey et al. 2017, Choubey et 279

al. 2016). Using reverse osmosis (RO) or membrane distillation (MD) systems to concentrate 280

the brines, the concentration can go further to a higher level of 25000-35000 ppm. In this 281

study, the Li+ extraction performance of VCT membrane was reported based on the reduction 282

in electrical conductivity (EC) due to the removal of Li+ from a saline solution with a 283

different concentration of LiCl (1000, 8000, 15000, 25000 and 35000 ppm) in an electrolysis 284

cell (See Figure S1c) with a different voltage of 3-12V. The volume of the three identical 285

reservoirs in Figure S1c was 60cc, each 20cc. For each measurement, the sample solutions 286

were diluted 3 times before EC measurement. The final EC value was adjusted before 287

reporting. As shown in Figure 3e, the EC reduces over time for all of the LiCl concentrations 288

tested. Results showed that the EC drop is more pronounced at higher concentrations of LiCl. 289

Figure 3f presents the effect of the change in applied power density (V). As can be seen in the 290

figure, the extraction rate increased significantly as the voltage increased, particularly at high 291

LiCl concentration, indicating a good efficiency for salt lakes Li+ extraction. Similar 292

experiments were performed for Na+ and K+ (see Figure S1 d-g). Also, EC of permeates was 293

measured for NaCl feed solution with an initial concentration of 8000 ppm with 2, 4, and 6 hr 294

electrodialysis (see Figure S1h). As can be seen, the diluted permeate ECs are almost half of 295

the feed solutions as the volume of the two reservoirs (3 and 1) is twice the feed reservoir. 296

Considering the fact that the ppm value of a sodium chloride solution happens to be very 297

close to half of its conductivity value (in milliSiemens/cm), we performed a simple mass 298

balance to confirm the results. From the Table in Figure S1h, after 4 hr of electrodialysis, the 299

concentration of NaCl in feed is around 500 ppm while that of the permeate is around 244 300

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

13

ppm. Considering that the feed volume reservoir is half of the permeate volume, there is a 301

near mass balance between feed and permeate. 302

303

304

305

3.3 Li+ ion transport mechanism and MD simulation 306 The proton conductivity of the VCT membrane was measured using the L.S.V technique and 307

apparatus which was shown in Figure S1b (see methods). Figure 4a represents I-V curves 308

recorded at different pH (different concentrations of HCl). From the figure, the higher the 309

proton concentration is, the higher the ion conductivity becomes, and the higher the slope of 310

the I-V curve. This higher conductivity at a higher concentration of ions was also observed 311

for KCl, NaCl and LiCl solutions (see Figure S6). Since the three electrolytes share the same 312

anion (Cl−), the differences of the ionic currents in I-V curves must be caused by the cations. 313

Figure 4b shows the pH-dependent proton conductivity of VCT membrane in comparison 314

with that of the bulk HCl solution. The proton conductivity of bulk is directly proportional to 315

the concentration of HCl. At low pH, the conductivity values of the membrane were similar 316

to that of nanofluidic channels, whereas at high pH the values deviated from the bulk 317

behavior. It seems that in the acidic regime the nanochannel conductivity above pH 5 was 318

somewhat constant, and was independent of the concentration of bulk. This pH-independent 319

conductivity is a signature of surface charge-governed transport(Shao et al. 2015), and also is 320

an indication of good sealing of VCT membrane by PDMS. At high pH values (basic 321

regime), the Debye length near the negatively charged VCT nanochannel is greater than the 322

interlayer spacing between the tetrahedral silicate sheets. Therefore, cations between the 323

layers are the dominant charge carrier, and the surface charge density determines their 324

concentration inside the channels rather than bulk electrolyte concentration(Salieb-Beugelaar 325

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

14

et al. 2009). This greater-than-bulk proton conductivity behavior was also observed for LiCl, 326

NaCl, and NaCl (see Figure S7). 327

328

329

330

MD simulation was performed to understand the effect of VCT nanochannel dimensions and 331

external electric field density on the Li+ ion transportation mechanism. In this work, all 332

molecular dynamics simulations were performed using Non-equilibrium, in the NVT 333

ensemble at a temperature of 298.15 K, maintained using the Nose-Hoover thermostat (please 334

refer to Supporting Information for more details). The simulation revealed that reducing the 335

nanochannel dimensions from nanometer to subnanometer scale resulted in two interesting 336

phenomena: a reduction in water density (number of H2O/nm3), and a change in Li+ ion 337

jumping transportation behavior from two-surface to one-surface charge-governed transport 338

mode. 339

Studies using density functional theory revealed that strong electric fields have a significant 340

impact on the structure and energetics of small Li+-water clusters (Daub et al. 2015). In the 341

absence of an electric field, the Li+.nH2O forms a symmetry of n = 4 tetrahedral energy 342

minimum structure. According to Daub et al. (Daub et al. 2015), the small electric field 343

strength is sufficient to break the symmetry into an asymmetric planar cluster with n=3, 344

resulting in the partial dehydration of Li+ ions. A similar transition was also observed for the 345

6-coordinated cluster, Li+.6H2O to 5- and 4-coordinated clusters at field strengths of 0.2 V/Å 346

and 0.3 V/Å, respectively. As can be seen in Figure 4c and d, the confinement effect resulted 347

in a substantial reduction in the water density in the first hydration shell of Li+. The second 348

hydration shell also nearly disappeared. In agreement with Daub et al. (Daub et al. 2015), the 349

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

15

increase in applied potential from 0 to 2 V/nm resulted in the reduction in coordination 350

number and RDF peaks and higher partial dehydration degree. 351

It is well-proven that saline water expels its salt content upon freezing(Nebbia and Menozzi 352

1968). It is also reported that, in a small nanoscale confined area, water exhibits layered 2D 353

structuring and can form ordered ice-like phases (Fumagalli et al. 2018, Karna et al. 2018, 354

Winarto et al. 2017). This might be the reason for increasing the migration rate of Li+ ions 355

towards the wall surfaces of VCT nanochannels. As can be seen in Figure 5b, d, f, and h, the 356

water density and number of layers inside the nanochannel increases when the channel 357

dimension increases from 0.4 to 1.2 nm. Reduction in water density could facilitate the Li+ 358

ion movement and lead to higher ion mobility. Water molecular dipoles also aligned with the 359

electric field such that oxygen-negative-end of the dipole faced the cathode (positive) while 360

hydrogen-positive-ends of the dipole faced the anode (negative). 361

Our MD simulation revealed that Li+ ions transportation behavior within the VCT 362

nanochannels with sub-nanometre dimension is different from that of nanometre dimension 363

(>1nm) channels. When the nanochannel dimension is below ~1 nm, the Li+ ions conduct via 364

the two-surface charge-governed transport mechanism; Li+ ions jump from one of the 365

nanochannel walls to the opposite wall while they move in the electric field direction 366

(depicted in Figure 5a as migration from left to right along the nanochannel). However, when 367

the channel dimension was increased to 1.2 nm, the Li+ ions moved by hopping only along 368

one surface of the nanochannel (one-surface charge-governed transport mechanism, see 369

Figure 5g). The reason may be related to the phenomenon of spontaneous symmetry breaking 370

of charge-regulated surfaces (Majee et al. 2018). When two surfaces are equally charged 371

similarly to our VCT nanochannels, their symmetry can become spontaneously broken with 372

decreasing the inter-surface distance. Their charge densities could then differ in magnitude 373

and even in sign. The origin of spontaneous symmetry breaking of charge on VCT surface is 374

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

16

a competition between the adsorption of Li+ ions from the solution to the surface and the 375

interaction between the adsorbed ions already on the surface or the functional groups existing 376

on the nanochannel wall surface. 377

The Li+ ions moved along the length of nanochannel due to an external electric field from left 378

to right toward the negative electrode, while the ions are attracted toward the VCT surfaces 379

due to the potential of the electrical double layer (EDL). The potential is at a maximum at the 380

VCT surface and reduces exponentially to the zeta potential ƺ (between the Stern layer and 381

the diffuse layer). For a VCT nanochannel with dimensions smaller than 1 nm, the absorbed 382

Li+ ions on one of the surfaces spontaneously break the charge symmetry resulting in the 383

reduction in charge density, potential of the surface’s EDL, and even sign. Since the EDL’s 384

potential on the opposite surface is higher, the ions will be attracted to that opposite surface 385

when they jump. This spontaneous breaking of charge symmetry of the surfaces repeats so 386

that the Li+ ions can move along the structure by jumping from side to side until they finally 387

exit the nanochannel. When the channel size increased to above 1 nm, the influence of the 388

potential of the opposite surface diminished to the point that it is not strong enough to attract 389

the Li+ ion, resulting in the one-surface charge-governed transport mechanism. 390

391

This phenomenon can also be explained based on the Lennard-Jones potential (1924). Table 392

S4 shows the Lennard-Jones potential parameters of oxygen and lithium. In the table, σ is the 393

finite distance at which the inter-particle potential is zero and is 2.87 and 3.11 Å for Li and 394

surface O, respectively. Therefore the inter-particle potential of lithium and oxygen at 2.99 Å 395

is zero. Figure 6a shows the RDF of available oxygen on the VCT surface and Li+. As can be 396

seen in the figure, the first unstable and sharp peak appeared at 1.86 Å (< 2.87 Å) for an 0.4 397

nm VCT nanochannel (see inset in Figure 6a), which is an indication of the repulsive force 398

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

17

between Li+ ions and oxygens of VCT surface. This spontaneous repulsive force reduces as 399

the channel dimension increased and reached a level that Li+ ions were no longer able to jump 400

between the surfaces, so hoped only along one surface. It was also discovered that the 401

transport behavior of Li+ ions inside VCT nanochannel not only changes in sub-nanometre 402

VCT channel but also has a different velocity distribution and gradient. Figures 6b and 6c 403

show that the velocity of the Li+ ions inside VCT nanochannel was zero in the absence of an 404

external electric field (0 V/nm), meaning no ions could diffuse into the membrane. 405

In the presence of an external electric field, Li+ ions exhibited an almost linear velocity 406

gradient and acceleration at the entry to the nanochannel, followed by a sudden drop 407

(acceleration-deceleration behavior). Change in the velocity gradient is very large for sub-408

nanometre (0.4 nm) VCT nanochannel (Figure 6b) as voltage increases, whereas it is much 409

smaller for 1.2 nm VCT nanochannel (Figure 6c) due to higher water fluidity and density (see 410

Figure S9 for velocity distribution of Li+ in 0.6 and 0.8 nm VCT nanochannel). The sudden 411

drop in the velocity gradient might be related to a phenomenon known as the “ion-enrichment 412

and ion-depletion effect” of nanochannels (Pu et al. 2004). Increasing the applied potential and 413

power density of the electrical field (voltage) beyond a certain value between the two sides of 414

the nanochannel causes the depletion of ions in the entrance (at Y: 1.5 nm in Figures 6b and 415

6c) and their enrichment on the other side (at Y: 4.5 nm) resulting in an increase in 416

polarization. The accumulation of Li+ ions at Y: 4.5 nm creates a strong repulsive force that 417

substantially reduces the velocity of incoming ions. 418

Figure S10 is the results of our MD simulation to investigate the effect of change in power 419

water density on the diffusion coefficient and ion flux of cations in the VCT membrane with 420

0.4 and 0.8 nm interlayer spacing. As can be seen, increasing the power density enhanced not 421

only the Li+ transport rate but also the Ca2+ transport rate inside the VCT nanochannel. 422

However, providing such a high-power density for Ca2+ ions to overcome the dehydration 423

and crosslinking energy barriers are practically challenging and might not be economically 424

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

18

viable. Therefore, for the Li extraction application at relatively low power density, our VCT 425

membrane exhibited great potential. 426

The Li+ ions velocity is compared with those of Na+, K+, and Ca2+ in Figure 6d and Figure 427

S11. The higher Li+ ions velocity might be related to higher ion mobility of Li+. According to 428

Wen et al. (Wen et al. 2016), the partially dehydrated Li+ ions have more compact structures 429

(shorter width of the hydration shell) than Na+, K+, and Ca2+ ions, and thus exhibit higher ion 430

mobilities and conductivities through nanochannels. 431

4 Conclusion 432 The industry is urgently seeking new and efficient membranes to extract Li+ from brine and 433

seawater to address the fast-growing lithium demand. Here, we showed that the reassembling 434

of exfoliated two-dimensional VCT sheets could lead to flexible free-standing membranes 435

which can selectively conduct Li+ while excluding other anions or divalent cations. Using 436

MD simulation, we discovered that when the nanochannel dimension is below ~1 nm, Li+ 437

ions jump from one of the nanochannel walls to the opposite wall while they move in the 438

electric field direction. However, when the channel dimension was increased to 1.2 nm, the 439

Li+ ions move by hopping only along one surface of the nanochannel. The inexpensive VCT 440

membrane introduced in this work exhibited either a significantly higher ion selectivity or 441

more cost-effective fabrication process when compared to similar works reported in the 442

literature. This provides an opportunity for many and varied commercial applications of this 443

advanced high-tech membrane. Although in this work the performance of the membrane was 444

evaluated for lithium recovery, it might also be considered for other applications, such as 445

lithium-based batteries and supercapacitors, sensors, solvent dehydration, gas purification, 446

and molecular sieving. 447

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

19

Acknowledgment 448

Authors acknowledge Munirah Mohammad for her contribution to this work. 449

References 450

(1924) On the determination of molecular fields. —II. From the equation of state of a gas. 451 Proceedings of the Royal Society of London. Series A 106(738), 463-477. 452 Abraham, J., Vasu, K.S., Williams, C.D., Gopinadhan, K., Su, Y., Cherian, C.T., Dix, J., Prestat, E., Haigh, 453 S.J., Grigorieva, I.V., Carbone, P., Geim, A.K. and Nair, R.R. (2017) Tunable sieving of ions using 454 graphene oxide membranes. Nature Nanotechnology 12, 546. 455 Ballard, D. and Rideal, G. (1983) Flexible inorganic films and coatings. Journal of Materials Science 456 18(2), 545-561. 457 Choubey, P.K., Chung, K.-S., Kim, M.-s., Lee, J.-c. and Srivastava, R.R. (2017) Advance review on the 458 exploitation of the prominent energy-storage element Lithium. Part II: From sea water and spent 459 lithium ion batteries (LIBs). Minerals Engineering 110, 104-121. 460 Choubey, P.K., Kim, M.-s., Srivastava, R.R., Lee, J.-c. and Lee, J.-Y. (2016) Advance review on the 461 exploitation of the prominent energy-storage element: Lithium. Part I: From mineral and brine 462 resources. Minerals Engineering 89, 119-137. 463 Cseri, L., Baugh, J., Alabi, A., AlHajaj, A., Zou, L., Dryfe, R.A.W., Budd, P.M. and Szekely, G. (2018) 464 Graphene oxide–polybenzimidazolium nanocomposite anion exchange membranes for 465 electrodialysis. Journal of Materials Chemistry A 6(48), 24728-24739. 466 Daiguji, H. (2010) Ion transport in nanofluidic channels. Chemical Society Reviews 39(3), 901-911. 467 Daub, C.D., Åstrand, P.-O. and Bresme, F. (2015) Lithium Ion–Water Clusters in Strong Electric Fields: 468 A Quantum Chemical Study. The Journal of Physical Chemistry A 119(20), 4983-4992. 469 Fumagalli, L., Esfandiar, A., Fabregas, R., Hu, S., Ares, P., Janardanan, A., Yang, Q., Radha, B., 470 Taniguchi, T., Watanabe, K., Gomila, G., Novoselov, K.S. and Geim, A.K. (2018) Anomalously low 471 dielectric constant of confined water. Science 360(6395), 1339-1342. 472 Goh, K., Jiang, W., Karahan, H.E., Zhai, S., Wei, L., Yu, D., Fane, A.G., Wang, R. and Chen, Y. (2015) All-473 Carbon Nanoarchitectures as High-Performance Separation Membranes with Superior Stability. 474 Advanced Functional Materials 25(47), 7348-7359. 475 Guo, Y., Ying, Y., Mao, Y., Peng, X. and Chen, B. (2016) Polystyrene Sulfonate Threaded through a 476 Metal–Organic Framework Membrane for Fast and Selective Lithium-Ion Separation. Angewandte 477 Chemie International Edition 55(48), 15120-15124. 478 Han, Y., Xu, Z. and Gao, C. (2013) Ultrathin Graphene Nanofiltration Membrane for Water 479 Purification. Advanced Functional Materials 23(29), 3693-3700. 480 Horike, S., Umeyama, D. and Kitagawa, S. (2013) Ion Conductivity and Transport by Porous 481 Coordination Polymers and Metal–Organic Frameworks. Accounts of Chemical Research 46(11), 482 2376-2384. 483 Huang, H., Song, Z., Wei, N., Shi, L., Mao, Y., Ying, Y., Sun, L., Xu, Z. and Peng, X. (2013) Ultrafast 484 viscous water flow through nanostrand-channelled graphene oxide membranes. 4, 2979. 485 Hung, W.-S., Tsou, C.-H., De Guzman, M., An, Q.-F., Liu, Y.-L., Zhang, Y.-M., Hu, C.-C., Lee, K.-R. and 486 Lai, J.-Y. (2014) Cross-Linking with Diamine Monomers To Prepare Composite Graphene Oxide-487 Framework Membranes with Varying d-Spacing. Chemistry of Materials 26(9), 2983-2990. 488 Joshi, R.K., Carbone, P., Wang, F.C., Kravets, V.G., Su, Y., Grigorieva, I.V., Wu, H.A., Geim, A.K. and 489 Nair, R.R. (2014) Precise and Ultrafast Molecular Sieving Through Graphene Oxide Membranes. 490 Science 343(6172), 752-754. 491 Jung, M., Kim, H., Baek, K. and Kim, K. (2008) Synthetic Ion Channel Based on Metal–Organic 492 Polyhedra. Angewandte Chemie International Edition 47(31), 5755-5757. 493

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

20

Karna, N.K., Rojano Crisson, A., Wagemann, E., Walther, J.H. and Zambrano, H.A. (2018) Effect of an 494 external electric field on capillary filling of water in hydrophilic silica nanochannels. Physical 495 Chemistry Chemical Physics 20(27), 18262-18270. 496 Liu, G., Jin, W. and Xu, N. (2015) Graphene-based membranes. Chemical Society Reviews 44(15), 497 5016-5030. 498 Majee, A., Bier, M. and Podgornik, R. (2018) Spontaneous symmetry breaking of charge-regulated 499 surfaces. Soft Matter 14(6), 985-991. 500 Marcos, C., Arango, Y.C. and Rodriguez, I. (2009) X-ray diffraction studies of the thermal behaviour of 501 commercial vermiculites. Applied Clay Science 42(3-4), 368-378. 502 Nebbia, G. and Menozzi, G.N. (1968) Early experiments on water desalination by freezing. 503 Desalination 5(1), 49-54. 504 Obut, A. and Girgin, İ. (2002) Hydrogen peroxide exfoliation of vermiculite and phlogopite. Minerals 505 Engineering 15(9), 683-687. 506 Oener, S.Z., Ardo, S. and Boettcher, S.W. (2017) Ionic Processes in Water Electrolysis: The Role of 507 Ion-Selective Membranes. ACS Energy Letters 2(11), 2625-2634. 508 Okada, T., Xie, G., Gorseth, O., Kjelstrup, S., Nakamura, N. and Arimura, T. (1998) Ion and water 509 transport characteristics of Nafion membranes as electrolytes. Electrochimica Acta 43(24), 3741-510 3747. 511 Orooji, Y., Liang, F., Razmjou, A., Li, S., Mofid, M.R., Liu, Q., Guan, K., Liu, Z. and Jin, W. (2017) 512 Excellent Biofouling Alleviation of Thermoexfoliated Vermiculite Blended Poly(ether sulfone) 513 Ultrafiltration Membrane. ACS Applied Materials & Interfaces 9(35), 30024-30034. 514 Orooji, Y., Liang, F., Razmjou, A., Liu, G.P. and Jin, W.Q. (2018) Preparation of anti-adhesion and 515 bacterial destructive polymeric ultrafiltration membranes using modified mesoporous carbon. 516 Separation and Purification Technology 205, 273-283. 517 Pu, Q., Yun, J., Temkin, H. and Liu, S. (2004) Ion-Enrichment and Ion-Depletion Effect of Nanochannel 518 Structures. Nano Letters 4(6), 1099-1103. 519 Razmjou, A. (2019) The Role of Defects in Li+ Selective Nanostructured Membranes: Comment on 520 “Tunable Nanoscale Interlayer of Graphene with Symmetrical Polyelectrolyte Multilayer Architecture 521 for Lithium Extraction”. Advanced Materials Interfaces 6(2), 1801427. 522 Razmjou, A., Barati, M.R., Simon, G.P., Suzuki, K. and Wang, H.T. (2013a) Fast Deswelling of 523 Nanocomposite Polymer Hydrogels via Magnetic Field-Induced Heating for Emerging FO 524 Desalination. Environmental Science & Technology 47(12), 6297-6305. 525 Razmjou, A., Liu, Q., Simon, G.P. and Wang, H.T. (2013b) Bifunctional Polymer Hydrogel Layers As 526 Forward Osmosis Draw Agents for Continuous Production of Fresh Water Using Solar Energy. 527 Environmental Science & Technology 47(22), 13160-13166. 528 Razmjou, A., Simon, G.P. and Wang, H.T. (2013c) Effect of particle size on the performance of 529 forward osmosis desalination by stimuli-responsive polymer hydrogels as a draw agent. Chemical 530 Engineering Journal 215, 913-920. 531 Salieb-Beugelaar, G.B., Dorfman, K.D., van den Berg, A. and Eijkel, J.C.T. (2009) Electrophoretic 532 separation of DNA in gels and nanostructures. Lab on a Chip 9(17), 2508-2523. 533 Shao, J.-J., Raidongia, K., Koltonow, A.R. and Huang, J. (2015) Self-assembled two-dimensional 534 nanofluidic proton channels with high thermal stability. Nat Commun 6. 535 Shirani, E., Razmjou, A., Tavassoli, H., Landarani-Isfahan, A., Rezaei, S., Kajani, A.A., Asadnia, M., Hou, 536 J.W. and Warkiani, M.E. (2017) Strategically Designing a Pumpless Microfluidic Device on an "Inert" 537 Polypropylene Substrate with Potential Application in Biosensing and Diagnostics. Langmuir 33(22), 538 5565-5576. 539 Su, Y., Kravets, V.G., Wong, S.L., Waters, J., Geim, A.K. and Nair, R.R. (2014) Impermeable barrier 540 films and protective coatings based on reduced graphene oxide. 5, 4843. 541 Sun, P., Wang, K. and Zhu, H. (2016) Recent Developments in Graphene-Based Membranes: 542 Structure, Mass-Transport Mechanism and Potential Applications. Advanced Materials 28(12), 2287-543 2310. 544

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

21

Tagliazucchi, M. and Szleifer, I. (2015) Transport mechanisms in nanopores and nanochannels: can 545 we mimic nature? Materials Today 18(3), 131-142. 546 Valaskova, M. and Martynkova, G.S. (2013) Vermiculite: Structural Properties and Examples of the 547 Use. Clay Minerals in Nature - Their Characterization, Modification and Application, 209-238. 548 Walker, G. and Garrett, W. (1967) Chemical exfoliation of vermiculite and the production of colloidal 549 dispersions. Science 156(3773), 385-387. 550 Wang, P., Wang, M., Liu, F., Ding, S., Wang, X., Du, G., Liu, J., Apel, P., Kluth, P., Trautmann, C. and 551 Wang, Y. (2018) Ultrafast ion sieving using nanoporous polymeric membranes. Nature 552 Communications 9(1), 569. 553 Wen, Q., Yan, D., Liu, F., Wang, M., Ling, Y., Wang, P., Kluth, P., Schauries, D., Trautmann, C., Apel, P., 554 Guo, W., Xiao, G., Liu, J., Xue, J. and Wang, Y. (2016) Highly Selective Ionic Transport through 555 Subnanometer Pores in Polymer Films. Advanced Functional Materials 26(32), 5796-5803. 556 Winarto, Yamamoto, E. and Yasuoka, K. (2017) Water Molecules in a Carbon Nanotube under an 557 Applied Electric Field at Various Temperatures and Pressures. Water 9(7), 473. 558 Yoshizuka, K., Kitajou, A. and Holba, M. (2006) Selective recovery of lithium from seawater using a 559 novel MnO2 type adsorbent III - Benchmark evaluation. Ars Separatoria Acta 4, 78-85. 560 Zhang, H., Hou, J., Hu, Y., Wang, P., Ou, R., Jiang, L., Liu, J.Z., Freeman, B.D., Hill, A.J. and Wang, H. 561 (2018a) Ultrafast selective transport of alkali metal ions in metal organic frameworks with 562 subnanometer pores. Science Advances 4(2). 563 Zhang, H., Hou, J., Hu, Y., Wang, P., Ou, R., Jiang, L., Liu, J.Z., Freeman, B.D., Hill, A.J. and Wang, H. 564 (2018b) Ultrafast selective transport of alkali metal ions in metal organic frameworks with 565 subnanometer pores. Science Advances 4(2), eaaq0066. 566 Zhao, Y., Shi, W., Van der Bruggen, B., Gao, C. and Shen, J. (2018) Tunable Nanoscale Interlayer of 567 Graphene with Symmetrical Polyelectrolyte Multilayer Architecture for Lithium Extraction. Advanced 568 Materials Interfaces 5(6), 1701449. 569

570

571

572

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

22

573

Figure 1 typical VCT structural units (two magnesium aluminosilicate layers plus an interlayer space) (a), Raw VCT 574 crystals (b), VCT after thermal shock at 900

oC (c), Wide-angle XRD patterns of VCT crystals at different thermal shock 575

treatments (d), Adsorption-desorption isotherms of VCT crystals at a different temperature (e), and HRTEM of VCT flakes 576 after exfoliation (inset: a dispersion of VCT crystals in water) (f). 577

578

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

23

579

Figure 2 A flexible free-standing VCT membrane obtained by vacuum filtration (a), Cross-section AFM and FESEM images 580 of VCT membrane (b), 3D model and TEM images of nano-fluidic VCT membrane (c), and XRD patterns of raw and 581 exfoliated VCT flake (d). 582

583

584

585

586

587

588

589

590

591

592

593

594

595

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

24

596

597

Figure 3 I-V curves of different electrolytic solutions indicating selective transportation of Li+ ions (a), Nyquist plots of 598

VCT membrane for different concentrations of LiCl electrolyte solutions E.I.S (b), the effect of membrane orientation on 599 the I-V performance of the membranes (c), a comparison between the I-V curves of VCT membrane and its commercial 600 counterparts (Nafion and CMI-7000S) (d), reduction in EC as a function of (e) time at fixed voltage of 12V and (f) applied 601 voltage over 2 hr of experiments for different LiCl solution indicating a great potential for Li

+ extraction. 602

603

604

605

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

25

606

607

Figure 4 Representative I-V curves of VCT membrane recorded at different pH (a), Proton conductivity as a function of 608 pH, which shows the deviation from bulk solution at low proton concentration indicating surface charge-governed-609 transport mechanism (b), and radial distribution function (RDF) and coordination number of Li and oxygen of water 610 inside nanochannel with channel dimension of 0.4 and 1.2nm (c and d), for 0.6 and 08 nm channel dimension please see 611 Figure S8 in supporting information. 612 613

614

615

616

617

618

619

620

621

622

623

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

26

624

625

626

Figure 5 Li+ ions transportation behavior and water molecules orientation inside VCT nanochannels with different inter-627

surface distance (1.2 V/nm). The MD simulation revealed that the Li+ ions jump between the surfaces of the 628

nanochannel when the inter-surface distance is in subnanometer range. 629

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

27

630

631

Figure 6 RDF of Li and Oxygen of VCT nanochannel surface (1.2 V/nm) inside nanochannel with different channel 632 dimension (a), the velocity of Li

+ ions inside 0.4 and 1.2 nm VCT nanochannel with different power density of electrical 633

field (b and c), and a comparison between the velocity of Li+ and that of Na

+, K

+ and Ca

2+ ions in 0.4 nm VCT nanochannel 634

with power density of 2 V/nm (d). 635

636

MAN

USCR

IPT

ACCE

PTED

ACCEPTED MANUSCRIPT

Highlights

Li+ ion-selective membranes for mineral resource recovery are highly demanded

2D subnanofluidic hydrous phyllosilicate channels are used to prepare the membranes

The membrane showed alkali metal ion selectivity in the order of Li+> Na+>K+

For below 1 nm channels, Li+ conducted by jumping between 2D nanochannels walls

The Li+ selective membrane exhibited two-surface-charge-governed transport mechanism