Capturing Lithium from Wastewater Using a Fixed Bed Packed with 3

Oct 24, 2016 - The Dubinin-Ashtakhov (DA) site energy distribution model based on Polanyi theory described the linear increase of Li adsorption capaci...
0 downloads 4 Views 3MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Capturing Lithium from Wastewater Using A Fixed Bed Packed with 3-D MnO2 Ion Cages Xu-Biao Luo, Kai Zhang, Jinming Luo, Shenglian Luo, and John Crittenden Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b02247 • Publication Date (Web): 24 Oct 2016 Downloaded from http://pubs.acs.org on October 24, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 45

Environmental Science & Technology

1

Capturing Lithium from Wastewater Using A Fixed Bed Packed with

2

3-D MnO2 Ion Cages Xubiao Luoa, Kai Zhanga, Jinming Luob,c, Shenglian Luoa*, John Crittendenc*

3

4

a

Key Laboratory of Jiangxi Province for Persistent Pollutants Control and Resources Recycle,

5

6 7

8 9

Nanchang Hangkong University, Nanchang 330063, PR China

b

Key Laboratory of Drinking Water Science and Technology, Research Center for

Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China

c

Brook Byer Institute for Sustainable Systems and School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA

10 11

*Corresponding author

12

Tel: +86 73183953371

13

Fax: +86 73183953373

14

E-mail address: [email protected], [email protected]

15

16 17 18 19 20 21

ACS Paragon Plus Environment

Environmental Science & Technology

22

Visual Abstract /TOC

23 24 25 26 27 28 29 30 31 32 33 34

35

36

37

38

ACS Paragon Plus Environment

Page 2 of 45

Page 3 of 45

Environmental Science & Technology

39

ABSTRACT

40

3-D MnO2 ion cages (CMO) were fabricated and shown to have a high capacity for

41

lithium removal from wastewater. CMO had a maximum Li(I) adsorption capacity of

42

56.87 mg/g, which is 1.38 times greater than the highest reported value (41.36 mg/g).

43

X-ray photoelectron spectroscopy indicated that the stability of the -Mn-O-Mn-O-

44

skeleton played an essential role in Li adsorption. The lattice clearance had a high

45

charge density, forming a strong electrostatic field. The Dubinin-Ashtakhov (DA) site

46

energy distribution model based on Polanyi theory described the linear increase of Li

47

adsorption capacity (Q0) with increasing temperature ( Q0 = k3 × Em + d3 = k3 ×(a ×T ) + d3 ).

48

Furthermore, the pore diffusion model (PDM) accurately predicted the lithium

49

breakthrough (R2 ≈ 0.99). The maximum number of bed volumes (BVs) treated were

50

1,374, 1972 and 2493 for 200 µg/L at 20, 30 and 40 °C, respectively. Higher

51

temperatures increased the number of BVs that may be treated, which implies that

52

CMO will be useful in treating industrial Li(I) wastewater in regions with different

53

climates (e.g., Northern or Southern China).

54

55

56

57

58

ACS Paragon Plus Environment

Environmental Science & Technology

59

INTRODUCTION

60

Lithium (Li) is a critical material for energy-related technologies1. Accordingly,

61

there has been increasing demand for lithium and Li compounds, particularly in the

62

rapid expansion of rechargeable lithium ion batteries (LIBs).2,3 We are now facing a

63

contraction the of the Li supply and increasing demand. Currently, approximately

64

200-500 MT of spent LIBs are produced each year.4 The Li content in spent LIBs is

65

2-7 wt.%, which is an important resource of lithium. Effective and efficient recovery

66

of Li could become a very important because it not only can relieve Li shortages, but

67

also can resolve environmental pollution problem caused by spent LIBs. Additionally,

68

research on lithium recovery from LIBs have attracted considerable interest in recent

69

years.5-7 Conventional technological approaches, including nanofiltration and

70

electrolysis, have been used to remove lithium from wastewater.8 According to

71

Tsuruta, the biological recovery of Li(I) can also be achieved using various

72

microorganisms.9 However, precipitation is the most common technology used for

73

lithium recovery. However, precipitation is difficult to use on lithium wastewater with

74

high Mg/Li contents because Li2CO3 is highly soluble (approximately 1.3 g/100 mL

75

H2O).10 Adsorption is one of the most promising methods for lithium recovery from

76

lithium wastewater because it is one of the most cost-effective and environmentally

77

friendly methods.11

78

The adsorbent plays a vital role in determining the performance of packed bed

79

adsorption systems; however, the majority of adsorbents have a lower adsorption

80

capacity for smaller cations, such as Li.12 Moreover, many adsorbents have been used

ACS Paragon Plus Environment

Page 4 of 45

Page 5 of 45

Environmental Science & Technology

81

to remove Li(I) but with low capacities. Abe et al.13 synthesized an adsorbent, called

82

the LiSbO3 precursor, and found that the adsorption capacity is only 1.0-1.4 mg/g.

83

Baumant et al.14 prepared a LiCl·2Al(OH)3·nH2O-type adsorbent, which exhibited a

84

Li(I) adsorption capacity of 2-3 mg/g. Song et al.15 investigated the Li(I) adsorption

85

capacity of D751 resin and found that the maximum adsorption capacity was also

86

only 20.82 mg/g and that the Li selectivity was poor as compared to other cations.

87

The waste streams for Li recovery contains large amounts of interfering metal ions

88

such as K, Na, Ca, Mg, Zn, Co, Cr, Cu, Al, Ni, Pb, and Cd, and the process stream is

89

usually alkaline.16 Therefore, high-capacity and selective adsorbents must be

90

developed to selectively separate Li(I) from wastewater that contains a variety of

91

cations.

92

Ionic-cage17,18 adsorbents possess a “memory effect” and that makes them ideal

93

for extracting target ions with an extremely high selectivity. The specific adsorption

94

sites for CMO were prepared by removing target ions from a stable inorganic solid

95

that contain Li in the lattice. The ionic-cage adsorbents have the advantages of low

96

toxicity and high chemical stability,19,20 but they have the disadvantage of slow

97

adsorption kinetics. 3-D MnO2 ionic cages (CMO) have: (1) high selectively for Li(I)

98

in the presence of competing ions, (2) high adsorption capacity and (3) a fast

99

adsorption kinetics. These attributes are the result of 8a-16d-8a channels in a

100

three-dimensional interstitial space provided by the -Mn-O-Mn-O- framework in the

101

Li4Mn5O12 spine.

ACS Paragon Plus Environment

Environmental Science & Technology

102

The adsorption maximum capacity (Q0) increases because CMO is endothermic

103

adsorbent. Previous studies have not examined the relationship between Q0 and T. We

104

explored the relationship between Q0 and T using the Dubinin-Ashtakhov (DA) Site

105

Energy Distribution model. The increase of the adsorption capacity with temperature

106

adsorption is important for Li(I) removal, because wastewater from southern China

107

can be considerably warmer than wastewater from northern China.

108

We investigated the performance of a CMO adsorbent for Li(I) removal. To

109

examine the practical applications of the CMO, fixed bed adsorption experiments and

110

mathematical model were developed.21 Some of the classical models include as those

111

developed by Thomas,22 Adams–Bohart,23 Clark,24 and Yoon–Nelson.25 These models

112

typically have simple mathematical solutions. However, they are empirical models,

113

which limits their ability to predict the performance of full-scale adsorption treatment

114

systems. Accordingly, we used the pore diffusion model (PDM)26 to predict and

115

validate small column breakthrough data. Thus, we can use the model to predict

116

lithium breakthrough in full-scale adsorption systems. In brief, our goal is to: (1)

117

explore the specific relationship between the Li adsorption capacity (Q0) and

118

temperature (T), (2) verify a model that can predict the fixed bed performance for

119

removal Li, and (3) provide specific guidance for engineering application.

120

MATERIALS AND METHODS

121

Chemicals and materials

122

The sources and grades of the chemicals used in this study are provided in

ACS Paragon Plus Environment

Page 6 of 45

Page 7 of 45

Environmental Science & Technology

123

Section S1 of the Supporting Information, and the characterization methods are

124

provided in Section S2. Synthetic Li solutions of different concentration were

125

prepared in DI water for adsorption isotherm, kinetic studies, and shorter column

126

fixed-bed experiments. The solutions with the concentration of 200 µg/L of Li, Na, K,

127

Ca, Mg, pH = 10, TOC = 48.6 mg/g were used for competitive adsorption

128

experiments and IX model because Na, K, Ca, and Mg ions have the similar ionic

129

radius. The concentration of all ions was adjusted to 200 µg/L in order to better study

130

selective adsorption experiments.

131

Synthesis of 3-D MnO2 Ion Cages

132

The process for preparing the 3-D MnO2 ion cages is described in detail in Section

133

S3 of the Supporting Information. In brief, the 3-D MnO2 ion cages (CMO) were

134

prepared using a two-step process (Figure 1). First, Li4Mn5O12 (LMO) precursor

135

particles were prepared by hydrothermal synthesis via a low-temperature solid-phase

136

reaction. The LMO particles were then washed with aqueous solutions of HCl to

137

remove the Li(I) ions to prepare a selective absorbent with Li(I) transport channels.

138

These channels consist of interconnected 3-D MnO2 ion cages within the 3-D lattice

139

of the Li4Mn5O12 spinels within each absorbent particle.

140

Adsorption Isotherm Procedure

141

Adsorption isotherms were obtained by shaking a mixture containing 20 mg of

142

CMO and 20 mL of the Li(I) solution at pH 10.1 using an incubator shaker at 160 rpm.

143

The adsorption isotherms of Li(I) were obtained at temperatures of 20, 30 and 40 °C.

ACS Paragon Plus Environment

Environmental Science & Technology

144

The initial concentrations of Li(I) ranged from 20 to 500 mg/L. After being shaken for

145

12 h, the resulting mixture was centrifuged at 12,000 rpm (which resulted in 22,550 G)

146

to remove the adsorbent. The concentrations of Li(I) in the supernatant before and

147

after adsorption were measured using atomic absorption spectrometry (AAS). The

148

adsorption capacity of the Li(I) ion was calculated using Eq. 1:

149

qe =

150

where qe (mg/g) denotes the equilibrium adsorption capacity, C0 and Ce are the initial

151

and equilibrium concentrations (mg/L) of Li(I), respectively, V is the volume (mL) of

152

adsorption solution, and m is the mass (mg) of the adsorbent.

153

Adsorption Selectivity

C0 − C e ⋅V m

(1)

154

The selectivity of Li+, Na+, K+, Mg2+, and Ca2+ ions with respect to H+ ions was

155

determined. Fifty milligrams of CMO was mixed with a 50-mL water solution

156

containing competing ions with a concentration of approximately 200 µg/L and pH 10

157

at 20 °C. The adsorption equilibrium was reached after 12 h of shaking, and the ion

158

concentrations were determined using AAS. In a multicomponent system, separation

159

factors can be converted to different reference ions by using Eq. 2-3:

160

α ki = α ij ⋅ α kj

161

xi =

Ci

,

m

∑C k =1

( α ij =

k

xj =

qi C j

=

Ci q j

yi x j xi y j

)

Cj m

∑C

k

k =1

ACS Paragon Plus Environment

(2)

Page 8 of 45

Page 9 of 45

Environmental Science & Technology

162

yi =

xi n

∑ (x α ) i

k =i

k i

,

yj =

xj n

∑ (x α ) j

(3)

k j

k= j

163

where α ki is separation factor of counter ion i with respect to ion k, unitless; q and C

164

represent the equivalent solid phase and liquid phase concentration (mg/L) for ions i

165

and j, respectively; x and y represent the liquid phase and the solid phase equivalent

166

fraction or mole fraction of ions i and j, unitless; i denotes the counter ion and j

167

represents the presaturant ion.

168

Continuous Flow Fixed-Bed Column Adsorption Experiments

169

A short bed adsorber (SBA) test was conducted to validate the calculated kf and

170

Dp values.27,28 In the SBA test, a column was packed with a CMO. It has a bed depth

171

of 0.9 cm (a mass of 0.65 g) and a diameter of 1.3 cm. The bed had an empty bed

172

approach velocity of 0.09 m/hr and had an influent lithium concentration, C0, of 200

173

µg/L. The exhausted Li+ on the CMO column was desorbed using 0.1 mol/L HCl

174

solution and a flow rate of 1.0 mL/min.

175

Determining the Parameters for Pore Surface Diffusion Modeling

176

The PSDM assumes intraparticle mass flux is described by surface and pore

177

diffusion and adsorption equilibrium of individual compounds can be represented by

178

the Freundlich isotherm equation.

179

Based on the Gnielinski correlation,29 the external mass transport coefficient (kf )

180

was estimated using Eqs. S1-3 in Section S4 of the Supporting Information. The pore

181

diffusion coefficient (Dp) was estimated using Eq. 4.29

ACS Paragon Plus Environment

Environmental Science & Technology

182

Dp =

ε p × D1 τ

(4)

The variable τ was estimated by Mackie and Meares (Eq. 5) for electrolyte

183 184

solutions using the following expression.29

185

τ=

186

where τ is the toruosity factor and εp is the particle porosity (εp ≈ 0.741). This equation

187

yielded a τ of 2.14 with the estimated Dp ≈ 1.603 × 10-10 m2/s. We assumed that the

188

surface diffusion coefficient was equal to zero because the Li ions could not be

189

transported down the surface of the CMO and were transported into the ion channels

190

for removal. So in actuality we used the pore diffusion model (PDM) to predict the

191

lithium breakthrough curve.26,30,31 PDM simulations were conducted using the

192

AdDesignS software (Michigan Technological University).32

(2 − ε )

2

p

εp

(5)

193

The wall effect on mass transfer can be neglected for dcolumn/dp ratios > 20.33,34

194

The relative importance of the internal (pore) and external (film) mass transport

195

resistances was evaluated using the pore Biot number (Bip) (Eq. 6) according to

196

Sontheimer et al. as follows:29

197

Bip =

198 199

200

kf ×dp 2 × Dp

(6)

The calculated value of Bip was 29.8. Accordingly, intraparticle diffusion controls the overall mass transport of the system because Bip ≥ 20.29 The maximum number of bed volumes (BVMAX) (Eq. 7) could be estimated

ACS Paragon Plus Environment

Page 10 of 45

Page 11 of 45

Environmental Science & Technology

201

according to Sontheimer et al. as follows:26

202

BVMAX =

203

where ρBED is the bed density of the media in the packed bed (g/cm3), and q0 is the

204

adsorption capacity calculated at C0.

205

RESULTS AND DISCUSSION

q0 × ρBED C0

(7)

206

As discussed below, we evaluated CMO adsorbent performance for different

207

concentrations, different temperatures, and pH values using isotherm tests, kinetic

208

studies, and fixed bed experiments. With respect to adsorption thermodynamics, we

209

determined the Li site energy distribution using various adsorption isotherm models

210

and data. We verified the PDM by comparing it to batch kinetic and fixed bed

211

experiments. We used the verified model to predict full scale performance and the

212

impact of competing ions on the recovery of Li ion. In addition, we determined the

213

recovery of Li using fixed bed adsorption by conducting adsorption and regeneration

214

experiments.

215

Adsorbent Characterization

216

The pore size distribution is shown in Figure S1. The specific surface area (BET)

217

of CMO was approximately 54.1 m2/g (see the figure inset). The XRD patterns of the

218

MO, LMO and CMO particles are shown in Figure 2. The crystal phases of MO,

219

LMO and CMO are pure tetragonal phase (T) β-MnO2 (according to JCPDS 24-0735),

220

pure cubic phase (C) c-Li4Mn5O12 (according to JCPDS 46-0810) and λ-MnO2

ACS Paragon Plus Environment

Environmental Science & Technology

221

(according to JCPDS 44-0992), respectively. The substitution of H for Li in the lattice

222

causes the crystal cell to shrink, whereas the crystal skeletons of LMO and CMO

223

remain unchanged. As a result, the -Mn-O-Mn-O- skeleton will be extremely stable

224

during adsorption because the lattice clearance contains vacancies with a high charge

225

density, forming a strong electrostatic field and are the perfect size for Li ions. Thus,

226

the MnO2 ion cages can adsorb the lithium ions. Figure 3 shows the SEM

227

micrographs of the obtained samples. Figures 3a and 3b illustrate that the MO

228

nanorods are delicate, smooth and uniform before calcination. In Figure 3c, the

229

resulting LMO from the calcined MO are still nanorods. Figure 3d clearly illustrates

230

that CMO maintains a similar morphology as LMO. However, the magnified view

231

shows that the CMO nanorods are partially cracked and the surface is coarse. The

232

particle surface appears to be partially dissolved during the acid wash that removes Li

233

ions (pickling process). This phenomenon was confirmed by the results of the XRD

234

analysis, shown in Figure 2.

235

As shown in Figures 3 and S1, the LMO and CMO have a high degree of porosity

236

and large pores between the nanorods, which increases the intraparticle pore diffusion

237

and reduces the impact of counter diffusion. The lack of pore connectivity (nanorods

238

contacting one another as opposed to a solid support that contains pores) and strong

239

electrostatic interaction between ions and the surface eliminates surface diffusion to

240

the intraparticle transport flux. As shown below, liquid-phase pore diffusion is rapid

241

and produces a steep breakthrough curve.

242

XPS Analyses. The chemical composition of the CMO before and after Li adsorption

ACS Paragon Plus Environment

Page 12 of 45

Page 13 of 45

Environmental Science & Technology

243

was measured using X-ray photoelectron spectroscopy (XPS). As shown in Figure 4a

244

and 4b, the peaks corresponding to Mn 3p, Li 1s, O 1s and Mn 2p are distinct. The

245

XPS spectrum of Mn 2p is plotted in Figure 4c and 4d. Two peaks located at 642.1

246

and 653.8 eV correspond to Mn 2p3/2 and Mn 2p1/2, respectively, with a spin energy

247

separation of 11.7 eV, which is consistent with the well-characterized MnO2 XPS

248

spectra in the literature.35 The ratio of the Mn/O atomic concentration (Mn: 25.89%, O:

249

51.72%) is 1 : 2, indicating the dominance of Mn4+ in the CMO. There was no change

250

in the proportion of Oad/Olatt before and after Li adsorption (Figure 4e and 4f), which

251

was approximately 48.2% on the CMO, after Li adsorption, indicating that the

252

chemical environment is stable and unaffected by the adsorbed Li(I). Figure 4g and 4h

253

were XPS spectrum of Li 1s. In Figure 4h, the appearance of the Li 1s peak at the

254

binding energy of 54.9 eV indicates that Li(I) is adsorbed on the CMO. This finding

255

indicates that stability of the Mn-O bond, which plays an indispensable role in

256

forming Li cages and it has a high charge density. The high charge density forms a

257

strong electrostatic field, which aids in the adsorption of Li(I). Researchers believe

258

that there are three mechanisms that explain the adsorption behavior of Li in ion cages:

259

(i) redox reactions36; (ii) ion exchange37; and (iii) a complexation mechanism from a

260

redox reaction during ion exchange38. Our results demonstrate that the Mn valence

261

cannot be reduced when CMO adsorbs Li(I), which was confirmed by XPS (Figures

262

4c and 4d). Consequently, the adsorption of Li on the CMO is an ion exchange

263

reaction, which is in good agreement with the results of the analysis using the D-R

264

isotherm model.

ACS Paragon Plus Environment

Environmental Science & Technology

Page 14 of 45

265

Effect of Initial Concentration and Temperature. Figure 5 shows the adsorption

266

isotherms of Li(I) on CMO for three temperatures, i.e., 20, 30 and 40 °C. The

267

adsorption capacity of Li(I) on the CMO increases with increasing concentration and

268

temperature.

269

approximately 61.32 mg/g at equilibrium concentration about 300 mg/L at 40 °C,

270

which is higher than those reported for other lithium ion-cage adsorbents. For

271

example, the adsorption capacity is 20.82 mg/g for D751 resins,15 17.0 mg/g for

272

H2.0Li0.1Mn4.0O7.8·0.22H2O,39 31.23 mg/g for LiAlMnO440 and 41.36 mg/g on

273

Li4Mn0.5Ti0.5O4.41 Hence, the CMO exhibits a significantly higher adsorption capacity

274

than these other adsorbents. Additionally, the maximum Li(I) adsorption capacity on

275

CMO is 15 times greater than on Fe3O4@SiO2-IIP (4.1 mg/g).16 In addition, the

276

synthesis cost of the CMO adsorbent is much less than Fe3O4@SiO2-IIP, which makes

277

industrial lithium recycling more feasible. The adsorption capacity also increased with

278

the increasing temperature, suggesting that the adsorption of lithium is endothermic.

279

Effect of pH. pH plays an important role in determining the performance of an

280

adsorbent. Figure S2 shows that the adsorption capacity of Li(I) increases with

281

increasing pH. When the pH is below pH 4.0, the adsorption capacity is low due to

282

the protonation of the -OH group, when the pH is higher than pH 4.0, high adsorption

283

capacity of Li+ is observed. The maximum adsorption capacity for Li(I) was obtained

284

at a pH of 12.95. The CMO ion exchange/adsorption site is an -OH group and can

285

exchange the H+ for Li(I). At high pH, the -OH group ionizes and adsorbs Li(I).42

286

Adsorption Selectivity. Competitive adsorption experiments of Li+, Na+, K+, Mg2+

Moreover,

the

maximum

equilibrium

ACS Paragon Plus Environment

adsorption

capacity

is

Page 15 of 45

Environmental Science & Technology

287

and Ca2+ were conducted to determine the selectivity of these ions on CMO. The

288

selectivities are summarized in Table 1. As can be seen in Table 1, the CMO exhibit a

289

much higher selectivity separation factor ( α ki ) and affinity of Li(I) as compared with

290

other competitive ions. Hence, the CMO exhibits a higher selectivity toward Li(I)

291

than other ions that may be found in wastewater. At approximately 0.7 Å, the

292

three-dimensional (1 × 3) tunnels in the spinel lattice are a suitable size for adsorbing

293

lithium ions in the CMO nanocrystal cubic phase.43 In other words, lithium ions can

294

be adsorbed in three-dimensional (1 × 3) tunnels, which have a strong electrostatic

295

field, whereas other larger metallic ions can only be adsorbed to a lesser extent on the

296

surface sites. The ion cages had a higher selectivity for Li+ than for Mg2+, despite their

297

similar ionic sizes and different valence.

298

In generally, multivalent ions tend to have higher selectivity than monovalent

299

ions. Because the hydration energy of Mg2+ is approximately 4 times larger than that

300

of Li+ (Mg(II)hydration energy = - 455 kcal/mol vs. Li(I)hydration energy = - 122 kcal/mol),44 a

301

higher energy would be required for dehydration to enter the cavity of a lithium ion

302

cages. Accordingly, it is easier to remove the water of hydration from Li(I), which

303

results in a higher selectivity for Li(I) ions. Therefore, there are 2 criteria that must be

304

met for adsorption to occur in the three-dimensional tunnels: (1) the ion must fit, and

305

(2) the adsorption free energy must be greater than dehydration energy which is

306

required to removal the water of hydration. The dehydration energy causes Li

307

adsorption to be endothermic.

308

Adsorbent Regeneration in a Batch Reactor. A good adsorbent should have

ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 45

309

superior desorption and regeneration performance. The regeneration of Li(I) was

310

desorbed using 0.1 mol/L HCl solution, the adsorbent was separated and washed with

311

deionized water, then dried. After that, regenerated CMO was placed into a lithium

312

adsorption solution (the experiment is similar to isotherm procedure). To assess the

313

CMO’s performance, the CMO adsorbent was subjected to six repetitive

314

adsorption-desorption cycles for the removal of Li(I) ions from aqueous solutions. As

315

shown in Figure 6, the adsorption capacity for Li(I) decreased slightly after each

316

regeneration,

317

adsorption-desorption cycle for Li(I) was only 7.6% less than that for the first cycle.

318

During the adsorbent regeneration, the lost Mn can destroy the 3-D tunnel in the

319

spinel structure, which can cause a decrease in the adsorption capacity. These results

320

demonstrate that the CMO can be reused for a large number of adsorption-desorption

321

cycles.

322

Adsorption Isotherm Models. The Langmuir and Freundlich isotherm models,

323

shown here, respectively, were used to fit the CMO isotherm data.

324

qe =

325

qe = k F Cen

326

where qe (mg/g) and Ce (mg/L) are the amount of lithium adsorbed at equilibrium and

327

the equilibrium concentration, respectively. kL (L/mg) is the equilibrium constant and

328

and Q0 the maximum adsorption capacity. kF and 1/n are the capacity factor and

329

intensity for the Freundlich isotherm equation.

Q0 k L C e 1 + k LCe

and

the

adsorption

capacity

of

the

CMO

for

the

sixth

(8)

1

(9)

ACS Paragon Plus Environment

Page 17 of 45

Environmental Science & Technology

330

The isotherm parameters that were determined by fitting the data in Figure 5 are

331

summarized in Table 2. A comparison of the correlation coefficients (R2) illustrates

332

that the Freundlich equation described the data slightly better than the Langmuir

333

equation, suggesting that the adsorption of Li(I) on CMO occurs on sites that have a

334

distribution of site energies (i.e., heterogeneous site energies). In contrast, the

335

Langmuir equation assumes that all of the site energies are identical.

336

Adsorption Site Energy Thermodynamics. We used the Dubinin-Radushkevich

337

(D-R) isotherm45 to determine whether the adsorption process originated from

338

physical (Van der Walls) or chemical (surface reaction) forces. The D-R isotherm is

339

given by Eq. 10:

340

ln qe = ln Q0 − βε 2

341

where qe is the amount of metal ions adsorbed on the adsorbent per unit weight

342

(mol/g), Q0 is the maximum adsorption capacity (mol/g), βDR is the activity coefficient

343

related to the mean free energy of adsorption (mol2/J2) and ε (J/mol) is the Polanyi

344

potential ( ε = RT ln (1 + 1/Ce ) ). The D-R isotherm model fit the equilibrium data well

345

(R2 > 0.99, Figure S3 and Table S1). The intercept of the plots for 20, 30 and 40 °C

346

yield qm values of 5.81, 6.21 and 6.91 mol/g, respectively. The mean free energy of

347

adsorption, E (kJ/mol), is determined using Eq. 11:

348

E = (2 β DR )

349

The mean free adsorption energy was approximately 8.6 kJ/mol, which is in the

350

energy range of an ion-exchange reaction (i.e., 8–16 kJ/mol).46

-1 / 2

(10)

(11)

ACS Paragon Plus Environment

Environmental Science & Technology

351

Thermodynamic calculations were performed to further explain the endothermic

352

nature of the process. The thermodynamic parameters (namely, the Gibbs free energy

353

∆G0 (kJ/mol), enthalpy ∆H0 (kJ/mol) and entropy ∆S0 (kJ/mol K) were estimated by

354

Eqs. 12 and 13, respectively:47

355

∆G 0 = − RT ln K

356

ln K =

357

where R is the universal gas constant (8.314 J/mol K) and T is the temperature (K). K

358

is the thermodynamic equilibrium constant in the adsorption process, which was

359

determined using the method of Khan and Singh48 by plotting ln(qe/Ce) versus qe and

360

extrapolating to zero qe.

∆S 0 ∆H 0 1 − × R R T

(12)

(13)

361

Based on Eq. 13, The thermodynamic parameters of ∆H0 and ∆S0 can be

362

determined from the slope and the intercept, respectively, as shown in Figure S4 and

363

Table S2. The positive value of ∆H0 (11.34 kJ/mol) proves that the adsorption process

364

is endothermic between 20 and 40 °C. The positive value of ∆S0 (0.08 kJ/mol K)

365

suggests that the degree of randomness increases during Li(I) adsorption. This may be

366

due to the loss of the water of hydration from the Li. The ∆G0 was calculated to be -

367

12.20, - 13.05 and - 13.77 kJ/mol for 20, 30 and 40 °C, respectively, indicating that

368

Li(I) adsorption is thermodynamically favored. A smaller negative ∆G0 indicates that

369

adsorption is less favored thermodynamically. Moreover, ∆G0 became more negative

370

(the absolute value of ∆G0 increased) as temperature increased at the same adsorbate

371

loading, suggesting that the driving force of the adsorption increased with increasing

ACS Paragon Plus Environment

Page 18 of 45

Page 19 of 45

Environmental Science & Technology

372

temperature, which consistent the adsorption capacity increasing with increasing

373

temperature. This conclusion is in good agreement with the results obtained from the

374

D-R isotherm model.

375

Dubinin-Ashtakhov (DA) Model-Based Site Energy Distribution. The Polanyi

376

theory49 assumes that the adsorption potential is independent of temperature and

377

application of the theory describes the adsorption process. The nonlinear DA model

378

based on Polanyi theory was used to fit the experimental isotherm data according to

379

Eq. 14:

380

log Qe = log Q0 − (ε sw /E d )

381

where Qe (mg/kg) is the amount of adsorbate compared to adsorbent, Q0 (mg/kg) is the

382

maximum adsorption capacity, εsw = RTln(Cs/Ce) (J/mol) is the effective adsorption

383

potential, Cs (mg/L) is the water solubility of the adsorbate, Ce (mg/L) is the

384

equilibrium concentration of the adsorbate in the liquid phase, Ed (J/mol) is the

385

“normalized constant”, and b is the empirical isotherm fitting parameter.

b

(14)

386

The adsorption capacity results from the adsorption onto adsorption sites that

387

have a distribution of energies F(E*) (see section S5 in the Supporting Information)

388

and is given by Eqs. S5-8. Because the resulting site energy distributions are not

389

normalized, the area under the distribution is equal to the maximum adsorption

390

capacity Q0.

391

Figure S5a and Table S3 illustrate that the experimental data fit the DA model

392

well (R2 > 0. 99). The site energy distributions were all unimodal distributions at

ACS Paragon Plus Environment

Environmental Science & Technology

393

different temperatures, as shown in Figure S5b. With increasing site energy (E*), the

394

frequency function F(E*) increased until it reached the F(E*) peak apex, after which it

395

decreased to approach zero. Theoretically, the area under the peak represents the

396

number of available adsorption sites in a specific energy range. Increasing the

397

temperature from 20 °C to 40 °C increased the area under the peak. The affinity of Li

398

for the CMO surface was related to the position of the energy distribution mean on the

399

energy axis.50 A larger value of the mean energy resulted in a higher adsorption

400

affinity and thus a higher adsorption capacity. The ∆G0 analysis was consistent with

401

the energy analysis, confirming that the affinity of CMO increased with increasing

402

temperature.

403

DA Model-Based Average Site Energy and Adsorption Site Heterogeneity.

404

The DA Model-based average site energy distribution was employed to determine

405

the interaction forces between adsorbent and adsorbate, and its width describes the

406

surface energy heterogeneity of the adsorbents. The mean site energy (Em) and

407

adsorbent site energy heterogeneity index (σe*) were calculated (see section S6 in the

408

Supporting Information) from Eq. S10. The adsorbent site energy heterogeneity can

409

be described by determining the standard deviation (σe*) using Eqs. S11 and S12.

410

Figure S6 describes the site energy distributions with variations in Ed and b,

411

demonstrating that both the mean and shape of the distribution changed as either Ed or

412

b were varied. This result implies that b could not represent the site energy

413

heterogeneity. Em increased significantly with increasing temperature (Figure S7 and

ACS Paragon Plus Environment

Page 20 of 45

Page 21 of 45

Environmental Science & Technology

414

Table S4), which can be ascribed to the three-dimensional tunnels structure in the

415

CMO, forming additional adsorption sites. The term σe* decreased with increasing

416

temperature, indicating that the stronger adsorption affinity for Li on the CMO.

417

The mean site energy (Em) was examined to further understand the relationship

418

between the maximum adsorption capacity (Q0) and adsorption temperature (T). In Eq.

419

15, Em is considered to be a function of T only. In Eq. 16, Q0 is determined by Em by

420

plugging Eq. 15 into Eq. 16 and then incorporating the result into in Eq. 17. The

421

adsorption capacity (Q0) was found to be linearly related to the adsorption

422

temperature (T) based on Eqs. 15-17:

423

Em = k1 × T + d1

(15)

424

Q0 = k2 × Em + d 2

(16)

425

Q0 = k3 × Em + d 3 = k3 × (a × T ) + d 3

(17)

426

where k1 (1/K), k2 and k3 (µmol/J) are regression coefficients and d1, d2 and d3 (mg/kg)

427

are fitting parameters. The terms k1, k2 and k3 are dimensionless and represent the

428

contribution fractions of T to Em, Em to Q0 and T to Q0, respectively.

429

The regression results revealed the contribution of the average site energy (Table

430

S5). The maximum adsorption capacity (Q0) was found to be linearly related to the

431

adsorption temperature. Furthermore, Em increased and Q0 significantly increased

432

with increasing temperature, indicating that the number of adsorption sites also

433

increases with increasing temperature. For endothermic reactions with increasing

434

adsorption temperature, the positive value of (k1 × T) increased. In contrast, an

ACS Paragon Plus Environment

Environmental Science & Technology

435

increasing negative value of (k1 × T) was observed with increasing temperature for

436

exothermic reactions. The increases in the positive value of (k1 × T) with increasing

437

temperature are in good agreement with our thermodynamic study.

438

Using the Pore Diffusion Model (PDM) and the Homogeneous Surface Diffusion

439

Model to Describe Adsorption Kinetics

440

In the next few sections, we first use batch kinetic tests to determine intraparticle

441

diffusivities. Then we show that the models can predict Li breakthrough in short fixed

442

bed column tests. Next we will use the validated models to predict full scale

443

performance and we evaluated the impact of multiple ions (that are found in a typical

444

waste water from LIB recovery operations) on Li breakthrough. We also determine

445

the Li recovery regenerating the fixed bed using acid.

446

Batch Kinetic Studies. Adsorption kinetics are important in determining the fixed

447

bed performance. A kinetic experiment was performed with CMO. The initial

448

concentration of Li was 300 mg/L, and the CMO dose was 1 g/L. Figure 7 shows that

449

the Li(I) adsorption capacity grew rapidly for the first 50 min. Then, the Li(I) uptake

450

gradually reached equilibrium at about 120 min. In addition, the equilibrium

451

adsorption capacity increased with the adsorption temperature, further supporting the

452

notion that the adsorption of Li(I) on CMO is an endothermic process.

453

To describe the adsorption kinetic data, a simplified version of the homogeneous

454

surface diffusion model (HSDM)51 was used to determine the surface diffusion

455

coefficient (See section S7 in the Supporting Information). The appropriate

ACS Paragon Plus Environment

Page 22 of 45

Page 23 of 45

Environmental Science & Technology

456

Freundlich isotherm model parameters for a given temperature were used in the

457

HSDM. The 2 h data obtained from the kinetics test are given in the first two columns

458

of Table S7. The equilibrium concentration was 250 mg/L and Ce/C0 = 0.8. The

459

experimental data and solution to the HSDM were used to fit the dimensionless model

460

concentrations, as illustrated in Figure S8. The best fit values are shown in Table S7.

461

The most appropriate simplified HSDM equation is given in Table S6, and Ds was

462

found to be 4.63×10-10 m2/s at 20 °C. We used the simplified HSDM because other

463

researchers could use the simplified HSDM solution that we have provided in the

464

literature.51 In addition, we have provided a simplified HSDM for fixed bed

465

calculations that also could be used for fitting short fixed bed experiments and full

466

scale predictions when competing ions are not present.

467

The most appropriate model to use to fit the kinetic data is the PDM with a Ds

468

value of zero; because, as discussed elsewhere, surface diffusion is not possible for

469

CMO. We calculated the Dp using Eq. 4 and we can calculate a Ds value that would

470

give the same result as the PDM using the surface to pore diffusion flux ratio (SPDFR)

471

equal to 1.0. This gives a Ds value of 2.63× 10-10 m2/s. However, when we fit the data

472

with the simplified HSDM, we found that the SPDFR was 2. (See Section S8 of the

473

Supporting Information.) When we ran the PDM with a Dp that was calculated from

474

Eq. 4, we had an excellent prediction SBA data as discussed below.

475

Hence, when competing ions are not present, we can use either the HSDM or the

476

PDM because they give similar results. And the two model could be used for

477

preliminary design if there are no competing ions.52

ACS Paragon Plus Environment

Environmental Science & Technology

478

Short Bed Adsorber (SBA) Test and Pore Diffusion Model (PDM). We used the

479

simplified batch HSDM to fit the adsorption kinetic data and determined the surface

480

diffusivity. Next, we used the HSDM to predict SBA data using that surface

481

diffusivity (see Figure S9). The HSDM did not predict the data well (R2 ≈ 0.86),

482

because the St for the SBA is smaller than the Stmin (required to establish constant

483

pattern). Consequently, the HSDM prediction has an earlier breakthrough. (However,

484

the simplified HSDM could be used for St values (or EBCT values) above those that

485

correspond to Stmin.) Consequently, we used the PDM to predict Li breakthrough

486

curve and the Dp value that was calculated from Eq. 4. As shown in Figure 8a, we

487

predicted the SBA data using the PDM and the breakthrough curves for 20, 30 and

488

40 °C show increasing column capacity with increasing temperature. The Freundlich

489

isotherm values came from fitting the equilibrium data. The pore diffusion coefficient,

490

film transfer coefficient, and sphericity were calculated from Eq. 4,Eq. S1 and Eq. S4

491

(Dp = 1.60 × 10-10 m2/s, 1.64 × 10-10 m2/s, 1.68 × 10-10 m2/s, kf = 4.79 × 10-5 m/s, 4.91×

492

10-5 m/s, 5.03× 10-5 m/s, and Φ = 0.62 for 20, 30 and 40 °C, respectively). The PDM

493

predictions for the SBA were excellent (R2 ≈ 0. 99). As shown in Figure 8a, in fact,

494

the integrated capacities (areas above the curves as a function of BVs fed) were 1,374,

495

1972 and 2493 BVs, respectively. These are the maximum number of BVs that can be

496

treated, if we had an infinite mass transfer rate (i.e., the breakthrough curve would be

497

a vertical line that occurs when the bed is exhausted).

498

As shown in Figure 8b, the approximately 98.7% of the Li+ was extracted after

499

approximately 66.41 BVs using a 0.1 mol/L HCl solution at 20 °C. Additional, we

ACS Paragon Plus Environment

Page 24 of 45

Page 25 of 45

Environmental Science & Technology

500

predict desorption curve using PDM. The PDM can fit the data well at 1 min EBCT

501

and predict 10 min EBCT data. We calculated the enrichment factor using the BVmax,

502

which the 10 minute EBCT approaches and the number of bed volumes (using 0.1

503

mol/L HCl acid) that are required to desorb Li. The number desorption bed volumes

504

were assumed to be at the point where the Li effluent concentration, C/Co, was less

505

than 0.05. This corresponds to 66.42 and 82.41 desorption bed volumes for 1 and 10

506

minutes of EBCT, respectively. Using these values, we determined that the

507

enrichment factors were 20 and 15, respectively, by dividing BVmax by these

508

desorption volumes. This means that the regenerated solution has a very high

509

concentration of Li and it is easy for recovery Li using evaporation.

510

The above results confirm that a higher temperature is beneficial for increasing

511

the treated number of BVs. Furthermore, reasonable breakthrough and regeneration

512

curves for Li can be predicted using the PDM. As a result, the model can be used to

513

predict full-scale performance, thus reducing the time and cost required as compared

514

to pilot studies.

515

Predicting Full-Scale Performance Using the Validated PDM. The performance of

516

full-scale fixed bed systems was simulated using PDM with the same operating

517

parameters as the SBA tests. The full-scale operating conditions were as follows: (1)

518

the superficial velocity was the same as in the SBA tests, (2) the empty bed contact

519

times (EBCTs) were 2.5, 5 and 10 min; and (3) the influent concentrations were 20,

520

50 and 200 µg/L, which are realistic values for LIB waste water. The treatment

521

objective of 10 µg/L Li(I). The input data used for the PDM are shown in Table S8.

ACS Paragon Plus Environment

Environmental Science & Technology

522

Figure 9a shows the full-scale breakthrough curves for EBCTs of 2.5, 5 and 10

523

min and an influent lithium concentration of 200 µg/L. For a treatment objective of 10

524

µg/L Li(I), the number of BVs that can be treated increased with increasing EBCT.

525

The PDM predicted that approximately 740 BVs can be treated at an EBCT of 2.5

526

min. When EBCT was increased to 5 and 10 min, the number of treated BVs

527

increased to 1,051 and 1,237, respectively, approaching the BVMAX of 1,374 BVs.

528

Figure 9b illustrates the same lithium breakthrough predictions described according to

529

the liters of water treated per gram of dry media. When the EBCT was 2.5, 5 and 10

530

min, approximately 2.2, 2.9 and 3.2 L of water can be treated when the treatment

531

objective of 10 µg/L is reached.

532

For an EBCT of 10 min, the lithium breakthroughs for influent lithium

533

concentrations C0 of 20, 50 and 200 µg/L were calculated and are presented in Figure

534

9c. When C0 was increased from 20 to 50 and 200 µg/L, the number of BVs that could

535

be treated until reaching the 10 µg/L gradually decreased from 1,681 to 1,414 and

536

1237 BVs, respectively. Figure 9d presents the same lithium breakthrough predictions

537

based on the liters of treated water per gram dry media at different C0 values. When

538

C0 was 20, 50 and 200 µg/L, approximately 6.3, 4.5 and 3.2 L of water could be

539

treated per gram of dry media, respectively, until the treatment objective of 10 µg/L

540

was reached.

541

PDM Predictions for Multiple Ions Breakthrough. The PDM that describes the fate

542

of ions in a fixed bed is also was called the IX model when pore diffusion is the only

543

intraparticle transport mechanism and multiple ions are present.52 We used the

ACS Paragon Plus Environment

Page 26 of 45

Page 27 of 45

Environmental Science & Technology

544

selectivity values that were measured and the multicomponent Langmuir equation to

545

description IX equilibria. The PDM can be used for preliminary design and predicts

546

the breakthrough of all competing ions. Figure 10 shows the predictions for an

547

influent containing Li+, Na+, K+, Mg2+ and Ca2+ all at 200 µg/L and pH 10. The other

548

competing ions broke through earlier than Li+ which is a result of a high selectivity

549

for Li+ over the other ions. The CMO can treat 420 BVs at EBCT 10 min BVs for a Li

550

treatment objective of 10 µg/L, which is much larger than other competing ions. The

551

BVs treated when Li was by itself 1,237 for 10 minutes of EBCT and a treatment

552

objective of 10 µg/L. This is 150, 93, 34, 4.6 times greater (for Na, K, Mg, Ca,

553

respectively) than when the other ions are present at concentrations that are typical in

554

LIB waste water.

555

Practical Significance

556

The maximum adsorption capacity (Q0) increased with increasing temperature,

557

and this relationship was described by using the DA model and its site energy

558

distribution (F(E*)). This increase in the adsorption capacity with temperature is

559

useful because the technology can be used in warm and cold climates. The PDM can

560

be used to predict the lithium ion breakthrough for full-scale processes. The IX model

561

can predict the lithium ion breakthrough when the wastewater containing multiple

562

ions. The predictive power of this model will allow engineers to evaluate the

563

performance of a fixed bed containing CMO when multiple ions are present.

564

ACKNOWLEDGEMENTS

ACS Paragon Plus Environment

Environmental Science & Technology

565

This study was financially supported by the National Science Foundation of

566

China (51238002, 51272099), the National Science Fund for Excellent Young

567

Scholars (51422807), the Key Project of Science and Technology Department of

568

Jiangxi Province (20143ACG70006) and the Cultivating Project for Academic and

569

Technical Leader of Key Discipline of Jiangxi Province (20153BCB22005). The

570

authors appreciate support from the Brook Byers Institute for Sustainable Systems

571

(BBISS), Hightower Chair and Georgia Research Alliance at Georgia Institute of

572

Technology.

573

574

REFERENCES

575

(1) Feng, G. F.; Zhang, X. Variation in feature of world lithium industry and its

576

influence on lithium industry in China. Chin. J. Rare Met. 2003, 27, 57-61.

577

(2) Armstrong, A.R.; Bruce, P. G. Synthesis of layered LiMnO2 as an electrode for

578

rechargeable lithium batteries. Nature 1996, 381, 499.

579

(3) Tang, W. P.; Yang, X. J.; Liu, Z. H.; Ooi, K. Preparation of β-MnO2

580

nanocrystal/acetylene black composites for lithium batteries. J. Mater. Chem. 2003,

581

13, 2989.

582

(4) Sun, L.; Qiu, K. Q. Vacuum pyrolysis and hydrometallurgical process for the

583

recovery of valuable metals from spent lithium-ion batteries. J. Hazard. Mater. 2011,

584

194, 378-384.

585

(5) Wang, R. C.; Lin, Y. C.; Wu, S. H. A novel recovery process of metal values from

ACS Paragon Plus Environment

Page 28 of 45

Page 29 of 45

Environmental Science & Technology

586

the cathode active materials of the lithium-ion secondary batteries. Hydrometallurgy

587

2009, 99, 194-201.

588

(6) Li, L.; Ge, J.; Chen, R. J.; Wu, F.; Chen, S.; Zhang, X. X. Environmental friendly

589

leaching reagent for cobalt and lithium recovery from spent lithium-ion batteries.

590

Waste Manage. 2010, 30, 2615-2621.

591

(7) Sun, L.; Qiu, K. Q. Vacuum pyrolysis and hydrometallurgical process for the

592

recovery of valuable metals from spent lithium-ion batteries. J. Hazard. Mater. 2011,

593

194, 378-384.

594

(8) Lemaire, J.; Svecova, L.; Lagallarde, F.; Laucournet, R.; Thivel, P. X. Lithium

595

recovery from aqueous solution by sorption/desorption. Hydrometallurgy 2014, 143,

596

1-11.

597

(9) Tsuruta, T. Removal and recovery of lithium using various microorganisms. J.

598

Biosci. Bioeng. 2005, 100, 562-566.

599

(10) Fan, Y. Q., Jiang, X. X., Wang, S. D.; Zhao, L.; Feng, L. Y. Study on

600

modification of deep-sea manganese nodules to adsorb lithium ion. J. Min. Metall.

601

2008, 3, 41-45.

602

(11) Zhu, G.; Zhu, P.; Qi, P. F.; Gao, C. J. Adsorption and desorption properties of Li+

603

on PVC-H1.6Mn1.6O4 lithium ion-sieve membrane. Chem. Eng. J. 2014, 235, 340-348.

604

(12) Lu, Y. K.; Yan, X. P. An imprinted organic-inorganic hybrid sorbent for selective

605

separation of cadmium from aqueous solution. Anal. Chem. 2004, 76, 453-457.

606

(13) Abe, M., Chitrakar, R. Recovery of lithium from seawater and hydrothermal

ACS Paragon Plus Environment

Environmental Science & Technology

607

water by titanium (IV) antimonate cation exchanger. Hydrometallurgy 1987, 19,

608

117-128.

609

(14) Bauman, W. C.; Burba III, J. L. Composition for the lithium values from brine

610

and process of making/using said composition. US 2001: 6280693.

611

(15) Song, J. J.; Huang, P. P.; Chen L. F.; Qi, Z. W.; Yu, J. G. Properties and

612

mechanism of Li(I) adsorption based on D751 resin. Chem. Ind. Eng. Prog. 2012, 31,

613

370-375.

614

(16) Luo, X. B.; Guo, B., Luo, J. M.; Deng, F.; Zhang, S. Y.; Luo, S. L.; Crittenden, J.

615

Recovery of Lithium from Wastewater Using Development of Li Ion-Imprinted

616

Polymers. ACS Sustain. Chem. Eng. 2015, 3 (3), 460–467.

617

(17) Volkhin, V. V.; Leonteva, G. V.; Onorin, S. A. Izv. Akad. Nauk. SSSR. Neorg.

618

Mater. 1973, 9(6): 1041-1046.

619

(18) Ooi, K.; Miyai, Y.; Katoh, S. Recovery of lithium from seawater by manganese

620

oxide adsorbent. Sep. Sci. Technol. 1986, 21 (8): 755-766.

621

(19) Wang, L.; Meng, C.; Ma, W. Study on Li(I) uptake by lithium ion-sieve via the

622

pH technique. Colloid Surfaces.; A. 2009, 334, 34-39.

623

(20) Han,Y.; Kim, H.; Park, J. Millimeter-sized spherical ion-sieve foams with

624

hierarchical pore structure for recovery of lithium from seawater. Chem. Eng. J. 2012,

625

210, 482-489.

626

(21) Lin, S. H.; Juang, R. S.; Wang, Y. H. Adsorption of acid dye from water onto

ACS Paragon Plus Environment

Page 30 of 45

Page 31 of 45

Environmental Science & Technology

627

pristine and acid-activated clays in fixed beds. J. Hazard. Mater. 2004, 113, 195-200.

628

(22) Thomas, H. C. Heterogeneous ion exchange in a flowing system. J. Am. Chem.

629

Soc. 1944, 66, 1664-1666.

630

(23) Bohart, G. S.; Adams, E. Q. Some aspects of the behavior of charcoal with

631

respect to chlorine. J. Am. Chem.Soc. 1920, 42, 523-544.

632

(24) Clark, R. M. Evaluating the cost and performance of field-scale granular

633

activated carbon systems. Environ. Sci. Technol. 1987, 21, 573-580.

634

(25) Yoon,Y. H.; Nelson, J. H. Application of gas adsorption kinetics I. A theoretical

635

model for respirator cartridge service life. Am. Ind. Hyg. Assoc. J. 1984, 45, 509-516.

636

(26) Crittenden, J. C.; Hutzler, N. J.; Geyer, D. G.; Oravitz, J. L.; Friedman, G.

637

Transport of organic compounds with saturated groundwater flow: Model

638

development and parameter sensitivity. Water Resour. Res. 1986, 22, 271-284.

639

(27) Weber, W. J.; Smith, E. H. Simulation and design models for adsorption

640

processes. Environ. Sci. Technol. 1987, 21 (11), 1040-1050.

641

(28) Smith, E. H.; Weber, W. J. Modeling activated carbon adsorption of target

642

organic-compounds from leachate-contaminated groundwaters. Environ. Sci. Technol.

643

1988, 22 (3), 313-321.

644

(29) Sontheimer, H.; Crittenden, J.; Summers, S. Activated Carbon for Water

645

Treatment, 2nd ed.; DVGW-Forschungsstelle, EnglerBunte Institut, Universitat

646

Karlsruhe: Karlsruhe, Germany, 1988.

647

(30) Friedman, G. Mathematical Modeling of Multicomponent Adsorption in Batch

ACS Paragon Plus Environment

Environmental Science & Technology

648

and Fixed-Bed Reactors. Master’s Thesis, Michigan Technological University,

649

Houghton, MI, 1984.

650

(31) Hand, D. W.; Crittenden, J. C.; Hokanson, D. R.; Bulloch, J. L. Predicting the

651

performance of fixed-bed granular activated carbon adsorbers. Water Sci. Technol.

652

1997, 35, 235-241.

653

(32) Mertz, K. A.; Gobin, F.; Hand, D. W.; Hokanson, D. R.; Crittenden, J. C. Manual:

654

Adsorption Design Software for Windows (AdDesignS); Michigan Technological

655

University: Houghton, MI, 1999.

656

(33) Benenati, R. F.; Brosilow, C. B. Void fraction distribution in bed of spheres.

657

AIChE J. 1962, 8 (3), 351-361.

658

(34) Chu, C. F.; Ng, K. M. Flow in packed tubes with a small tube to particle diameter

659

ratio. AIChE J. 1989, 35 (1), 148-158.

660

(35) Liu, M. K.; Tjiu, W. W.; Pan, J. S.; Zhang, C.; Gao, W and Liu, T. X. A high

661

impact peer reviewed journal publishing experimental and theoretical work across the

662

breadth of nanoscience and nanotechnology. Nanoscale 2014, 6, 4233-4242.

663

(36) Hunter, J. C. Preparation of a new crystal form of manganese dioxide: λ-MnO2. J.

664

Solid State Chem. 1981, 39, 142-147.

665

(37) Ooi, K.; Miyai, Y.; Katoh, S. Extraction reaction with λ-MnO2 in the aqueous

666

phase. Chem. Lett. 1988, 21, 989-992.

667

(38) Feng, Q., Miyai, Y.; Kanoh, H and Ooi, K. Li+ extraction/insertion with

668

spinel-type lithium manganese oxides. Characterization of redox-type and

ACS Paragon Plus Environment

Page 32 of 45

Page 33 of 45

Environmental Science & Technology

669

ion-exchange-sites type, Langmuir 1992, 8(7), 1861-1 867.

670

(39) Koyanaka, H.; Matsubaya, Osamu.; Koyanaka, Y.; Hatta, N. Quantitative

671

correlation between Li absorption and H content inmanganese oxide spinel λ-MnO2. J.

672

Electroanal. Chem. 2003, 559, 77-81.

673

(40) Liu, Y. F.; Feng, Q.; Ooi, K. The synthesis and ion exchange properties of Li+

674

memorized spinel LiAlMnO4. Ion Exch. and Adsorp. 1995, 11 (3), 216 -222.

675

(41) Jiang, J. H.; Dong, D. Q.; Li, J. L. Synthesis of Li4Mn0. 5Ti0. 5O4 and its selectivity

676

to Li+ exchange. Appl. Chem. 2006, 23 (4), 357-361.

677

(42) Zhong, H.; Hui A. Y. Study on the properties of the surface and absorb of Li+

678

ion-exchange of H2TiO3 type. Ion Exch. and Adsorp. 2003, 19 (1), 55-60.

679

(43) Ammundsen, B.; J, Jones. D.; Roziere, J. Mechanism of proton insertion and

680

characterization of the proton sites in Lithium manganate spinels. Chem Mater, 1995,

681

7(11), 2151-2160.

682

(44) Rosseinsky, D. R. Electrode Potentials and Hydration Energies. Theories and

683

Correlations. Chem. Rev. 1965, 65, 467.

684

(45) Dubinin, M. M.; Zaverina, E. D.; Radushkevich, L. V. Sorption and structure of

685

active carbons. I. Adsorption of organic vapors. Zh. Fiz. Khim. 1947, 2, 1351-1362.

686

(46) Helfferich, F. G. Ion exchange; Courier Corporation: New York, 1962.

687

(47) Zhang, W. B.; Gan, W. E.; Lin, X. Q. Electrochemical hydride generation atomic

688

fluorescence spectrometry for the simultaneous determination of arsenic and

ACS Paragon Plus Environment

Environmental Science & Technology

689

antimony in Chinese medicine samples. Anal. Chim. Acta. 2005, 539, 335-340.

690

(48) Khan, A. R.; Singh, R. P. Adsorption thermodynamics of carbofuran on Sn(IV)

691

arsenosilicate in H+, Na+ and Ca2+ forms. Colloids Surf. 1987, 24, 33-42.

692

(49) Dubinin, M. M.; Astakhov, V. A. Development of concepts of volume filling of

693

micropores in adsorption of gases and vapors by microporous adsorbents. Izv. Akad.

694

Nauk SSSR, Ser. Khim. 1971, 1, 5-11.

695

(50) Carter, M. C.; Kilduff, J. E.; Weber, W. J., Jr. Site energy distribution analysis of

696

preloaded adsorbents. Environ. Sci. Technol. 1995, 29 (7), 1773-1780.

697

(51) Zhang, Q.; Crittenden, J.; Hristovski, K.; Hand, D.; Westerhoff, P. User-oriented

698

batch reactor solutions to the homogeneous surface diffusion model for different

699

activated carbon dosages. Water. Res. 2009, 43 (7): 1859-1866.

700

(52) Hokanson, D. R.; Clancey, B. L.; Hand, J. C.; Crittenden, J. C.; Carter, D. L.;

701

Garr II, J. D. Ion exchange Model Development for the international space station

702

water processor, J. Aerospace. 1995, 104, 977-987.

703 704 705 706 707

ACS Paragon Plus Environment

Page 34 of 45

Page 35 of 45

Environmental Science & Technology

Figure and Table

Figure 1. Preparation of the lithium ion cages (CMO)

ACS Paragon Plus Environment

Environmental Science & Technology

Figure 2. XRD patterns of MnO2 oxide (MO), Li4Mn5O12 tri-oxide precursor (LMO) and MnO2 ion cages (CMO).

ACS Paragon Plus Environment

Page 36 of 45

Page 37 of 45

Environmental Science & Technology

Figure 3. (a) and (b) shows the SEM image of MO, (c) and (d) shows SEM images of LMO and CMO, respectively.

ACS Paragon Plus Environment

Environmental Science & Technology

Figure 4. XPS characterization of CMO before and after Li(I) adsorption. (a) and (b) XPS survey spectra of the samples, high-resolution XPS spectra of (c) and (d) Mn 2p, (e) and (f) O 1s (Olatt (lattice oxygen), Oad (adsorb oxygen) and OH2O (chemisorbed oxygen species)) and (g) and (h) Li 1s before and after Li(I) adsorption, respectively.

ACS Paragon Plus Environment

Page 38 of 45

Page 39 of 45

Environmental Science & Technology

Figure 5. Adsorption isotherm at different temperatures for Li(I) adsorption on CMO. The initial Li(I) concentration was ranged in 20-450 mg/L. Adsorbent dose: 1g/L, Total solution volume: 20 mL, pH: 10.1, Temperature: 20, 30 and 40 °C.

ACS Paragon Plus Environment

Environmental Science & Technology

Figure 6. Regeneration cycles of CMO. Adsorbents mass: 200 mg, The initial Li(I) concentration: 300 mg/L, pH: 10.1, Temperature: 30 °C

ACS Paragon Plus Environment

Page 40 of 45

Page 41 of 45

Environmental Science & Technology

Figure 7. Adsorption kinetics for Li(I) adsorption on CMO at different temperatures.

ACS Paragon Plus Environment

Environmental Science & Technology

Figure 8. (a) The experimental data and the predictions from PDM for Li(I) breakthrough in SBA tests. The initial Li(I) concentration: 200 μg/L, pH: 10.1, Temperature: 20, 30 and 40 °C. (b) Fixed bed column regeneration test using 0.1 mol/L HCl solution at EBCT of 1 min, and PDM prediction for the desorption at EBCTs of 1 min and 10 min at 20 °C, respectively.

ACS Paragon Plus Environment

Page 42 of 45

Page 43 of 45

Environmental Science & Technology

Figure 9. (a) The lithium breakthroughs for Li(I) with the initial concentration of 200 μg/L, by using the validated pore surface diffusion model with empty bed contact times (EBCTs) at 2.5, 5 and 10 min, respectively. (b) The same lithium breakthrough predictions in Figure a, but with liters of treated water per gram dry media at different EBCTs at 20 °C. (c) The lithium breakthroughs for Li(I) with the initial concentration of 20, 50 and 200 μg/L, respectively, by using the validated pore surface diffusion model with a fixed EBCT at 10 min. (d) The same lithium breakthrough predictions in Figure c, but expressed as liters of treated water per gram dry media at different initial Li concentration at 20 °C.

ACS Paragon Plus Environment

Environmental Science & Technology

Figure 10. The mathematical model consists of ion exchange (IX) models for describing the selectivity behavior of the competition ion on CMO bed. The competition for adsorption of Li+, Na+, K+, Mg2+, and Ca2+ ions with respect to H+ ions was determined. The initial Li(I) concentration: 200 μg/L, The empty bed contact times (EBCT): 10 min, Temperature: 20 °C

ACS Paragon Plus Environment

Page 44 of 45

Page 45 of 45

Environmental Science & Technology

Table 1. Selectivity of various ions for CMO

Metal ions

Ionic radius (pm)

Li(I)

76

118.48

Na(I)

102

1.68

K(I)

138

2.35

Mg(II)

72

28.21

Ca(II)

100

49.09

α ki

Table 2. Langmuir and Freundlich parameters for Li(I) adsorption on CMO

Equations

Langmuir model

Freundlich model

T

Q0

kl

(oC)

(mg/g)

(L/mg)

20

56.05

0.013

0.9756

0.45

2.854

0.9837

30

71.22

0.011

0.9737

0.51

3.216

0.9879

40

76.80

0.012

0.9798

0.58

3.777

0.9902

R2

1/n

kF

R2

(mg1-(1/n)L1/n/g)

ACS Paragon Plus Environment