Influence of Active Layer on Separation Potentials of Nanofiltration

Subscriber access provided by Fudan University

Article

Influence of active layer on separation potentials of nanofiltration membranes for inorganic ions Shardul Sudhir Wadekar, and Radisav D. Vidic Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b05973 • Publication Date (Web): 17 Apr 2017 Downloaded from http://pubs.acs.org on April 19, 2017

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 26

Environmental Science & Technology

1

Influence of active layer on separation potentials of nanofiltration membranes for

2

inorganic ions

3 4 5 6 7 8 9 10 11 12 13 14 15 Shardul S Wadekar1, Radisav D Vidic*1,2

16 17

1

Department Chemical and Petroleum Engineering, University of Pittsburgh, PA, 15261

18

2

Department Civil and Environmental Engineering, University of Pittsburgh, PA, 15261

19 20

*Corresponding Author. Address: 742 Benedum Hall, 3700 O'Hara St, Pittsburgh, PA 15261 Email: [email protected]; Tel: 412-624-9870; Fax: 412-624-0135

21 22

ACS Paragon Plus Environment

1


23

Page 2 of 26

Abstract

24

Active layers of two fully aromatic and two semi-aromatic nanofiltration membranes were

25

studied along with surface charge at different electrolyte composition and effective pore size to

26

elucidate their influence on separation mechanisms for inorganic ions by steric, charge and

27

dielectric exclusion. The membrane potential method used for pore size measurement is

28

underlined as the most appropriate measurement technique for this application owing to its

29

dependence on the diffusional potentials of inorganic ions. Crossflow rejection experiments with

30

dilute feed composition indicate that both fully aromatic membranes achieved similar rejection

31

despite the differences in surface charge, which suggests that rejection by these membranes is

32

exclusively dependent on size exclusion and the contribution of charge exclusion is weak.

33

Rejection experiments with higher ionic strength and different composition of the feed solution

34

confirmed this hypothesis. On the other hand, increase in the ionic strength of feed solution when

35

the charge exclusion effects are negligible due to charge screening strongly influenced ion

36

rejection by semi-aromatic membranes. The experimental results confirmed that charge

37

exclusion contributes significantly to the performance of semi-aromatic membranes in addition

38

to size exclusion. The contribution of dielectric exclusion to overall ion rejection would be more

39

significant for fully aromatic membranes.

40 41

Keywords: Nanofiltration; active layers; inorganic ion separation; charge exclusion; steric

42

hindrance

43


2

Page 3 of 26

44


Introduction

45

Application of membrane technologies for water purification gained greater attention in

46

recent years due to population boom and worldwide industrialization.1 Major technological

47

advancement and cost reduction lead to increased use of reverse osmosis (RO) and nanofiltration

48

(NF) membranes in desalination, wastewater treatment and reclamation.2,

49

membranes are predominantly thin film composite (TFC) membranes consisting of three layers:

50

the topmost layer is the selective active layer followed by the microporous polysulfone support

51

layer and a non-woven fabric layer for mechanical strength.3 Of these, the topmost dense layer

52

with thickness of about a few hundred nanometers is the most important layer responsible for

53

permeability, ionic selectivity, fouling resistance, roughness and hydrophilicity of the composite

54

membrane.4 Ion rejection by this active layer is due to three different separation potentials: Steric

55

hindrance (pore size effects), Donnan exclusion (by fixed surface charge) and dielectric

56

exclusion (by Born effect and image forces).5, 6 Modeling efforts have attempted to explain these

57

effects and predict the rejection behavior of NF membranes.7-9 The separation by NF membranes

58

was initially modeled using the Donnan steric partitioning pore model (DSPM)7 but the steric,

59

electric and dielectric model (SEDE)8, 9 was developed later due to the inability of DSPM to

60

predict the rejection of divalent cations. SEDE model is a four parameter (i.e., membrane’s

61

effective pore size, thickness to porosity ratio, volume charge density and the dielectric constant

62

of solution inside the membrane pores) model and is able to predict the rejection performance of

63

NF membranes reasonably well.

3

Modern NF

64

Interfacial polymerization (IP) is the most commonly used technique for synthesizing these

65

TFC membranes. The amine monomers in water are brought into contact with acid chloride

66

monomers in solvent to form a thin film of polyamide on the substrate.10-13 Two of the most


3


Page 4 of 26

67

commercially successful recipes to make the polyamide films are: 1) 1,3-benzenediamine (m-

68

phenylenediamine) (MPD) with trimesoyl chloride (TMC) and piperazine (PIP) with TMC.3 In

69

the first reaction scheme, both monomers (i.e., MPD and TMC) are aromatic and hence the

70

membrane can be designated as fully aromatic (FA) while the membranes prepared using the

71

second scheme can be designated as semi-aromatic (SA) because PIP is an aliphatic monomer.

72

These membranes are often coated with different groups to alter membrane properties.14 For

73

example, it has been found that a neutral hydrophilic coating can affect surface charge, surface

74

roughness, permeability and salt rejection of TFC membranes.15 Thus, understanding the role of

75

active layers in achieving ion rejection by a particular separation mechanism is of utmost

76

importance to understand the potential use of these membranes for specific application.

77

Tang et al.4, 15 characterized seventeen commercially available RO and NF membranes and

78

have shown how the active layer chemistries and coatings affect hydrophilicity, surface

79

roughness and permeability of these membranes. Verissimo et al.10 evaluated the effect of

80

combining different aliphatic monomers (i.e., PIP, 1,4-bis(3-aminopropyl)-piperazine (DAPP),

81

N,N’-diaminopiperazine (DAP) and N,N’-(2-aminoethyl)-piperazine (EAP)) with TMC on the

82

performance, surface morphology and charge of composite semi-aromatic polyamide

83

membranes. They found that water permeability was the highest for DAP-TMC membrane but

84

that PIP-TMC membrane performed better in terms of salt rejection. In a similar study, Li et al.11

85

evaluated the effect of other aliphatic monomers on salt rejection and anti-fouling properties of

86

thin-films. Ahmad et al.12 found that permeate flux and separation capabilities of polyamide NF

87

membranes greatly depend on the diamine ratio and the IP reaction times while effects of

88

polyamide chemistry on amino acid separation have also been compared.13 Other studies have

89

also evaluated the ionization behavior of functional groups16-18 and surface heterogeneity19, 20 of


4

Page 5 of 26


90

active layer towards understanding the rejection by these membranes. However, the underlying

91

separation mechanisms of NF membranes with different active layers have not yet been fully

92

unraveled because none of the previous studies attempted to fully characterize all three

93

separation potentials as a function of active layer chemistry.

94

Since semi-aromatic (polypiperazine amide) and fully aromatic (polyamide) membranes are

95

the most commonly used NF membranes, the aim of this study was to investigate the separation

96

potentials of these active layer types in different applications. Separation potentials (i.e., steric,

97

Donnan (charge) and dielectric exclusion) of these active layers were characterized to elucidate

98

their relative contribution to rejection of inorganic ions in an effort to develop new applications

99

and to improve membrane selection process for the separation of inorganic ions. Sulfate was the

100

key ion selected for this study since it is found in many wastewaters, surface and ground waters

101

at widely different concentrations21 and since it is one of the contaminants of concern in

102

abandoned mine drainage, which is a pervasive problem in many parts of the US.22,

103

membranes used in this study are commercially available and hence the information about their

104

separation potential and performance is relevant to their application in practice.

105

Experimental

106

Membranes and chemicals. Four commercially available flat sheet NF membranes were used

107

in the study. NF90 and NF270 membranes were purchased from DOW Filmtech (Edina, MN);

108

TS40 and TS80 were purchased from Sterlitech Corporation (Kent, WA). Key properties of these

109

membranes reported in the literature and those provided by the manufacturers are shown in Table

110

1. Deionized (DI) water used for water permeability experiments (resistivity = 18 kohm.cm-1)

111

was obtained using MilliQ water system (Millipore, Billerica, MA). All chemicals used were

112

analytical grade and were purchased from Fisher Scientific (Pittsburgh, PA).


23

All

5


113

Table 1. Membranes used in this study Membranes Polymer MWCO

114

Page 6 of 26

21

Fully Aromatic (FA) Semi-Aromatic (SA) NF90 TS80 NF270 TS40 Polyamide Polyamide Polypiperazine- Polypiperazineamide amide

~200a

~150c

~200-300b

~200c

24 c

a , b , Provided by manufacturer

115 116

Apparatus and filtration process. All experiments were carried out in the laboratory-scale

117

SEPA-CFII test cell (GE Osmonics, Minnetonka, MN) shown in Fig. S2 (Supporting

118

Information) with usable membrane area of 140 cm2 and has been descriped in detail

119

elsewhere25. Pump power (Hydra-cell diaphragm pump, Wanner Engineering, MN), feed control

120

valve and concentrate control valve were used to adjust the desired feed pressure (20 bar) and

121

flow rate (1 GPM), which were held constant throughout the study. All experiments were

122

performed in total recirculation mode at a constant feed pH of 5.6±0.1. Temperature was

123

maintained at 22±1°C using an immersed cooling coil connected to a chiller (6500 series,

124

Polyscience, Niles, IL). Prior to the experiment, each membrane was immersed in DI water for at

125

least 24 hours to ensure complete wetting. Each membrane was first compacted with DI water at

126

50 bar and then used to filter DI water until a stable flux (LMH/bar) was reached (typical

127

stabilization times ranged between 20-24 hours). Once, a stable flux had been established, the

128

feed was adjusted to the required composition and the system was allowed to equilibrate for two

129

hours. The permeate flux was measured over the next two hours during which samples were

130

collected every 15 min for chemical analysis.

131

Sulfate, magnesium and calcium were introduced as Na2SO4⋅10H2O, CaCl2⋅2H2O and

132

MgCl2⋅6H2O salts. The filtration experiments were carried out at dilute (low) and high


6

Page 7 of 26


133

electrolyte concentrations to evaluate the relative importance of the Donnan (charge) exclusion

134

effects. Experiments with dilute feed were performed at sulfate concentration of 96 mg/L with

135

calcium and magnesium concentrations up to 40 mg/L and 24 mg/L (i.e., 1mM each),

136

respectively. For experiments at high ionic strength, sulfate concentration was adjusted to 650

137

mg/L and magnesium and calcium ions were introduced at 1,000 mg/L each. This feed

138

composition was chosen for two reasons: it assured that the surface charge is screened at high

139

concentrations and it also represents abandoned mine drainage26, which is a pervasive

140

environmental concern in many areas of the US.23 All cations and anions were analyzed using

141

inductively coupled plasma-optical emission spectroscopy (ICP-OES) system (5100 ICP-OES,

142

Agilent Technologies, Santa Clara, CA) and ion chromatography (IC) system (Dionex ICS-1100

143

with IonPac AS22 carbonate eluent anion-exchange column, Dionex, Sunnywale, CA),

144

respectively.

145

Attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR). FTIR

146

was used to determine the chemical composition of the active layer for all four nanofiltration

147

membranes selected for this study. Infrared spectra were obtained using Nicolet 6700 (Thermo

148

Scientific, Pittsburgh, PA) FTIR spectrometer with the active layer of the membrane pressed

149

tightly against the crystal. At least two replicates were obtained for each membrane type and

150

each spectrum was averaged from 256 scans collected from 1800 cm-1 to 800 cm-1.

151

Membrane pore size measurement. Membrane potential technique27 was used to measure the

152

effective pore sizes of the NF membranes. The membrane sample (exposed area of 12.5 cm2)

153

was held between two acrylic half-cells (700 cm3 each) filled with NaCl solutions at different

154

concentrations but identical pH, temperature and hydrostatic pressure. The ratio of ion

155

concentrations in the two half cells, /∆ , was maintained at a constant value of 2, with the


7


Page 8 of 26

156

active layer always facing towards the half-cell with higher concentration. NaCl concentrations

157

ranged between 3-250 mM. Each experiment was repeated at least twice and the electrodes were

158

also interchanged between the two compartments to cancel the asymmetric potential effect.27

159

Prior to each experiment, the membrane was immersed in solution of lower concentration for at

160

least 24 hours to ensure saturation of the support layer and to avoid any interference from the

161

concentration gradient in the support layer. All the experiments were carried out at ambient

162

temperature of 22°C with continuous stirring of each cell using magnetic stirrers. The output

163

from Ag/AgCl electrodes (RE-5B, BASi Electronics, West Lafayette, IN) submerged in each cell

164

was amplified (INA826EVM, Gain = 97.76, Texas Instruments, Dallas, TX) and measured by a

165

multimeter (Fluke 21 Series II, Fluke Corporation, Everett, WA). The membrane pore radius was

166

calculated using the procedure described in Supporting Information.

167

Zeta potential measurement. Zeta potential of the membranes was analyzed using Surpass 3

168

Electrokinetic Analyzer (EKA) equipped with the Adjustable Gap Cell (AGC) (Anton Paar,

169

Ashland, VA). For each measurement, two 10 mm × 20 mm membrane samples were inserted

170

into the AGC and 1 mM KCl solution was used as electrolyte to obtain the isoelectric point of

171

each membrane. An automatic pH sweep from ~5.6 to 2 was accomplished by the addition of

172

0.05 M HCl solution and from ~5.6 to 10 using 0.05 M NaOH. Following the isoelectric point

173

determination, the membranes were also tested with the following electrolyte solutions at pH

174

5.6±0.1: 1mM Na2SO4, 1mM Na2SO4 + 1mM CaCl2 and 1mM Na2SO4 + 1mM MgCl2. These

175

experiments were designed to determine the change in zeta potential of these NF membranes

176

with the addition of divalent cations using sulfate as the base anion. Each of these experiments

177

was repeated at least four times with a maximum standard deviation of 4 mV.

178


8

Page 9 of 26


179 180 181

Results and Discussion

182

ATR-FTIR. ATR-FTIR spectra of the four NF membranes in the range 1800–800 cm-1 are

183

shown in Fig. 1. This range would reflect both the polyamide active layer and the polysulfone

184

support layer as the FTIR signal has relatively deep penetration (>300 nm).4 Since the focus of

185

this study is to differentiate between fully and semi-aromatic membranes, only the relevant peaks

186

are discussed here and information about all other peaks is included in Supporting Information.

187

As seen from Fig. 1, the peaks at ~1664, 1610 and 1545 are present only for NF90 and

188

TS80 membranes and are absent for NF270 and TS40 membranes. The peak at ~1664 cm-1 can

189

be assigned to C=O stretching (dominant contributor), C–N stretching and C–C–N deformation

190

vibration in a secondary amine group.29, 30 The peak at ~1610 cm-1 is due to N–H deformation

191

vibration for the aromatic amide31 while the peak at ~1545 cm-1 is due to amide II band for the

192

N–H in-plane bending and N–C stretching vibration of CO–NH group.4 These three peaks are

193

clearly seen to be absent from the spectra obtained for the semi-aromatic membranes (NF270 and

194

TS40). On the other hand, the peak at ~1630 cm-1 is observed only in the case of NF270 and

195

TS40 and is absent for NF90 and TS80 membranes. This peak is due to amide I band

196

(poly(piperazineamide)).32 Tang el al.4 have shown that NF90 and NF270 are uncoated NF

197

membranes by comparing the FTIR and XPS spectra of several commercially available NF

198

membranes. Fig. 1 shows that the spectra for TS80 is identical to NF90 and that of TS40 is

199

identical to NF270. Based on these results, it can be concluded that TS80 and NF90 are uncoated

200

fully aromatic polyamide membranes and TS40 and NF270 are uncoated semi-aromatic

201

polyamide membranes. Therefore, this study included two membranes that are truly


9


Page 10 of 26

202

representative of each category of membrane chemistry (i.e., full aromatic (MPD-TMC) and

203

semi-aromatic (PIP-TMC)) without any coating or any modifications of the polyamide and

204

polypiperazine amide active layers.

205

Pore size measurement. Variation of membrane potential (∆ ) with chloride concentration is

206

shown in Fig. 2. Each point on this figure represents a mean of at least four measurements at

207

each concentration of the single salt (NaCl). As can be seen from Fig. 2, the membrane potential

208

first increases with an increase in chloride concentration and then plateaus, which corresponds to

209

the diffusion potential (i.e., limiting value at high concentration) where both the image forces

210

and the Donnan (charge) effects are screened.27 A maximum standard deviation of ±0.6 mV was

211

observed for membrane potential values when chloride concentration in solution was below 0.1

212

M and it was only ±0.1 mV at the plateau of membrane potential. The asymmetry potential was

213

below ±0.1 mV.

214

The effective mean pore sizes of the four NF membranes were evaluated using plateau

215

levels of the membrane potential (i.e., diffusion potential) where the diffusion potentials of 6.14

216

mV, 6.0 mV, 5.66 mV and 5.45 mV were measured for NF90, TS80, TS40 and NF270

217

membranes, respectively. Using the procedure described in Supporting Information, the effective

218

mean pore radii of 0.68±0.02 nm, 0.71±0.02 nm, 0.80±0.03 nm and 0.87±0.02 nm were

219

calculated for NF90, TS80, TS40 and NF270 membranes, respectively. Thus, the calculated

220

effective pore sizes of these four membranes differed only slightly (i.e., 0.19 nm difference

221

between NF90 and NF270). Also, it was noted that the two semi-aromatic membranes (TS40 and

222

NF270) had larger effective pore sizes as compared to the two fully aromatic membranes (NF90

223

and TS80). Lo et al.33 used Density Functional Theory (DFT) analysis to suggest that the

224

reaction between MPD and TMC (i.e., FA membranes) is much facile as compared to the


10

Page 11 of 26


225

reaction between PIP and TMC (i.e., SA membranes), which explains the greater crosslinking

226

and smaller effective pore size for fully aromatic membranes.

227

The mean pore size of NF membranes can been measured using three different

228

techniques: 1) atomic force microscopy (AFM)34, 35, 2) retention of neutral organic solutes of

229

different molecular weights7,

230

Selection of a particular method should be based on specific application because NF membranes

231

behave differently in different feed solutions. This study is focused on the rejection of inorganic

232

ions and hence membrane potential technique is more suitable for pore size measurements

233

because it is based on the diffusional potential of ionic species through the membrane pores.

234

Hilal et al.35 used AFM technique and reported the membrane pore radius of NF90 and NF270 to

235

be 0.257 nm and 0.341 nm, respectively. Nghiem et al.36 reported the pore radius of NF90 and

236

NF270 as 0.34 nm and 0.42 nm, respectively, by modelling the retention data of organic solutes

237

of different molecular weights. Mean pore sizes determined in this study differ for both

238

membranes but they agree that pore size of NF90 is smaller than the pore size of NF270

239

membrane. Similarly, by modelling the retention data of neutral organic solute, mean pore radius

240

of TS80 membrane (≈0.52 nm) has been reported to be smaller than that of TS40 membrane

241

(≈0.65 nm).37

242

AFM provides a semi-visual determination of the pore size since only the membrane surface is

243

evaluated and no transport of species takes place through the membrane. It is known that the

244

pores in the NF membranes are non-homogeneous and hence surface evaluation of pores cannot

245

accurately determine the mean effective membrane pore radius. In case of modeling the retention

246

data of neutral organic solutes of different molecular weights, actual retention experiments have

247

to be carried out in order to measure the rejections, which involves introducing a convective

36

and 3) membrane potential analysis (used in this study).9,


27

11


Page 12 of 26

248

factor in these experiments. Hence, in addition to the dependence of the effective membrane pore

249

radius on the steric partitioning coefficient, it is now also dependent on the convective hindrance

250

factor. In the case of membrane potential technique, the effective pore radius is dependent on

251

only the diffusive hindrance factor or the steric partitioning coefficient since there are no

252

convective forces at play. This difference in the convective and diffusive hindrance factors

253

contributes to the observed differences in the measured effective membrane pore radii between

254

the membrane potential technique and neutral organic molecule retention technique.

255

Apart from the membrane pore size measurements, the results in Fig. 2 can also be used to

256

determine the approximate solute concentration where the Donnan (charge) separation potential

257

becomes negligible. As can be seen from this figure, the Donnan (charge) separation potential

258

has already canceled out at chloride concentration of 0.09 M for all four membranes as

259

evidenced by the leveling of the membrane potential. This observation indicated that hindered

260

diffusion and convection are the only relevant transport mechanisms in NF systems where the

261

ionic strength of the feed is above about 0.1 M.

262

Zeta potential. Fig. 3(a) shows zeta potential of the four nanofiltration membranes in the pH

263

range between 2-10 using 1 mM KCl as the electrolyte. It can be seen from this figure that the

264

isoelectric point (IEP) for NF90, TS80, NF270 and TS40 is 4.60, 2.54, 2.43 and 2.40,

265

respectively. An IEP in the neighborhood of pH 4 typically indicates that the surface is either

266

neutral or inert.38 Hence, the zeta potential for NF90 suggests that it has similar concentrations of

267

dissociable acidic carboxylic groups and basic amine groups. The remaining three membranes

268

have low IEP, which indicates dominance of dissociable acidic carboxylic groups over

269

dissociable basic amine groups. Artug et al.24 and Tu et al.39 have reported the IEP’s of NF90 and

270

NF270 at 4.2 and 2.8 and 4 and 2.8, respectively. Since the reaction between MPD and TMC


12

Page 13 of 26


271

(i.e., FA membranes) is much facile as compared to the reaction between PIP and TMC (i.e., SA

272

membranes)33, there will be more unreacted acyl chloride in the active layer in the case of PIP-

273

TMC (i.e., SA type) and the charged carboxylic entities will impart more negative surface

274

potential to these membranes. TS80 membrane has IEP very close to that of the SA type

275

membranes even though it has been confirmed to be a FA type membrane (Fig. 1). The excess

276

carboxylic groups on TS80 membrane suggests that this membrane may have been immersed in

277

the TMC solution for a longer time during the interfacial polymerization process as compared to

278

NF90.

279

Zeta potentials of the selected membranes were also measured using three different

280

electrolytes at pH = 5.6±0.1 as shown in Fig. 3(b). The results in this figure suggest that both SA

281

type membranes (i.e., NF270 and TS40) have a more negative zeta potential than both FA type

282

membranes (i.e., NF90 and TS80), which suggests that the contribution of Donnan (charge)

283

exclusion towards the separation by SA type membranes would be greater than for FA type

284

membranes. Fig. 3(b) also shows an increase in zeta potential with the addition of divalent

285

cations to the electrolyte solution for all four membranes. Childress et al.40, 41 proposed that the

286

complex formation or electrostatic interactions between the divalent cations and the negatively

287

charged membrane surface lead to adsorption of cations on membrane surface and an increase in

288

zeta potential, which was also supported by other studies.42, 43 It can also be observed from Fig.

289

3(b) that the relative increase in zeta potential is about the same for TS80, TS40 and NF270

290

membranes but is less pronounced for NF90 membrane, which is due to the fact that NF90 is less

291

electronegative than the other three membranes (Fig. 3(a)). Also, the increase in zeta potential is

292

more pronounced with the addition of Ca2+ as compared to the Mg2+, which can be explained by

293

the fact that Ca2+ has higher diffusivity and smaller stokes radius than Mg2+ (Table S2,


13


Page 14 of 26

294

Supporting Information) and can approach closer to the membrane surface than Mg2+ to exert a

295

more pronounced impact on zeta potential. The main conclusion from the zeta potential study is

296

that the semi-aromatic membranes are more electronegative and have more fixed charges on the

297

membrane surface compared to the fully aromatic membranes.

298

Membrane Performance. Membrane permeability was measured with DI water where the

299

permeate flux was monitored for two hours and is shown in Fig. 4. Pure water permeability

300

correlates well with the membrane pore radius measured using membrane potential method. An

301

increase in the permeate flux of 143% (4.7 LMH/bar for NF90 to 11.4 LMH/bar for NF270) was

302

measured with an increase in the effective pore radii of 28% (0.68 nm for NF90 to 0.87 nm for

303

NF270). Similar values have previously been reported by Hilal et al.44 and Santafe-Moros et al.45

304

for NF90 and NF270 membranes.

305

Fig. 5 shows rejections of various ions for both dilute (Fig. 5(a), (b)) and concentrated (Fig.

306

5(c), (d)) feed strengths. These experiments were conducted to determine the dominant

307

separation mechanism knowing that the surface charge is screened at high ionic strength and that

308

the contribution of Donnan effect to separation potential would be negligible.

309

Fig. 5 shows that sulfate rejection was always greater than 98% and that the FA type

310

membranes performed better than SA type membranes in all cases. High sulfate rejection can be

311

explained by negative surface charge of all four membranes for all solution compositions

312

investigated in this study (Fig. 3). The rejection order of the ionic species observed in these

313

experiments is: R(SO42-) > R(Cl-) (for anions) and R(Mg2+) > R(Ca2+) > R(Na+) (for cations);

314

rejection of magnesium being marginally more or equal to the rejection of calcium . According

315

to the Donnan exclusion theory for single salt solutions and negatively charged membrane

316

surfaces, the sequence of rejection of cations should be in the order R(Na+) > R(Mg2+) ≈


14

Page 15 of 26


317

R(Ca2+).46 With single salt solutions, the effect of single valence cations on the electronegativity

318

of the negatively charged membrane surface will be less drastic as compared to that of multi-

319

valence cations47 and hence, the rejection of Na+ is expected to be greater than Mg2+ and Ca2+.

320

The rejection order of R(Mg2+) > R(Ca2+) > R(Na+) observed in Fig. 5 with multiple ions in the

321

feed can be explained by ionic diffusivity and Stokes radius of magnesium, calcium and sodium.

322

With multiple ions in the feed, the rejected sulfate will be largely electro-neutralized by divalent

323

cations that will also experience greater steric rejection potential than the monovalent cations

324

owing to their larger Stokes radii (Table S2, Supporting Information). Hence, rejection of Mg2+

325

and Ca2+ will be greater than Na+. When comparing the rejection of Mg2+ and Ca2+, the ionic

326

diffusivity and Stokes radius play a major role. Because Mg2+ has lower ionic diffusivity and

327

larger stokes radius than Ca2+ (Table S2, Supporting Information), it will be rejected more than

328

Ca2+. In the case of anions, SO42- with a valence of -2 will experience greater electronegative

329

repulsion from the negatively charged membrane surface as opposed to Cl-. In addition, SO42-

330

has larger Stokes radius and lower ionic diffusivity than Cl- (Table S2, Supporting Information),

331

which will also contribute towards greater rejection of SO42- than Cl-.

332

In the case of dilute feed composition (Fig. 5 (a) and (b)), both Donnan (charge) and steric

333

effects would contribute to ion rejection.47 FA type membranes achieved >98% rejection of

334

calcium and magnesium ions while the SA type membranes achieved 92-94% rejection of these

335

cations. Also, FA membranes achieved >95% rejection of sodium and chloride ions in each case

336

while the rejection of these ions by SA membranes ranged between 62-73%. The fact that SA

337

membranes achieved lower ion rejection despite having more electronegative surfaces than FA

338

membranes (Fig. 3) clearly suggests that the steric rejection potential can be more dominant in

339

determining the overall rejection by a particular membrane. Lower rejection by SA type


15


Page 16 of 26

340

membranes is explained by their larger effective pore size compared to FA type membranes.

341

Comparison between two membranes of the same category (i.e. FA membrane type) shows that

342

both FA membranes achieved similar rejection of all ions although TS80 membrane is more

343

electronegative than NF90 membrane (Fig. 3). Such behavior suggests that the contribution of

344

charge effects (Donnan potential) towards rejection by FA type membranes is weak. Slightly

345

better rejection of sodium and chloride ions by NF90 membrane is due to smaller pores

346

compared to TS80 membrane.

347

For high feed ionic strength (i.e., addition of 1,000 mg/L of calcium or magnesium to the

348

feed), the rejection of divalent ions (i.e., sulfate, magnesium and calcium) changed only slightly

349

(3% or less) for both FA and SA type membranes (Fig. 5(c) and (d)). However, the rejection of

350

sodium and chloride ions was affected by the elevated ion concentrations in the feed (Fig. 5 (c),

351

(d)) as compared to those achieved with dilute feed (Fig. 5 (a), (b)). The decrease in the rejection

352

of sodium with an increase in feed concentration was significant for SA membranes where it

353

decreased by at least 30% compared to the dilute feed conditions while it changed by 5% or less

354

for FA membranes. This decrease in rejection for semi-aromatic membranes occurs when the

355

Donnan separation potential has been screened out and is no longer assisting the separation,

356

which suggests that Donnan potential contributed significantly to separation at dilute feed

357

conditions for SA membranes. Additional ion rejection experiments were also carried out at low

358

and high ionic feed strengths with different feed composition. The results from these experiments

359

also support the conclusion of dominant dependence of fully aromatic membranes on steric

360

exclusion potential and that of semi-aromatic membranes on both steric and Donnan exclusion

361

potential and are discussed in the Supporting Information (Fig. S1). It is also important to note

362

that the addition of cations to the feed is accompanied by the increase in chloride concentration


16

Page 17 of 26


363

because all cations were added as their chloride salts. The quantum of chloride ions diffusing to

364

the permeate side is dependent on the sodium ions diffusing through the membrane to maintain

365

electroneutrality because the divalent cations are effectively rejected by all membranes. Hence,

366

there is a relative increase in chloride rejection with an increase in the ionic strength of the feed.

367

Crossflow rejection experiments with dilute feed indicated that fully aromatic membranes

368

achieved similar rejection despite the differences in surface charge and suggest that the rejection

369

by FA membranes is predominantly dependent on the pore size effects (i.e., size exclusion) and

370

that the contribution of Donnan (charge) effects is rather weak. Also, rejection experiments with

371

high ionic strength feed confirmed the weak contribution of Donnan (charge) exclusion effects

372

on the rejection of inorganic ions by these membranes. On the other hand, increase in the ionic

373

strength of the feed solution when the Donnan exclusion effects are negligible due to charge

374

screening strongly influenced ion rejection by semi-aromatic membranes, which confirmed that

375

the Donnan (charge) exclusion contributes significantly to the performance of SA membranes in

376

addition to steric hindrance. These results suggest that the fully aromatic nanofiltration

377

membranes would be preferred over semi-aromatic nanofiltration membranes in applications that

378

require complete removal of inorganic ions (e.g., desalination) while the later would be better

379

suited for applications requiring fractional rejections of ionic species (e.g., dairy industry).

380

Another important mechanism for the separation by nanofiltration membranes is by

381

dielectric exclusion, which is explained in terms of: (1) Born effects that occur by changes to the

382

equilibrium and dynamic properties of the solvent in the confined geometry of the nanopores,

383

and (2) image forces due to the difference in dielectric constants between the membrane matrix

384

and the solution.48 Comparison of dielectric exclusion potential for the two membrane types (i.e.,

385

fully aromatic and semi-aromatic) is difficult because dielectric exclusion occurs by several


17


Page 18 of 26

386

concomitant factors and only a qualitative discussion is possible. When considering only the

387

effect of image forces, the contribution of dielectric exclusion to rejection is expected to decrease

388

with an increase in the ionic strength of the feed because each ion will interact not only with its

389

own polarization charge (i.e., image force) but also with the polarization charges induced by

390

neighboring ions, which screens the image forces. Both membrane materials have similar

391

dielectric constants of ~349,

392

exclusion can be expected to be similar for both membrane types. In addition to screening of

393

interactions by polarization charges induced by neighboring ions, polarization charges are also

394

induced by fixed membrane charges. This screening will be stronger in the case of semi-aromatic

395

nanofiltration membranes than that for fully aromatic nanofiltration membranes because they

396

have greater fixed surface charge (Fig. 3). Considering the Born effects, the structural changes of

397

water in a confined medium affect the free energy of ion transfer from external solution into

398

nanopores of the NF membranes51. Yaroshchuk48 showed that the rejection by dielectric effects

399

is more prominent for smaller pores. Because fully aromatic membranes have narrower pores

400

than semi-aromatic membranes, the contribution of Born effects to rejection by dielectric

401

exclusion would be higher for FA membranes than SA membranes. The screening by fixed

402

membrane charges is expected to be greater in the case of semi-aromatic membranes and it is

403

reasonable to expect that the contribution to the overall rejection by dielectric exclusion would

404

be more prominent in the case of fully aromatic nanofiltration membranes.

50

and the contribution of image force to the overall dielectric

405

This study contributes significantly to understanding the separation mechanisms of two

406

types of commonly used nanofiltration membranes with a view of realizing new potential

407

applications. We first analyze the active layer chemistries of four commercially available

408

nanofiltration membranes with two different active layer chemistries: fully aromatic (1,3-


18

Page 19 of 26


409

benzenediamine (m-phenylenediamine) (MPD) with trimesoyl chloride (TMC) and semi-

410

aromatic (piperazine (PIP) with TMC) to prove that these membranes are representative of the

411

respective categories with no coatings or modification of the active layer. Effective membrane

412

pore size and zeta potential characterization of the four membranes suggests that semi-aromatic

413

membranes have relatively larger pore sizes and that they are more electronegative for all feed

414

compositions tested. Crossflow rejection experiments at low and high ionic strength feed suggest

415

that Donnan (charge) exclusion is significant for semi-aromatic membranes and that these

416

membranes should be preferred in applications requiring partial ion removal (e.g., dairy industry)

417

or charged based separations (e.g., charged organic separations), while fully aromatic

418

membranes should be considered in desalination applications. Qualitative analysis suggests that

419

the contribution of dielectric exclusion to overall rejection by fully aromatic membranes would

420

be more significant than semi-aromatic membranes. This study offers additional insigths into

421

how the separation potentials of polyamide and poly(piperazineamide) active layer chemistries

422

vary even with a very small difference in the effective membrane pore size.

423

Supporting Information. Results and discussion for ATR-FTIR section, Membrane pore size

424

measurement, Additional ion rejection experiments (Fig. S1). Schematic of the crossflow NF

425

filtration system (Fig. S2), SEDE model equations for binary electrolyte at high concentration to

426

calculate the pore size of membrane (Table S1), Ionic diffusivity and Stokes radius of ions used

427

in this study (Table S2).

428

References

429 430 431

1. Geise, G. M.; Lee, H.-S.; Miller, D. J.; Freeman, B. D.; McGrath, J. E.; Paul, D. R., Water purification by membranes: The role of polymer science. Journal of Polymer Science Part B: Polymer Physics 2010, 48, (15), 1685-1718.


19


432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477

Page 20 of 26

2. Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Marinas, B. J.; Mayes, A. M., Science and technology for water purification in the coming decades. Nature 2008, 452, (7185), 301-310. 3. Peterson, R. J., Composite reverse-osmosis and nano-filtration membranes. Journal of Membrane Science 1993, 83, 81-150. 4. Tang, C. Y.; Kwon, Y.-N.; Leckie, J. O., Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: I. FTIR and XPS characterization of polyamide and coating layer chemistry. Desalination 2009, 242, (1– 3), 149-167. 5. Szymczyk, A., Mechanistic modeling of transport in nanofiltration. Encyclopedia of Membrane Science and Technology 2013. 6. Nghiem, L. D.; Schäfer, A. I.; Elimelech, M., Pharmaceutical Retention Mechanisms by Nanofiltration Membranes. Environmental Science & Technology 2005, 39, (19), 7698-7705. 7. Schaep, J.; Vandecasteele, C.; Mohammad, A. W.; Bowen, W. R., Analysis of the Salt Retention of Nanofiltration Membranes Using the Donnan–Steric Partitioning Pore Model. Separation Science and Technology 1999, 34, (15), 3009-3030. 8. Lanteri, Y.; Szymczyk, A.; Fievet, P., Influence of Steric, Electric, and Dielectric Effects on Membrane Potential. Langmuir 2008, 24, (15), 7955-7962. 9. Lanteri, Y.; Fievet, P.; Szymczyk, A., Evaluation of the steric, electric, and dielectric exclusion model on the basis of salt rejection rate and membrane potential measurements. Journal of Colloid and Interface Science 2009, 331, (1), 148-155. 10. Veríssimo, S.; Peinemann, K. V.; Bordado, J., Influence of the diamine structure on the nanofiltration performance, surface morphology and surface charge of the composite polyamide membranes. Journal of Membrane Science 2006, 279, (1–2), 266-275. 11. Li, Y.; Su, Y.; Dong, Y.; Zhao, X.; Jiang, Z.; Zhang, R.; Zhao, J., Separation performance of thin-film composite nanofiltration membrane through interfacial polymerization using different amine monomers. Desalination 2014, 333, (1), 59-65. 12. Ahmad, A. L.; Ooi, B. S.; Mohammad, A. W.; Choudhury, J. P., Composite Nanofiltration Polyamide Membrane: A Study on the Diamine Ratio and Its Performance Evaluation. Industrial & Engineering Chemistry Research 2004, 43, (25), 8074-8082. 13. Vyas, B. B.; Ray, P., Preparation of nanofiltration membranes and relating surface chemistry with potential and topography: Application in separation and desalting of amino acids. Desalination 2015, 362, 104-116. 14. Petersen, R. J.; Cadotte, J. E., Thin film composite reverse osmosis membranes. Handbook of industrial membrane technology 1990, 307-348. 15. Tang, C. Y.; Kwon, Y.-N.; Leckie, J. O., Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: II. Membrane physiochemical properties and their dependence on polyamide and coating layers. Desalination 2009, 242, (1–3), 168-182. 16. Coronell, O.; Mariñas, B. J.; Zhang, X.; Cahill, D. G., Quantification of Functional Groups and Modeling of Their Ionization Behavior in the Active Layer of FT30 Reverse Osmosis Membrane. Environmental Science & Technology 2008, 42, (14), 5260-5266. 17. Coronell, O.; González, M. I.; Mariñas, B. J.; Cahill, D. G., Ionization Behavior, Stoichiometry of Association, and Accessibility of Functional Groups in the Active Layers of Reverse Osmosis and Nanofiltration Membranes. Environmental Science & Technology 2010, 44, (17), 6808-6814.


20

Page 21 of 26

478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523


18. Coronell, O.; Mariñas, B. J.; Cahill, D. G., Accessibility and Ion Exchange Stoichiometry of Ionized Carboxylic Groups in the Active Layer of FT30 Reverse Osmosis Membrane. Environmental Science & Technology 2009, 43, (13), 5042-5048. 19. Freger, V., Nanoscale heterogeneity of polyamide membranes formed by interfacial polymerization. Langmuir 2003, 19, (11), 4791-4797. 20. Coronell, O.; Mariñas, B. J.; Cahill, D. G., Depth Heterogeneity of Fully Aromatic Polyamide Active Layers in Reverse Osmosis and Nanofiltration Membranes. Environmental Science & Technology 2011, 45, (10), 4513-4520. 21. Krieg, H. M.; Modise, S. J.; Keizer, K.; Neomagus, H. W. J. P., Salt rejection in nanofiltration for single and binary salt mixtures in view of sulphate removal. Desalination 2005, 171, (2), 205-215. 22. Singer, P. C.; Stumm, W., Acidic Mine Drainage: The Rate-Determining Step. Science 1970, 167, (3921), 1121-1123. 23. Johnson, D. B.; Hallberg, K. B., Acid mine drainage remediation options: a review. Science of The Total Environment 2005, 338, (1–2), 3-14. 24. Artuğ, G.; Roosmasari, I.; Richau, K.; Hapke, J., A Comprehensive Characterization of Commercial Nanofiltration Membranes. Separation Science and Technology 2007, 42, (13), 2947-2986. 25. Wadekar, S. S.; Vidic, R. D., Optimization of abandoned mine drainage treatment by commercially available nanofiltration membranes. In Proceedings of the 2017 American Water Works Association (AWWA)-American Membrane Technology Association (AMTA) Membrane Technology Conference and Exposition, Long Beach, California, 2017. 26. Al-Zoubi, H.; Rieger, A.; Steinberger, P.; Pelz, W.; Haseneder, R.; Härtel, G., Optimization study for treatment of acid mine drainage using membrane technology. Separation Science and Technology 2010, 45, (14), 2004-2016. 27. Escoda, A.; Lanteri, Y.; Fievet, P.; Déon, S.; Szymczyk, A., Determining the Dielectric Constant inside Pores of Nanofiltration Membranes from Membrane Potential Measurements. Langmuir 2010, 26, (18), 14628-14635. 28. Lokare, O. R.; Tavakkoli, S.; Wadekar, S.; Khanna, V.; Vidic, R. D., Fouling in direct contact membrane distillation of produced water from unconventional gas extraction. Journal of Membrane Science 2017, 524, 493-501. 29. Saha, N. K.; Joshi, S. V., Performance evaluation of thin film composite polyamide nanofiltration membrane with variation in monomer type. Journal of Membrane Science 2009, 342, (1–2), 60-69. 30. Skrovanek, D. J.; Howe, S. E.; Painter, P. C.; Coleman, M. M., Hydrogen bonding in polymers: infrared temperature studies of an amorphous polyamide. Macromolecules 1985, 18, (9), 1676-1683. 31. Socrates, G., Infrared characteristic Group Frequencies. Wiley-Interscience 1994. 32. Oh, N.-W., Jegal, J. and Lee, K.-H., Preparation and characterization of nanofiltration composite membranes using polyacrylonitrile (PAN). I. preparation and modification of PAN supports. J. Appl. Polym. Sci. 2001, 80, 1854–1862. 33. Lo, R.; Bhattacharya, A.; Ganguly, B., Probing the selective salt rejection behavior of thin film composite membranes: A DFT study. Journal of Membrane Science 2013, 436, 90-96. 34. Freger, V., Swelling and Morphology of the Skin Layer of Polyamide Composite Membranes: An Atomic Force Microscopy Study. Environmental Science & Technology 2004, 38, (11), 3168-3175.


21


524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568

Page 22 of 26

35. Hilal, N.; Al-Zoubi, H.; Darwish, N. A.; Mohammad, A. W., Characterisation of nanofiltration membranes using atomic force microscopy. Desalination 2005, 177, (1–3), 187199. 36. Nghiem, L. D.; Schäfer, A. I.; Elimelech, M., Removal of Natural Hormones by Nanofiltration Membranes: Measurement, Modeling, and Mechanisms. Environmental Science & Technology 2004, 38, (6), 1888-1896. 37. Bowen, W. R.; Mohammad, A. W., Characterization and Prediction of Nanofiltration Membrane Performance—A General Assessment. Chemical Engineering Research and Design 1998, 76, (8), 885-893. 38. Luxbacher, T., The ZETA Guide: Principles of the Streaming Potential Technique. Anton Paar 2014. 39. Tu, K. L.; Nghiem, L. D.; Chivas, A. R., Coupling effects of feed solution pH and ionic strength on the rejection of boron by NF/RO membranes. Chemical Engineering Journal 2011, 168, (2), 700-706. 40. Childress, A. E.; Elimelech, M., Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes. Journal of Membrane Science 1996, 119, (2), 253-268. 41. Childress, A. E.; Elimelech, M., Relating Nanofiltration Membrane Performance to Membrane Charge (Electrokinetic) Characteristics. Environmental Science & Technology 2000, 34, (17), 3710-3716. 42. Teixeira, M. R.; Rosa, M. J.; Nyström, M., The role of membrane charge on nanofiltration performance. Journal of Membrane Science 2005, 265, (1–2), 160-166. 43. Tay, J.-H.; Liu, J.; Delai Sun, D., Effect of solution physico-chemistry on the charge property of nanofiltration membranes. Water Research 2002, 36, (3), 585-598. 44. Hilal, N.; Al ‐ Zoubi, H.; Darwish, N. A.; Mohammad, A. W., Nanofiltration of Magnesium Chloride, Sodium Carbonate, and Calcium Sulphate in Salt Solutions. Separation Science and Technology 2005, 40, (16), 3299-3321. 45. Santafé-Moros, A.; Gozálvez-Zafrilla, J.; Lora-García, J., Performance of commercial nanofiltration membranes in the removal of nitrate ions. Desalination 2005, 185, (1), 281-287. 46. Wang, D.-X.; Su, M.; Yu, Z.-Y.; Wang, X.-L.; Ando, M.; Shintani, T., Separation performance of a nanofiltration membrane influenced by species and concentration of ions. Desalination 2005, 175, (2), 219-225. 47. Peeters, J. M. M.; Boom, J. P.; Mulder, M. H. V.; Strathmann, H., Retention measurements of nanofiltration membranes with electrolyte solutions. Journal of Membrane Science 1998, 145, (2), 199-209. 48. Yaroshchuk, A. E., Dielectric exclusion of ions from membranes. Advances in colloid and interface science 2000, 85, (2), 193-230. 49. Weast, R. C.; Astle, M. J.; Beyer, W. H., CRC handbook of chemistry and physics. CRC press Boca Raton, FL: 1988; Vol. 69. 50. Szymczyk, A.; Fievet, P., Investigating transport properties of nanofiltration membranes by means of a steric, electric and dielectric exclusion model. Journal of Membrane Science 2005, 252, (1–2), 77-88. 51. Parsegian, A., Energy of an ion crossing a low dielectric membrane: solutions to four relevant electrostatic problems. Nature 1969, 221, (5183), 844-846.

569


22

Page 23 of 26


570 571 572 573 574 575 576

Figures:

577

578 579

Fig. 1. FTIR spectra of NF270, TS40, TS80 and NF90 nanofiltration membranes

580

581 582

Fig. 2. Membrane potential as a function of chloride concentration in solution


23


583 584 585 586 587

Page 24 of 26

Fig. 3. (a) Isoelectric Point (IEP) determination (with 1 mM KCl) (b) Zeta potentials with different solution composition (1: 1 mM Na2SO4; 2: 1 mM Na2SO4 + 1mM MgCl2; 3: 1 mM Na2SO4 + 1mM MgCl2 + 1mM CaCl2, pH = 5.6±0.1) for NF90, TS80, NF270 and TS40 membranes

588

589 590

Fig. 4. Pure water permeability as a function of effective pore radius


24

Page 25 of 26


591

592 593 594 595

Fig. 5. Rejection of ionic species with feed solutions: (a) 96 mg/l sulfate + 24 mg/l magnesium and (b) 96 mg/l sulfate + 40 mg/l calcium, (c) 650 mg/L sulfate + 1000 mg/L magnesium and (d) 650 mg/L sulfate + 1000 mg/L calcium

596 597 598 599 600 601 602 603 604 605 606 607 608 609 610


25


611

Page 26 of 26

TOC/Abstract Art

612

613


26

Influence of Active Layer on Separation Potentials of Nanofiltration

Recommend Documents