Fabrication and characterization of γ-Al2O3–clay composite ultrafiltration membrane for the separation of electrolytes from its aqueous solution

Journal of Membrane Science 340 (2009) 181–191

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Fabrication and characterization of ␥-Al2 O3 –clay composite ultrafiltration membrane for the separation of electrolytes from its aqueous solution Abhijit Majhi, P. Monash, G. Pugazhenthi ∗ Department of Chemical Engineering, Indian Institute of Technology Guwahati, Guwahati 781039, Assam, India

a r t i c l e

i n f o

Article history: Received 2 December 2008 Received in revised form 23 April 2009 Accepted 19 May 2009 Available online 27 May 2009 Keywords: ␥-Al2 O3 Clay Membrane MgCl2 AlCl3

a b s t r a c t In this paper, we have reported the preparation of low cost ␥-Al2 O3 membrane on a macroporous clay support by dip-coating method. For the synthesis of ␥-Al2 O3 top layer on the support, a stable boehmite sol is prepared using aluminium chloride salt as a starting material by sol–gel route. The structural properties of the composite membrane as well as ␥-Al2 O3 powder is carried out using scanning electron microscopy (SEM), X-ray diffraction (XRD), nitrogen adsorption–desorption isotherm data, Fourier transform infrared analysis (FTIR) and dynamic light scattering (DLS) analysis. The mean particle size of the boehmite sol used for coating is found to be 30.9 nm. The pore size distribution of the ␥-Al2 O3 –clay composite membrane is found to be in the range of 5.4–13.6 nm. Separation performance of the membrane in terms of flux and rejection of single salts solution such as MgCl2 and AlCl3 as a function of pH, salt concentration and applied pressure is also studied. The rejection and flux behavior are found to be strongly dependent on electrostatic interaction between the charged molecules and ␥-Al2 O3 –clay composite membrane. The intrinsic rejection has been determined by calculating the concentration at membrane surface (Cm ) using Speigler–Kedem model. It is found that the observed rejection shows anomalous trend with increase in applied pressure and the intrinsic rejection increases with increase in applied pressure, a trend typical of the separation of electrolyte through charged membranes. At acidic pH, both the salt solution shows higher rejection. With increase in the salt concentration, observed rejection of salt decreases due to the enhanced concentration polarization. The maximum rejection of MgCl2 and AlCl3 is found to be 72% and 88%, respectively for salt concentration of 3000 ppm. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Membrane technology has gained a huge importance in the last 30 years competing with long established technologies such as distillation, absorption, adsorption and extraction. The reason is the possibility of recovering valuable products from effluent and minimizing environmental problems, which are usually claimed to be caused by the chemical industry. Membrane-based separations are energy efficient and cost effective. So far, application of the organic membrane is more developed as compared to inorganic membrane. In recent years, a great deal of research has been directed to the development of new types of composite inorganic membranes which includes ␣-, ␥-Al2 O3 [1–3], zeolites [4,5], carbon [6–8] and dense metals [9,10]. Inorganic membranes are chemically more stable and can tolerate higher temperatures than polymeric-based membranes which make them a promising candidate for industrial applications. Sol–gel derived ␥-Al2 O3 membranes have emerged as perhaps the most intensively studied inorganic membrane sys-

∗ Corresponding author. Tel.: +91 361 2582264; fax: +91 361 2582291. E-mail address: [email protected] (G. Pugazhenthi). 0376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2009.05.030

tem because of their unique surface charge characteristic. ␥-Al2 O3 membranes may acquire either a positive or a negative charge due to the amphoteric behavior of their surface sites (hydroxyl groups). This allows controlling the sign and charge of the membrane through pH of the solution. This was demonstrated for mesoporous ␥-Al2 O3 membranes with a MWCO of 900 Da by Schaep et al. [11]. Alami-Younssi et al. [12] have prepared the ␥-Al2 O3 membrane using expensive aluminium alkoxide as raw material and investigated the rejection mechanism of mineral salts. They found the rejection mechanism of ionic species through gamma alumina (which was positively charged over the studied range of pH) exclusively depends on the size and charge of the filtered ionic species. Larbot et al. [13] studied the influence of the final calcination temperature on the crystalline structure of the alumina membrane prepared from boehmite sol. Majority of the reported ␥-Al2 O3 membranes were derived from the expensive aluminium alkoxides [3,11–13]. Unfortunately, these membranes are too expensive to consider economic applications in industry. So, the focus is to search for alternative membrane materials that can reduce the synthesis cost of the membrane. The fabrication of inorganic membranes using naturally occurring clay mineral is one of the approaches that have received only limited attention in the literature. Clay

182

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

minerals are well-known class of natural inorganic materials having good adsorption, rheological and thermal properties [14–17]. The development of clay-based inorganic membranes could lead to an important new technological application that would add economic value to the vast natural deposits of clay minerals that are located throughout the world, many of which are currently under utilized. Several authors have focused their research to prepare supports of natural clays to support micro or ultrafiltration top layer [18,19]. Potdar et al. [20] fabricated the ceramic–analcime zeolite composite membrane for the separation of surfactant by hydrothermal crystallization of zeolite over clay supports. Shukla and Kumar [21] prepared modified zeolite clay composite membrane for the separation of chloride salts (FeCl3 and AlCl3 ) from its aqueous solution. They showed that the observed rejection of FeCl3 was higher than AlCl3 which depends on the interaction between salt and the membrane. In our earlier work, we have reported the separation of chromium (VI) from its aqueous solution by clay supported carbon membrane [22] and cross-linked PMMA–EGDM clay composite anion exchange membrane [23]. We have obtained a maximum rejection of 84% and 90% for carbon membrane and PMMA composite membrane, respectively. Recently, Workneh and Shukla [24] have prepared sodalite octahydrate–zeolite clay composite ultrafiltration membrane for separation of sodium dodecyl sulfate and obtained 10–45% rejection. Khider et al. [25] studied purification of water effluent from a milk factory by ultrafiltration using Algerian clay support. The obtained results have shown a good retention for lactose and total proteins giving rise to pure water. Seffaj et al. [26] have used macroporous support made of Moroccan clay to support an intermediate ZrO2 layer and an active TiO2 /ZnAl2 O4 ultrafiltration layer for the separation of mineral salts. Liangxiong et al. [27] have successfully studied the separation of ionic solutes from oil field produced water using bentonite membrane. A performance study of ceramic microfiltration membrane from Tunisian clay for the treatment of cuttlefish effluent was done by Khemakhem et al. [28]. They have also studied the separation of dextran solution using membrane prepared from Tunisian natural illite clay [29]. The above extensive literature review reveals that utilization of clay materials is now of growing interest. In this paper, a low cost ␥-Al2 O3 composite membrane on a clay support (developed using locally available cheaper clay materials) has been prepared. The top layer (␥-Al2 O3 ) is fabricated using boehmite sol by dip-coating technique. In order to further reduce the fabrication cost of the membrane, aluminium chloride salt is used as a starting material for the preparation of boehmite sol. Boehmite sol characteristic is investigated by means of TGA, XRD, FTIR, N2 adsorption–desorption isotherm and DLS analysis. The structural characteristic of the composite membrane has also been determined by SEM, chemical stability test and pure water flux. Finally, separation characteristic of membrane has been studied using aqueous solution of MgCl2 and AlCl3 . 2. Theory The transport of the solute through a membrane by pressure driven membrane processes can be described by irreversible thermodynamics, where the membrane is considered as a black box. The solute flux through the membrane can be described as the sum of a convective (due to applied pressure gradient across a membrane) and diffusive (due to concentration difference) flux. In ultrafiltration (UF) and nanofiltration (NF), the rejection factor is defined as

where Cp is the concentration of solute in the permeate, Cb , the bulk concentration of retentate, Cm , the concentration on the surface of the membrane, Robs , observed rejection and Rint , intrinsic (real) rejection. In the above, Rint is an inherent property of the membrane while Robs depends strongly on the operating conditions. Therefore, it is desirable to report separation performance of a membrane in terms of Rint even though the determination of Cm is difficult. The Cm can be determined by either of the following two techniques. (i) Direct measurement of Cm through interoferometric [30,31] and (ii) optical shadow measurements [32]. Apart from the above two techniques, Cm can also be determined by solving transport equations in the polarization layer. Accuracy of the estimated Cm depends upon the validity of the hydrodynamic model used. In the second technique, the following equations are used for determination of Cm . The membrane surface concentration is calculated using the osmotic pressure model [31] Jv = Lp (P −  )

(3)

where Jv is the permeate flux, Lp , the pure water permeability, P, the applied pressure difference, , the membrane reflection coefficient and , the osmotic pressure difference. The osmotic pressure difference is calculated using the van’t Hoff equation for electrolytes  = T C

(4)

where  is the number of moles of ions given by each mole of the electrolyte in the solution,  is universal gas constant, T is temperature (Kelvin units) and C (=Cm − Cp ) is the difference in the concentration of the solute at the membrane surface (upstream side) and in the permeate. The reflection coefficient is related to the intrinsic rejection of the membrane through the equation given by Spiegler and Kedem [33] Rint =

(1 − F) 1 − F

(5)

where F is given by F=

exp{−(1 − )Jv } Pm

(6)

In this equation, Pm is the solute permeability of the membrane. According to Eq. (5), the rejection increases with increasing the water flux (Jv ). The reflection coefficient,  is a characteristic of the convective transport of the solute. If  = 1 means that the convective solute transport does not take place at all. This is the case for ideal reverse osmosis (RO) membranes where the membranes have no pores available for the convective transport. For the UF and NF membranes having pores, the  value will be less than 1 if the solutes are small enough to enter the membrane pores. The parameter Cm ,  and Pm are determined using Eqs. (3)–(6) following the iterative technique given by Ghose et al. [34]. The technique involves assuming a  value and using Eqs. (3) and (4) to calculate the values of Cm . The Pm values for various experimental data are then calculated using Eqs. (5) and (6) and the standard deviation in the Pm values is calculated and minimized by adjusting the  value. The  value obtained is used to calculate the surface concentration (Cm ) and the convergence criterion is taken as a change of less than 2% in the value of surface concentration of the membrane (Cm ).

Robs = 1 −

Cp Cb

(1)

3. Experimental

Rint = 1 −

Cp Cm

(2)

The preparation method of ␥-Al2 O3 –clay composite membrane was discussed in the following sections.

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191 Table 1 Composition of clay mixture. Clay material

Compositions (wt%)

Kaolin Ball clay Feldspar Quartz Calcium carbonate Pyrophyllite

14.45 17.58 05.60 26.59 17.14 14.73

3.1. Preparation of clay supports The clay powders were collected locally and the composition used for the fabrication of clay disc is given in Table 1. Fig. 1 depicts the schematic representation of the clay support preparation. First, Kaolin (Al2 Si2 O5 (OH)4 ), ball clay (3SiO2 Al2 O3 ), feldspar ((Na,Ca)(AlSi3 O8 ), quartz (SiO2 ), pyrophyllite (Al2 (Si2 O5 )2 (OH)2 ), calcium carbonate (CaCO3 ) were mixed in a ball mill with 4 ml of 2 wt% aqueous polyvinyl alcohol (PVA, M.W. 72,000). The resulting powder was then sieved using 40 mesh and a requisite amount of the mixture was compacted (uniaxial pressing) at a pressure of 50 MPa with the help of stainless steel die, designed for singleended pressing. It resulted in circular green disc specimens of 63 mm diameter and 4.5 mm thickness. The green supports thus prepared were first dried at 100 ◦ C for 24 h, at 250 ◦ C for 24 h in an oven and finally sintered at 950 ◦ C for 6 h with a heating rate of 2 ◦ C/min in a muffle furnace. The rate of heating in a muffle furnace was kept low to minimize the possibility of bending due to radiation shock waves. During sintering, the clay discs were placed vertically over the grooves of an insulation brick to ensure uniform

183

sintering. After sintering, the support dimension was measured to calculate the volume shrinkage. The sintered supports were finally polished at both sides using silicon carbide abrasive paper (no. C220) to get clay supports with uniform smooth surface of diameter 42 mm and thickness of 4 mm. To remove the loose particles created during polishing, clay discs were cleaned with Millipore water in an ultrasonic bath [make: Elma (India); model: T460] for 15 min. 3.2. Synthesis of -Al2 O3 –clay composite membrane Boehmite sol was synthesized from aluminium chloride, AlCl3 ·6H2 O (99.5%, Merck, India) by controlled precipitation followed by peptization using dilute nitric acid at pH 3–3.5. Aluminium chloride solution was prepared by dissolving 4 g of the salt in Millipore water. After which, an aqueous ammonia solution (30 wt%) was added slowly to the aluminium chloride solution at room temperature under constant stirring until the pH of the precipitate reaches 8.0. The precipitate was filtered and washed repeatedly with Millipore water to remove the chlorides. After that, it was transferred into a beaker and diluted with Millipore water. A stable sol having pH of 3.5 was obtained by peptizing it by the addition of 0.3 wt% HNO3 . Additionally, 0.5 wt% PVA was mixed with the sol and refluxed at 85 ◦ C for 24 h to get a stable and clear or slightly translucent dipping sol. The organic binder allows the adjustment of the sol viscosity (as a thickener) as well as it protects the thin alumina layer from cracking during calcination [35]. PVA generally burns off during calcination without leaving ash. Finally, the ␥-Al2 O3 –clay composite membrane was prepared by dipping the support into the above-prepared sol for 1 min. For this, one side of the clay support was wrapped with polytetrafluoroethylene (PTFE) sealing wrap and the coating was applied only on the other side. Before dipping into boehmite sol, the clay supports were heated to 200 ◦ C and then cooled in a desiccator. After dipping, the dip-coated clay support was dried overnight at room temperature followed by heating at 50 ◦ C for 24 h to remove maximum moisture. Thereafter, the calcination was done at a rate of 0.5 ◦ C/min to 600 ◦ C, where they were kept for 3 h. 3.3. Characterization

Fig. 1. Flow chart for the preparation of the membrane support discs.

3.3.1. Characterization of boehmite and -Al2 O3 particles For characterization purpose, ␥-Al2 O3 powder was prepared by pouring the boehmite sol (without adding PVA) on a glass plate and then dried overnight at 50 ◦ C. Then, this as-dried sample (boehmite) was calcined to ␥-Al2 O3 at the same conditions as followed for ␥-Al2 O3 –clay composite membrane. To identify phases and their crystallinity of as-synthesized (boehmite) and calcined sample (␥Al2 O3 ), X-ray powder diffraction (XRD) pattern was recorded in a Bruker AXS instrument using Ni filtered Cu K␣ radiation ( = 1.5406 Å) operating at 40 kV and 40 mA. Diffraction intensities were measured by scanning from 10◦ to 80◦ (2) with a step size of 0.05◦ /s. The IR spectrum of the ␥-Al2 O3 powder was obtained on a Nicolet Impact-410 Fourier transform infrared spectrometer (Nicolet Impact-410). Thermogravimetric analyses of boehmite was conducted in a TGA instrument of METTLER TOLEDO with model No. TGA 851® in flowing air atmosphere at a heating rate of 10 ◦ C min−1 from 25 ◦ C to 900 ◦ C. The transformation temperature of boehmite to ␥-Al2 O3 was determined from the DTG curve. The BET (Brunauer–Emmet–Teller) specific surface area, pore volume and pore size distribution of the ␥-Al2 O3 powder was determined by N2 adsorption–desorption isotherm at 77 K measured in a surface area analyzer (make: Beckman-Coulter; model: SA 3100). The pore size distribution was determined by a BJH (Barett–Joyner–Halenda) model from the adsorption branch of the nitrogen isotherms. The pore volume was calculated from the amount of nitrogen adsorbed at a relative pressure, P/P0 of 0.99,

184

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

Fig. 2. Experimental setup for filtration test.

where it was assumed that all the pores were filled. The particle size distribution of dipping boehmite sol was examined by dynamic light scattering equipment (make: HORIBA; model LB-550V) with a He–Ne laser source.

mination of pore size distribution of the membrane. This method essentially measures the differential pressure needed to overcome the resistance offered to the displacement of a wetting liquid by a non-wetting liquid. The differential is given by the Laplace equation

3.3.2. Characterization of clay support and -Al2 O3 –clay composite membrane 3.3.2.1. Chemical stability test. Static chemical stability test of the clay support was carried out in terms of mass loss after keeping it separately in the acidic (pH 1.5) and alkali solution (pH 13). For this, clay supports were placed in H2 SO4 and NaOH solution for different times (1–5 days). After each day, the supports were taken out from the solution (acid and alkali) washed with Millipore water and dried at 100 ◦ C. The mass of the support before and after the test was noted down and mass losses of samples were adopted to characterize the chemical stability [36].

P =

3.3.2.2. SEM study. Scanning electron microscopy (make: LEO; model: 1430VP) was used to determine the surface morphology of clay support as well as ␥-Al2 O3 –clay composite membrane. The top layer thickness of the alumina thin films was determined from SEM image of the cross-section of the composite membrane. 3.3.2.3. Pore size distribution. The membrane pore size is an important characteristic to evaluate the influence of the solute size on its retention. The bubble-point technique was adopted for the deter-

2 cos  r

(7)

where r is the radius of capillary shaped pore,  is the interfacial tension between the two fluids and  is the contact angle between the fluids and the ␥-Al2 O3 . The bubble point setup used for this experiment is shown in Fig. 2. An unstirred batch cell made of stainless steel 316 having a capacity of 300 ml was used for this purpose. It consists of two parts, the cylindrical top part and a base plate with a provision to keep membrane. The membrane was placed in a SS casing and sealed with epoxy resin. Then it was placed in the membrane housing provided on the base plate. The fluid was filled in the tubular section from the top and the cell was pressurized with compressed nitrogen. The contact angle and surface tension of water–butanol system was taken from the literature [37]. 3.3.3. Determination of water flux and hydraulic permeability The setup used for this experiment was same as the one used for the bubble point experiment. Pure water was passed through the fresh membrane at higher pressure till a constant flux was obtained and thereafter, we measured the pure water flux at different pressures. For water flux determinations at each pressure,

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

185

Fig. 3. SEM images of top surface of (a) clay support, (c) ␥-Al2 O3 –clay composite membrane and cross-section of (b) clay support, and (d) ␥-Al2 O3 –clay composite membrane.

50 ml of permeate was allowed to pass and the time required for collections of the next 25 ml permeate was noted down. This was used to determine the water flux using the following equation: Jw =

Q A T

(8)

where Jw is pure water flux (m s−1 ), Q is volume of water permeated (m3 ), A is effective membrane area (m2 ) and T is sampling time (s). The pure water flux through a membrane can be described by the Darcy’s law, which states that volumetric flux is directly proportional to the applied pressure gradient Jw = Lp Papplied

(9) (m s−1

kPa−1 )

and P is applied where Lp is hydraulic permeability pressure (kPa). The reported values of water flux were average of three readings and the data was regressed by linear curve to obtain the hydraulic permeability.

by conductivity measurements (Model: VSi-04, India). After each experiment, the membrane was thoroughly washed using Millipore water followed by flushing with pure water at higher pressure to regain its original water permeability. The retention and permeate flux of electrolyte solutions was investigated as a function of working parameters such as pH, salt concentration, applied pressure and nature of cation present in the solution. The studied pH in this paper is in the range of 3–8 at constant pressure of 207 kPa. The pH of the salt solution was adjusted by adding dilute NaOH and HCl solution. Experiments were also carried out at five different applied pressures for a fixed concentration of salt solution (3000 ppm) and at different electrolyte concentrations for a fixed pressure of 207 kPa to study the influence of pressure and feed concentration on the permeate flux and rejection, respectively. The pressures used were in the ranges between 138 kPa and 550 kPa and the concentrations were in the ranges of 1000–5000 ppm.

3.4. Filtration experiments 4. Results and discussion The separation capability of the above-prepared membrane was studied by performing filtration experiments of single salt solution (MgCl2 and AlCl3 ) using dead-end filtration setup as shown in Fig. 2. All the experiments were carried out at a constant temperature of 25 ◦ C. All the salts used were of analytical grade (Merck, India). The electrolyte solutions were prepared using Millipore water. For each run, the cell was filled with 250 ml of electrolyte solution and the first 45 ml of permeate passed through the membrane was discarded. Permeate flux was calculated by measuring the time interval corresponding to next 25 ml permeate volume. The concentrations of the salts in the feed and permeate were determined

4.1. Characterization of clay support A uniaxial pressing method has been used to obtain a more compact clay support with higher mechanical strength. The pore formation takes place in two steps. In the first step, the pore formation is due to the evolution of gases. During the sintering of circular green clay discs, phase transformation and dehydroxylation of the raw materials occurs. The CaCO3 present in the mixture forms CaO and the exiting CO2 makes the pores on the surface of the clay support. Second step is the pore rearrangement due to phase

186

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

Fig. 5. XRD pattern (a) boehmite sol and (b) ␥-Al2 O3 . b, boehmite; ␥, ␥Al2 O3 .

Fig. 4. Chemical stability of the clay support.

transformation. The phase transformation of kaolin to metakaolinite and mullite occurs in the studied sintering temperature which can be confirmed by XRD analysis [38]. The releasing of the gaseous products also changes the volume of clay disc. Here, we have used PVA as a binder in the clay mixture. The degradation of the organic binder occurs in two steps; elimination of the side-groups at lower temperature between 200 ◦ C and 350 ◦ C and the decomposition of the main chain of PVA between 450 ◦ C and 650 ◦ C. The PVA chain enhances the thermal stability of the clay support as well as it acts as a pore-forming agent. Volume shrinkage of the sintered clay support is measured and found to be ∼4.5%, which is very less compared to that of commercially available supports. Porosity of the support is calculated using Archimedes’ principle and is found to be 46%. Fig. 3(a and b) shows the top surface and cross-section of clay support where the macroporous structure of the support is clearly visible. To estimate the average pore size of the support, the individual pore diameters are measured for about 500 pores using ImageJ free software (provided by National Institute of Health (NIH), weblink: http://rsbweb.nih.gov/ij/download.html) for different pores visible in the SEM images. The calculated average pore diameter of the support is 1.01 ␮m. Fig. 4 depicts the test result of chemical stability of the support in terms of weight loss with time. It is clearly visible that the resistance to alkaline is better than that to acid. The poor acid resistance of the support is mainly due to the nature of clay materials as well as the existence of the alkaline metal or alkaline earth metal oxides in the starting material. However, clay support shows excellent chemical stability with weight loss less than 1.5 wt%, which is comparable with ␣-alumina support having weight loss of 1.25 wt% in 20 wt% H2 SO4 solution.

The bands at 744 cm−1 , 628 cm−1 and 1075 cm−1 can be assigned to the defined characteristics of boehmite [39]. It is observed an intense band at 1075 cm−1 and a shoulder at 1171 cm−1 , which are assigned to the symmetric and asymmetric bending modes of Al–O–H, respectively [40]. The intense band at 1634 cm−1 and broad band centered at 3414 cm−1 are assigned to the bending and stretching modes of adsorbed water molecule [40]. The OH torsional mode expected around 750 cm−1 could not be observed because it overlaps with the Al–O stretching vibrations. The weak band at 2098 cm−1 observed for boehmite is assigned to a combination band, which disappears in the spectra of the calcined samples (see Fig. 6b). In aluminium oxides, aluminium can have different types of coordination with oxygen. If the coordination is octahedral (AlO6 ), the Al–O stretching and bending modes are expected in the region 500–750 cm−1 and 330–450 cm−1 , respectively. However, a tetrahedral coordination (AlO4 ) is expected to give stretching modes in the narrow range of 750–870 cm−1 and bending modes between 250 cm−1 and 320 cm−1 . For boehmite (Fig. 6a), the bands observed at 744 cm−1 and 628 cm−1 are assigned to the stretching modes of AlO6 [41]. Fig. 6(b) depicts the FTIR spectra of the ␥-Al2 O3 . Upon controlled calcinations at 600 ◦ C, the IR bands observed at ∼1075 cm−1 and 2098 cm−1 for boehmite sample disappears suggesting the complete dehydration or dehydroxylation of boehmite and its conversion to ␥-Al2 O3 . Broad bands observed around 3445 cm−1 and 1644 cm−1 are due to the water absorbed during pelletization. The peak at 592 cm−1 is assigned to AlO6 and the shoulder observed at 867 cm−1 is assigned to AlO4 .

4.2. Characterization of boehmite and -Al2 O3 particles The XRD pattern of boehmite and ␥-Al2 O3 particle is depicted in Fig. 5. All the reflections of boehmite (Fig. 5a) matches well with the JCPDS file no. 21-1307, which corresponds to boehmite. The XRD patterns of calcined samples matches well with ␥-Al2 O3 phase, which is confirmed by the JCPDS file of 10-0425. The strongest (4 4 0) and (4 0 0) diffraction pattern of ␥-Al2 O3 confirms the presence of ␥-Al2 O3 nano-crystallites in the framework walls. The crystallite size of the boehmite and ␥-Al2 O3 is calculated using well known Scherer equation from the broadening of the d (0 2 0)-peak and d (4 0 0)-peak, respectively, which is found to be 2.4 nm and 3.3 nm, respectively. Fig. 6(a) shows the FTIR spectrum of boehmite.

Fig. 6. FTIR spectra of (a) boehmite sol and (b) ␥-Al2 O3 .

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

187

Fig. 7. TG and DTG profile of boehmite sol.

From these results, we conclude that ␥-Al2 O3 (calcined samples) contain tetrahedral and octahedral Al–O, where as boehmite is purely octahedral in nature [42]. The thermogravimetric (TG) and differential thermogravimetric (DTG) curves of boehmite is depicted in Fig. 7. The boehmite appears to undergo three stages of decomposition during heating. The first stage of decomposition (<120 ◦ C) is the liberation of physically adhered water present in the pores. The second stage of weight loss between 120 ◦ C and 320 ◦ C can be assigned to the removal of the crystal water in the sample. The final step (>320 ◦ C) corresponds to dehydroxylation of boehmite to ␥-Al2 O3 [43]. As can be seen from DTG curve, an endothermic peak at around 100 ◦ C corresponds to the removal of water molecules adsorbed on the gel. The second endothermic peak at about 278 ◦ C corresponds to the loss of crystal water from this sample. The third endothermic peak around 390 ◦ C is ascribed to boehmite decomposition to produce ␥-Al2 O3 involving elimination of OH groups [43]. N2 adsorption–desorption isotherm of ␥-Al2 O3 powder is shown in Fig. 8. The ␥-Al2 O3 particles give an isotherm of type IV with an H2 hysteresis loop according to the IUPAC classification. This kind of hysteresis loop is an indication of a network of interconnected pores with narrower parts. The pore size distribution of ␥-Al2 O3 was computed from the adsorption isotherm by the BJH method [44] as shown in Fig. 9. This plot confirms that the pore size of

Fig. 8. N2 adsorption–desorption isotherm of ␥-Al2 O3 particles.

Fig. 9. BJH pore size distribution of ␥-Al2 O3 particles.

␥-Al2 O3 is in the mesoporous range. It also shows a unimodal narrow pore size distribution with pores in the range of 3.6–10 nm in diameter. Moreover, 85% of the pores are smaller than 6.0 nm for ␥Al2 O3 . BET surface area and pore volume of ␥-Al2 O3 are calculated to be 216.89 m2 /g and 0.30 ml/g, respectively, which are strongly higher than the values reported by Kandri et al. [45] and Buelna and Lin [46]. Fig. 10 shows DLS particle size analysis of boehmite sol. It is well established that the dipping sols with large particle size will not cover the surface uniformly leaving patches of exposed untreated surface. Our work reports the narrow particle size distribution of the boehmite sol with a mean particle size of 31 nm, which is suitable for coating. 4.3. Characterization of -Al2 O3 –clay composite membrane SEM photographs of the top and cross-sectional view of ␥-Al2 O3 composite membrane is presented in Fig. 3(c and d). After coating, reduction in the pore sizes is clearly visible when compared with pore sizes of clay support. The top layer (␥-Al2 O3 ) has a finer structure because of smaller particle size of the sol. Fig. 3(d) also clearly demonstrates a uniform interphase between the ␥-Al2 O3 layer and the macroporous clay support, suggesting that the ␥-Al2 O3 layer

Fig. 10. Particle size distribution of boehmite sol.

188

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

Fig. 11. Flux–pressure curve for pore size distribution of ␥-Al2 O3 –Clay composite membrane (a, butanol flux; b, butanol flux when membrane pores are filled with water).

has good adhesion to the support. Furthermore, it is found that no infiltration of sol particles occurred during the dip-coating process. The thickness of the coated ␥-Al2 O3 layer is measured using the cross-section of SEM image of the composite membrane and is found to be ∼2.6 ␮m. Two distinct layers is also clearly seen from the SEM image. The pore size distribution of the composite membrane is determined employing bubble point technique. Here, water was selected as a wetting liquid and butanol used as a non-wetting liquid. First, butanol flux was measured through a dry membrane as a function of pressure and a straight line was obtained as shown in Fig. 11. Then the membrane was wetted with water and again butanol flow rate was determined as a function of applied pressure. Up to a certain pressure, there was no flow due to the resistance offered by wetting liquid (water). It was observed at a certain minimal pressure of 276 kPa, butanol flow started where it was assumed that largest pore was empty. Then at a certain higher pressure of ∼700 kPa, the butanol flow of the dry membrane was equal to wet membrane, i.e. all the pores were available for butanol flow. By using Eq. (7) with the knowledge of these pressures, the pore size range for the membrane was calculated and found in the range of 5.4–13.6 nm. Fig. 12 shows the pure water flux of the clay support and ␥Al2 O3 –clay composite membrane. It is found that the water flux is much lower than that of the clay support, which is attributed to decrease in the pore size on coating. As a result, the hydraulic permeability of ␥-Al2 O3 –clay composite membrane decreases resulting in lower flux. The hydraulic permeability (Lp ) of the membrane is obtained by linear regression of pure water flux vs. applied pressure by using Eq. (9). The values of Lp for support and the composite membrane are found to be 0.4838 × 10−5 and 0.02357 × 10−5 m s−1 kPa−1 , respectively.

Fig. 12. Pure water flux obtained from clay support and ␥-Al2 O3 –clay composite membrane as a function of pressure.

4.4.1. Effect of pH of salt solution The surface charge of the material which depends on the pH of the solution is an important parameter realizing the efficiency of a membrane separation process, especially when removing ionic species. Keeping this in mind, we have measured the rejection over the ranges of pH (2.5–8) for a constant MgCl2 concentration (3000 ppm) and transmembrane pressure (207 kPa). Fig. 13 indicates that the observed rejection of MgCl2 strongly depends on the operating pH value and the highest rejection (75%) is found at pH 2.5. It is already reported in the literature [48,49] that the isoelectric point (IEP) of the ␥-Al2 O3 membrane is 8–9 and hence is electrically positive for pH values lower than 8–9 and negative for pH values greater than 8–9. So the composite membrane is positively charged at the studied pH during the filtration experiments. With increasing pH of the feed solution from 2.5 to 8, the observed rejection of MgCl2 shows a decreasing trend. This can be explained by the variation of interaction between membrane surface and Mg2+ ion at different pH. When a charged membrane is in contact with an electrolyte solution, the concentration of co-ions (ions with the same charge as the membrane) near the surface of the membrane will be lower than that in solution, whereas the counter-ions, which have the opposite charge, have a higher concentration in the membrane than in the solution. Because of this concentration difference,

4.4. Separation of salts solution During the separation of electrolytes, ceramic support can adsorb the salt ions. To avoid the errors in calculating observed rejection, it is an important to determine the extent of adsorption of rejection characteristics of the membranes. This experiment was carried out at a pressure of 207 kPa by following the procedure reported in our earlier work [47]. It was found that the rejection was constant after passages of 45 ml of permeate. We performed the same experiments for both salt solutions.

Fig. 13. Permeate flux and retention of MgCl2 solution as a function of pH (feed concentration = 3000 ppm and applied pressure = 207 kPa).

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

Fig. 14. Variation of permeate flux with applied pressure for MgCl2 and AlCl3 solution (feed concentration = 3000 ppm and pH = 3).

a potential difference is generated at the interface between the membrane and the solution to maintain electrochemical equilibrium between solution and membrane. By this potential, which is called the Donnan potential, co-ions are repelled by the membrane [50]. When pH of the solution increases from 2.5 to 8, the surface potential (charge) of the membrane decreases since the IEP of membrane is 8–9. Therefore the repulsion between Mg2+ and positively charged membrane decreases and consequently reduces the rejection of Mg2+ . Since the cation and anion cannot act independently, Cl− is also rejected to maintain electroneutrality. When the magnitude of the surface potential decreases, the permeate flux increases due to the decreased electroviscous effect (i.e. decrease in the apparent viscosity of the fluid permeating through the membrane pores) which can be seen from Fig. 13 [51,52]. Electroviscous considerations imply a maximum in permeate flux at the IEP of the membrane [51]. 4.4.2. Effect of type of salts and pressure In Fig. 14, the permeate flux of AlCl3 and MgCl2 salts are plotted against the applied pressure for a fixed concentration of 3000 ppm and pH of 3. The solution permeability for both the salts is lower than that of pure water with increasing pressure. It shows that Al3+ and Mg2+ give an additional resistance to the flow due to their presence. It can be seen from Fig. 14 that the permeate flux increases with applied pressure. This can be explained by considering salt transport through the membrane as a result of diffusion and convection, which are respectively due to a concentration and a pressure gradient across the membrane. At lower applied pressure, diffusion contributes and convection dominates the overall process at higher applied pressure. The permeate flux of MgCl2 is higher than AlCl3 salt solution, which depends upon the electrostatic interaction between electrolyte and membrane surface charge giving additional resistance for flow through membrane. The effect of pressure on the rejection of Mg2+ and Al3+ ions is shown in Fig. 15. The membrane shows the following rejection sequence: R(AlCl3 ) > R(MgCl2 ), i.e. rejection increases when the valence of associated cation increases. This is in accordance with Donnan exclusion model. Donnan exclusion effect will be more if cation charge is high. Since the membrane is positively charged, Al3+ ion is excluded more than the Mg2+ which enhances the rejection of AlCl3 salt from its aqueous solution. Additionally, the diffusion coefficient of AlCl3 salt is also lower than MgCl2 . As expected, a lower diffusion contribution to the salt transport through the membrane yields a

189

Fig. 15. Rejection of MgCl2 and AlCl3 as a function of applied pressure (feed concentration = 3000 ppm and pH = 3).

higher retention. A similar type of rejection trend was reported for positively charged polymeric membranes [53]. The observed rejection for both the salts decreases with increase in the pressure as has been reported in literature [21]. With increasing pressure, there is more accumulation of solute particles on the membrane surface that causes severe concentration polarization and also increases the permeate concentration due to increasing convective flux. Thus observed rejection is less as it is calculated using Eq (1) which is based on bulk feed concentration. With this observation, the Spiegler–Kedem model has been used to determine the intrinsic (real) rejection, which calculates the concentration of solute at the membrane surface by taking into account the concentration polarization effects. The model result shows that the concentration at the membrane surface is 2–3 times higher than the bulk concentration for both the solutes. As can be seen from Fig. 15, the intrinsic (real) rejection increases with the applied pressure, which is a common trend followed by electrolytes through charged membranes [21,23]. The maximum rejection of MgCl2 and AlCl3 salt solution is found to be 72% and 88%, respectively. The values of reflection coefficient () and the solute permeability (Pm ) have been calculated from experimental data of the observed rejection and the permeate flux using Eqs. (2)–(6). The  value for MgCl2 and AlCl3 solute is found to be 0.77 and 0.94, respectively. The solute permeability (Pm ) value for MgCl2 and AlCl3 solute is found to be 8.442 × 10−6 m/s and 2.51 × 10−6 m/s, respectively. It is well established that the values of Pm and  depends on the salt concentration, type of salts and the type of membrane [54]. The value of Pm is higher for MgCl2 than AlCl3 which is due to the high amount of salt passing through the membrane. So the  value is lower for MgCl2 salt separation due to the decreasing of the salt rejection. The relationship between permeate flux and intrinsic rejection is generally described by Kimura–Sourirajan model [55,56]. J = Pm

 R  int 1 − Rint

(10)

where J, Pm are the permeate flux (m/s) and solute permeability of the membrane (m/s), respectively. In this model, the maximum intrinsic rejection can reach 100% when permeate flux tends to be infinite; and the minimum rejection equals to zero when the permeate flux tends to zero. It can be seen from Figs. 14 and 15 that the intrinsic rejection increases with increasing permeate flux. The strong dependency of separation on the permeate flux was also observed for negatively charged membrane, neutral membrane [57] and positively charged membrane [58]. Our experimental result

190

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191

tration of 5000 ppm) and 70% (permeate flux of ∼2.0 × 10−5 m/s at 1300 kPa with the salt concentration of 20 ppm), respectively. The iron modified silica membrane [61] and ␥-Al2 O3 ceramic membrane (having pore size of 1 nm) supported on macroporous ␣-alumina [49] showed 10% (permeate flux of ∼ 6.9 × 10−5 m/s at 1725 kPa and salt concentration of 6000 ppm) and 96% (permeate flux of ∼0.7 × 10−5 m/s at 1000 kPa and salt concentration of 2000 ppm) MgCl2 rejection, respectively. As compared to the results reported in the literature, the permeate flux of the membrane reported in this work (9.6 × 10−5 m/s for MgCl2 and 4.4 × 10−5 m/s for AlCl3 ) is one order higher even at lower applied pressure of 550 kPa. In addition, the net separation, which is permeate flux times the difference in concentration of solute in the retentate and permeate is also high. Based on this comparison study, we can conclude that the membrane reported in this work is comparable or even better than the commercial membranes and other inorganic membranes reported in the literature. Fig. 16. Variation of rejection and permeate flux of MgCl2 and AlCl3 salt solution as a function feed concentration (applied pressure = 207 kPa and pH of salt solution = 3).

also confirms that it is true for ␥-Al2 O3 composite membrane, which is positively charged at the studied pH. 4.4.3. Effect of salt concentration Fig. 16 depicts the effect of salt concentration on rejection and permeate flux for a fixed applied pressure (207 kPa) and pH (=3). It is observed that the permeate flux decreases with the increase in salt concentration because of the concentration polarization and partial plugging of pores of the membrane at higher concentration. At the studied pH, the membrane is positively charged. So AlCl3 is much better retained than MgCl2 . It is seen that the observed rejection decreases with increase in salt concentration. The higher value of observed rejection is obtained for both salts at lower feed concentration of 1000 ppm. At higher salt concentration, the membrane charge will be shielded to a large extent, resulting in a lower effective charge and consequently a decrease in rejection is observed. The decrease in retention with increasing salt concentration is typical of charged membranes, for which Donnan exclusion plays an important role [50]. The effect of Donnan exclusion reduces with increasing electrolyte concentrations. With increase in salt concentration, surface concentration also increases leading to severe concentration polarization that enhances solute permeation by diffusion. Consequently, the permeate concentration also increases. In case of AlCl3 (Fig. 16), only a small decrease in rejection is observed with increasing salt concentration, which indicates that the charge effect stays nearly constant for these concentrations. This means that either the effect of membrane charge is so small or the charge effect is still important but does not further decrease at higher concentrations. The rejection (72% for MgCl2 and 88% for AlCl3 ) reported in this work at 550 kPa with salt concentration of 3000 ppm is comparable or even higher than those reported in the literature [21,49,54,59–61]. Shukla and Kumar [21] have obtained the enhanced rejection for AlCl3 (90% for both Z2 and Z3 with permeate flux of ∼1.75 × 10−6 m/s) using chemically modified zeolite–clay composite membranes than the unmodified membrane (84% for Z1 with permeate flux of ∼4.25 × 10−6 m/s) at 420 kPa for salt concentration of 500 ppm. In another work [59], a negatively charged polyamide (PA) composite membrane showed 96.5% (permeate flux of 5.3 × 10−6 m/s) and 92.4% (permeate flux of 4.2 × 10−6 m/s) rejection, respectively for MgCl2 and AlCl3 at 1600 kPa for salt concentration of 2000 ppm. However, the rejection of MgCl2 using commercial membranes, N30F [54] and Desal G-20 [60] was about 6% (permeate flux of ∼0.9 × 10−5 m/s at 900 kPa for the salt concen-

5. Conclusion A ␥-alumina–clay composite membrane has been successfully fabricated using low cost clay materials. Dip coating is used to coat the top layer on clay support using boehmie sol derived from inexpensive aluminium chloride salt by sol–gel route. The particle size distribution of the boehmite sol used for coating is very narrow with a mean particle size of 31 nm. The pore size of the membrane thus prepared is measured by employing bubble point technique and found to be in the range of 5.4–13.6 nm. Separation characteristic of the membrane is checked by performing filtration studies of MgCl2 and AlCl3 salt solution individually as a function of pH, applied pressure and salt concentration. The obtained result confirms that the rejection of MgCl2 strongly depends on the operating pH and the highest rejection (75%) is found at pH 2.5. The rejections to inorganic electrolytes (MgCl2 and AlCl3 ) declined with the growth of electrolyte concentrations accordance to Donnan theory. The intrinsic rejections are found to increase with an increase in the applied pressure, a trend typical of the separation of electrolytes through the charged membranes. The maximum rejection of MgCl2 and AlCl3 is found to be 72% and 88%, respectively for salt concentration of 3000 ppm. References [1] K.A. DeFriend, M.R. Wiesner, A.R. Barron, Alumina and aluminate ultra-filtration membranes derived from alumina nanoparticles, J. Membr. Sci. 224 (2003) 11. [2] Y. Cho, K. Han, K. Lee, Separation of CO2 by modified ␥-Al2 O3 membranes at high temperature, J. Membr. Sci. 104 (1995) 219. [3] X. Changrong, W. Feng, M. Zhaojing, L. Fanqing, P. Dingkun, M. Guangyao, Boehmite sol properties and preparation of two-layer alumina membrane by a sol–gel process, J. Membr. Sci. 116 (1996) 9. [4] M.C. Lovallo, M. Tsapatsis, Preferentially oriented submicron silicilite membranes, AIChE J. 42 (1996) 3020. [5] T. Bein, Synthesis and applications of molecular sieve layers and membranes, Chem. Mater. 8 (1996) 1636. [6] H. Kita, H. Maeda, K. Tanaka, K. Okamoto, Carbon molecular sieve membrane prepared from phenolic resin, Chem. Lett. 26 (1997) 179. [7] F.K. Katsaros, T.A. Steriotis, A.K. Stubos, A. Mitropoulos, N.K. Kanellopoulos, S. Tennison, High pressure gas permeability of microporous carbon membranes, Micropor. Mater. 8 (1997) 171. [8] V. Linkov, R. Sanderson, E. Jacobs, Highly asymmetrical carbon membranes, J. Membr. Sci. 95 (1994) 93. [9] G. Saracco, H.W.J.P. Neomagus, G.F. Versteeg, W.P.M. vanSwaaij, Hightemperature membrane reactors: potential and problems, Chem. Eng. Sci. 54 (1999) 1997. [10] J. Zaman, A. Chakma, Inorganic membrane reactors, J. Membr. Sci. 92 (1994) 1. [11] J. Schaep, C. Vandecasteele, B. Peeters, J. Luyten, C. Dotremont, D. Roels, Characteristics and retention properties of a mesoporous ␥-Al2 O3 membrane for nanofiltration, J. Membr. Sci. 163 (1999) 229. [12] S. Alami-Younssi, A. Larbot, M. Persin, J. Sarrazin, L. Cot, Rejection of mineral salts on a gamma alumina nanofiltration membrane: application to environmental process, J. Membr. Sci. 102 (1995) 123. [13] A. Larbot, J.A. Alary, C. Guizard, L. Cot, New inorganic ultrafiltration membranes: preparation and characterisation, Int. J. High Technol. Ceramics 3 (1987) 143.

A. Majhi et al. / Journal of Membrane Science 340 (2009) 181–191 [14] B.F. Jones, E. Galan, Sepiolite and palygorskite, in: S.W. Bailey (Ed.), Reviews in Mineralogy, Hydrous Phyllosilicates, vol. 19, Mineralogical Society of America, Washington, 1988, pp. 631–674. [15] C. Serna, J.L. Ahlrichs, J.M. Serratosa, Folding in sepiolite crystals, Clay Clay Miner. 23 (1975) 452. [16] H. Nagata, S. Shimoda, T. Sudo, On dehydration of bound water of sepiolite, Clay Clay Miner. 22 (1974) 285. [17] J.L. Pérez-Rodrıguez, E. Galán, Determination of impurity in sepiolite by thermal analysis, J. Therm. Anal. 42 (1994) 131. [18] S. Rakib, M. Sghyar, M. Rafiq, A. Larbot, L. Cot, New porous ceramics for tangential filtration, Sep. Purif. Technol. 25 (2001) 385. [19] M.R. Weir, E. Rutinduka, C. Detellier, C.Y. Feng, Q. Wang, T. Matsuura, R. Le Van Mao, Fabrication, characterization and preliminary testing of all inorganic ultrafiltration membrane made of a naturally occurring sepiolite clay mineral, J. Membr. Sci. 182 (2001) 41. [20] A. Potdar, A. Shukla, A. Kumar, Effect of gas phase modification of analcime zeolite composite membrane on separation of surfactant by ultrafiltration, J. Membr. Sci. 210 (2002) 209. [21] A. Shukla, A. Kumar, Characterization of chemically modified zeolite–clay composite membranes using separation of trivalent cations, Sep. Purif. Technol. 41 (2005) 83. [22] G. Pugazhenthi, S. Sachan, N. Kishore, A. Kumar, Separation of chromium (VI) using modified ultrafiltration charged carbon membrane and its mathematical modeling, J. Membr. Sci. 254 (2005) 229. [23] G. Pugazhenthi, A. Kumar, Chromium (VI) Separation from aqueous Solution using anion exchange membrane, AIChE J. 51 (2005) 2001. [24] S. Workneh, A. Shukla, Synthesis of sodalite octahydrate zeolite–clay composite membrane and its use in separation of SDS, J. Membr. Sci. 309 (2008) 189. [25] K. Khider, D.E. Akretche, A. Larbot, Purification of water effluent from a milk factory by ultrafiltration using Algerian clay support, Desalination 167 (2004) 147. [26] N. Seffaj, S. Alami Younsi, M. Persin, M. Cretin, A. Albizane, A. Larbot, Processing and characterization of TiO2 /ZnAl2 O4 ultrafiltration membranes deposited on tubular support prepared from Moroccan clay, Ceram. Int. 31 (2005) 205. [27] L. Liangxiong, T.M. Whitworth, R. Lee, Separation of inorganic solutes from oil field produced water using a compacted bentonite membrane, J. Membr. Sci. 217 (2003) 215–225. [28] S. Khemakhem, A. Larbot, R. Ben Amar, Study of performances of ceramic microfiltration membrane from Tunisian clay applied to cuttlefish effluents treatment, Desalination 200 (2006) 307. [29] S. Khemakhem, R. Ben Amar, A. Larbot, Synthesis and characterization of a new inorganic ultrafiltration membrane composed entirely of Tunisian natural illite clay, Desalination 206 (2007) 210. [30] D. Mahlab, N.B. Yosef, G. Belfort, Interferometric measurement of concentration polarization profile for dissolved species in unstirred batch hypofilteration (reverse osmosi), Chem. Eng. Commun. 226 (1980) 225–243. [31] M. Clifton, Polarization de concentration dans divers procedes de separation a membrane, These, Universite Paul Sabatier, Toulhouse, France, 1982. [32] V.L. Vilker, C.K. Colton, K.A. Smith, Part I: An optical shadow technique for measuring concentration profiles near solution-membrane interface, Part 2: Theoretical and experimental studies of albumin ultrafiltered in an unstirred cell, AIChE J. 27 (4) (1981) 632–645. [33] K.S. Spiegler, O. Kedem, Thermodynamics of hyperfiltration (reverse osmosis): criteria for efficient membranes, Desalination 1 (1966) 311–326. [34] S. Ghose, C. Bhattacharjee, S. Datta, Simulation of unstirred batch ultrafiltration process based on reversible pore-plugging model, J. Membr. Sci. 169 (2000) 29–38. [35] G. Jinghua, Preparation of Catalytic inorganic membranes by sol-gel technique, Doctoral Thesis, Dalian Institute of Chemical Physics, 1994. [36] T.V.G.C. Vandecasteele, A. Buekenhoudt, C. Dotremont, J. Luyten, R. Leysen, B. Van der Bruggen, G. Maesc, Alumina and titania multilayer membranes for nanofiltration: preparation, characterization and chemical stability, J. Membr. Sci. 207 (2002) 73.

191

[37] E. Jakobs, W.J. Koros, Ceramic membrane characterization via the bubble point technique, J. Membr. Sci. 124 (1997) 149. [38] J.S. Reed, Introduction to the Principles of Ceramic Processing, Wiley, New York, 1989. [39] R.L. Frost, J. Kloprogge, S.C. Russell, J.L. Szetu, Dehydroxylation and structure of alumina gels prepared from trisecbutoxyaluminium, Thermochim. Acta 329 (1999) 47. [40] Ph. Colomban, Raman study of the formation of transition alumina single crystal from protonic ␤/␤ aluminas, J. Mater. Sci. Lett. 7 (1988) 1324. [41] P. McMillan, B. Piriou, The structures and vibrational spectra of crystals and glasses in the silica–alumina system, J. Non-Cryst. Solids 53 (1982) 279. [42] A.D. Cross, An Introduction to Practical IR Spectroscopy, 2nd ed., Butterworth, London, 1964. [43] T.V. Mani, P.K. Pillai, A.D. Damodaran, K.G.K. Warrier, Dependence of calcination conditions of boehmite on the alumina particulate characteristics and sinterability, Mater. Lett. 19 (1994) 237. [44] C.D. Jones, A.R. Barron, Porosity, crystal phase, and morphology of nanoparticle derived alumina as a function of the nanoparticle’s carboxylate substituent, Mater. Chem. Phys. 104 (2007) 460. [45] N.I. Kandri, A. Ayral, C. Guizard, E.H.E. Ghadraoui, L. Cot, Synthesis, characterization and first application of new alumina hydrosols, Mater. Lett. 40 (1999) 52. [46] G. Buelna, Y.S. Lin, Sol–gel-derived mesoporous ␥-alumina granules, Micropor. Mesopor. Mater. 30 (1999) 359. [47] C. Neelakandan, G. Pugazhenthi, A. Kumar, Preparation of NOx modified PMMAEGDM composite membrane for the recovery of chromium (VI), Eur. Polym. J. 39 (2003) 2383. [48] G.A. Parks, The isoelectric points of solid oxides, solid hydroxides and aqueous hydroxo complex systems, Chem. Rev. 65 (1965) 177. [49] G.M. Rios, R. Joulie, S.J. Sarrade, M. Carles, Investigation of ion separation by microporous nanofiltration membranes, AIChE J. 42 (9) (1996) 2521. [50] C.V. Chung, N.Q. Buu, N.H. Chau, Influence of surface charge and solution pH on the performance characteristics of a nanofiltration membrane, Sci. Technol. Adv. Mater. 6 (2005) 246. [51] F.F. Nazzal, M.R. Wiesner, pH and ionic strength effects on the performance of ceramic membranes in water filtration, J. Membr. Sci. 93 (1994) 91. [52] I.H. Huisman, G. Tragardh, C. Tragardh, A. Pihlajamaki, Determining the zetapotential of ceramic microfiltration membranes using the electroviscous effect, J. Membr. Sci. 147 (1998) 187. [53] J.M.M. Peeters, J.P. Boom, M.H.V. Mulder, H. Strathmann, Retention measurements of nanofiltration membranes with electrolyte solutions, J. Membr. Sci. 145 (1998) 199. [54] N. Hilal, H. Al-Zoubi, N.A. Darwish, A.W. Mohammad, Nanofiltration of magnesium chloride, sodium carbonate and calcium sulphate in salt solutions, Sep. Sci. Technol. 40 (2005) 3299. [55] S. Kimura, S. Sourirajan, Analysis of data in reverse osmosis with porous cellulose acetate membranes used, AIChE J. 13 (1967) 497. [56] R. Levenstein, D. Hasson, R. Semiat, Utilization of the Donnan effect for improving electrolyte separation with nanofiltration membranes, J. Membr. Sci. 116 (1996) 77. [57] T. Tsuru, M. Urairi, S.I. Nakao, S. Kimura, Reverse osmosis of single and mixed electrolytes with charged membranes: experiment and analysis, J. Chem. Eng. Jpn. 24 (4) (1991) 518. [58] Y. Xu, R.E. Lebrun, Investigation of the solute separation by charged nanofiltration membrane: effect of pH, ionic strength and solute type, J. Membr. Sci. 158 (1999) 93. [59] C. Wu, S. Zhang, D. Yang, J. Wei, C. Yan, X. Jian, Preparation, characterization and application in wastewater treatment of a novel thermal stable composite membrane, J. Membr. Sci. 279 (2006) 238. [60] M.D. Afonso, G. Hagmeyer, R. Gimbel, Streaming potential measurements to assess the variation of nanofiltration membranes surface charge with the concentration of salt solutions, Sep. Purif. Technol. 22–23 (2001) 529. [61] J.M. Skluzacek, M.I. Tejedor, M.A. Anderson, An iron-modified silica nanofiltration membrane: Effect of solution composition on salt rejection, Micropor. Mesopor. Mater. 94 (2006) 288.