Desalination 279 (2011) 104–114

Contents lists available at ScienceDirect

Desalination

j ourna l homepage: www.e lsev ie r.com/ locate /desa l

Effect of TiO2 addition on the fabrication of ceramic membrane supports: A study onthe separation of oil droplets and bovine serum albumin (BSA) from its solution

P. Monash, G. Pugazhenthi ⁎Department of Chemical Engineering, Indian Institute of Technology Guwahati, Guwahati, Assam 781039, India

⁎ Corresponding author. Tel.: +91 361 2582264; fax:E-mail address: pugal@iitg.ernet.in (G. Pugazhenthi)

0011-9164/$ – see front matter © 2011 Elsevier B.V. Adoi:10.1016/j.desal.2011.05.065

a b s t r a c t

a r t i c l e i n f o

Article history:Received 9 February 2011Received in revised form 21 May 2011Accepted 27 May 2011

Keywords:Membrane supportPermeabilitySeparationOil dropletBSA

Low cost porous ceramic membrane support was prepared using kaolin, pyrophyllite, feldspar, ball clay,quartz and calcium carbonate mixture along with PVA as a binder. The main intention of this work is to studythe changes in the properties of the membrane supports (pore size, porosity, mechanical strength and purewater permeability) by the addition of titanium dioxide (TiO2). Three membrane supports, namely, support-I,3G and 6G supports (naming is based on the loading of TiO2) were prepared by uniaxial compaction method.All the membrane supports were sintered at 950 °C and systematically analyzed using thermogravimetric(TG), particle size distribution (PSD), X-ray diffraction (XRD), N2 adsorption–desorption and scanningelectron micrograph (SEM) analysis. The porosity of the support-I, 3G and 6G supports was found to be 44, 38and 36% with an average pore diameter of 0.98, 0.93 and 0.83 μm, respectively. The flexural strength of themembrane supports was increased with the addition of TiO2 and ranges between 28 and 33 MPa. Solventpermeation studies through these membrane supports revealed that the non polar solvents were morepermeable than the polar solvents and the transport mechanism was mainly controlled by viscosity of thesolvents. Performance of the membrane supports was investigated for oil-in-water emulsion separation byvarying the feed concentration and applied pressure. A maximum rejection of 99% was obtained with 6Gsupport for an oil concentration of 200 ppm. Also, an attempt has been made for the separation of bovineserum albumin (BSA) from its solution using these membrane supports and a reasonable rejection of 40% wasobtained for 100 ppm of BSA solution.

+91 361 2582291..

ll rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Ceramic membranes have been gaining more attention inindustrial application, such as food, pharmaceutical and electronicindustries etc., due to its excellent thermal, chemical and mechanicalstability, higher separation efficiency, longer life time and ease ofcleaning [1–4]. In general, performance of any ceramic membranesmainly depends on the porous support. An excellent membranesupport should possess higher mechanical strength, higher perme-ability, narrow pore size distribution and low manufacturing cost[5,6]. The aforesaid parameters mainly depend on the starting rawmaterials, sintering temperature and method of fabrication.

Many fundamental research works have been well explainedwith α-alumina [7–9]. However, owing to higher cost of α-alumina,many research works were focused on low cost clay/ceramic materialcontaining aluminosilicates [10–17]. Few literature were found forthe fabrication of membrane supports using mixture of clays [12,15].In general, calcium carbonate is considered as a porosity controlling

agent. The addition or removal of CaCO3 has a significant effecton the pore size and pore size distribution of the membranesupport [18,19]. Moreover, the pore size and its distribution of themembrane support have also a significant effect on the selective layerformation.

In general, macroporous membrane supports were applied forvarious separation applications, such as separation of salt, dye, heavymetals, oil emulsion and proteins [4,14,16,20–23]. Very few reportswere available on the separation of oil droplets from oil-in-wateremulsion systems and protein separations [24–31].

Large volumes of oily wastewater with oil concentrations rangingbetween 50 and 1000 ppm have been produced from petroleumrefineries, petrochemical, metallurgical and transportation industries[21,25–27]. This ecologically hazardous oily wastewater should betreated before being discharged due to environmental regulations. Theexisting tolerance limit of total oil and grease concentrations inwastewater is ranging between 10 and 15 ppm [28,30,32,33]. Thismay vary based on the environment regulations of that country. Varioustraditional methods, such as gravity settling (API separator), skimming,dissolved air flotation, coalescence and centrifuging have been appliedfor treating the unstable oil/water emulsion [24,26,28,31]. Manytechnological solutions have been reported for the treatment of oilywastewaters of feed oil concentrations ranging between 500 and

105P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

3000 ppm [24,27,29]. Higher concentration of the feed containsunstable oil droplets having sizes greater than 50 μm that could betreated easily [24,27,29]. However, the oil emulsion containing oildroplets size of less than 20 μm (exists with oil concentrations below100 ppm) is so stable. This can't be treated effectively by the traditionalmethods [24,31,34]. In such cases, membrane technology has beenfound to be more effective than the conventional methods for thetreatment of oil at low concentration [27,28,34]. Both microfiltrationand ultrafiltration have been used for concentrating/treating oilemulsions, as they are highly efficient for removing oil, do not requirechemical additives and more economical than conventional separationtechniques.

This work is focused on the investigation of membrane supportproperties by the addition/removal of TiO2/CaCO3 to the rawmaterialsused for the fabrication of porous membrane support (support-I).Porousmembrane support (support-I)was preparedwith amixture ofkaolin, ball clay, feldspar, pyrophyllite, quartz and calcium carbonatewith poly vinyl alcohol (PVA) as a binder. 3g and 6g of TiO2were addedby removing an equal amount of CaCO3 from the raw material com-position of support-I to form the newmembrane supports (3G and 6Gsupports). The porous membrane support without TiO2 was named assupport-I and the two different amounts (3g and 6g) of TiO2 loadedmembrane supports were named as 3G and 6G supports, respectively.All the membrane supports were fabricated by uniaxial pressingmethod and sintered at 950 °C. The prepared membrane supportswere investigated for its porosity, pore size distribution, mean poresize, flexural strength, permeate flux and rejection performance. Forthe aforesaid investigation, the rawmaterials andmembrane supportswere characterized with various analytical techniques and testingmethods. Rejection performance of the membrane supports wasstudied for oil–water emulsion system. A brief study on fouling of themembrane supports for oil water system was also been carried out inthiswork to understand the rejectionmechanism. An attemptwas alsobeen made for the separation of bovine serum albumin (BSA) proteinfrom its aqueous solution.

2. Materials and methods

The raw materials (kaolin, ball clay, feldspar, pyrophyllite andquartz) were collected from Kanpur, India. Calcium carbonate, titaniumdioxide, HCl, NaOH, BSA and Poly vinylalcohol (PVA, M.W. 72000) wereprocured from Merck, Mumbai, India. Crude oil was collected fromGuwahati Refinery, Indian Oil Corporation Limited (IOCL), India. Oil-in-water emulsions were prepared without any further treatment of thecrude oil. Water used in this work was collected from Millipore system(ELIX-3).

2.1. Membrane support fabrication

Rawmaterials of different compositions (as given in Table 1) weremixed with 4 ml of PVA solution (2 wt.%) in a ball mill at 40 rpm for1200 s and the resulting mixture was sieved in a 40 mesh standardscreen. Circular disk shaped greenmembrane supports were preparedby taking a required amount of sieved powder and pressed at 50 MPa

Table 1Composition of the raw materials used for the preparation of porous membranesupports.

Raw materials Support-I (wt.%) 3G support (wt.%) 6G support (wt.%)

Kaolin 14.5 14.5 14.5Ball clay 17.6 17.6 17.6Feldspar 05.6 05.6 05.6Quartz 26.6 26.6 26.6Pyrophyllite 14.7 14.7 14.7Calcium carbonate 17.0 14.0 11.0Titanium dioxide – 03.0 06.0

using a hydraulic press. The fabricated greenmembrane supportsweredried and sintered at 950 °Cwith a controlled heating and cooling rate.Sintered membrane supports were polished and shaped with a siliconcarbide abrasive paper (C-120, C-220 and C-320) and sonicated in anultrasonic bath to remove the loosematerials adheredon the surface ofthe membrane support during polishing. Detailed fabrication proce-dure of the membrane supports was reported elsewhere [35].

2.2. Membrane support characterization

Rawmaterials andmembrane supportswere characterized inMettlerToledo thermo gravimetric analyzer (TGA/SDTA 851® model) undernitrogen atmosphere at a heating rate of 10 °C/min from 25 to 950 °C in a900 μl alumina crucible. Particle size of the raw materials was analyzedusing Malvern Mastersizer 2000 (APA 5005® model, hydro MU) in wetdispersion mode. The X-ray diffraction (XRD) patterns were acquired fordiffraction angle (2θ) ranges between 5 and 75 degree with a scan speedof 0.05°/s using Bruker AXS instrument with Cu Kα (λ=0.154506 nm)radiation operating at 40 kV and 40 mA. Nitrogen adsorption/desorptionisotherms were measured by the BET method at −196 °C usingBeckman-Coulter surface area analyzer (SATM 3100 model). Prior to N2

adsorption/desorption analysis, themembrane supports (3 mmdiameterand 2 mm thickness) were degassed at 150 °C for 3 h. Scanning electronmicrograph (SEM) analysiswas carried out using variable pressure digitalscanning electron microscope (LEO 1430VP® model). Porosity of themembrane support was calculated by Archimedes' principle using wateras an immersing medium. Three point flexural strength measurementswereperformedon60×5×5mmrectangularbarsusinguniversal testingmachine (M/s Deepak Polyplast, Model: DUTT-101, Mumbai, India) witha span length of 50 mm and crosshead speed of 0.5 mm/min.

Water and solvent permeation experiments were conducted in adead-end filtration setup made up of stainless steel 316. Rejectionperformances of the membrane supports for oil-in-water emulsionand BSA were also carried out with the same permeation setup. Theconcentration of oil-in-water and BSA was determined using a standardcalibration graph of absorbance versus concentration prepared with aUV–visible spectrophotometer (Perkin-Elmer, model: Lambda 35, USA)at the wavelength of 236 and 280 nm, respectively. pH of the solutionswas measured using a calibrated pHmeter (Eutech instruments, model:pH 510). Surface tension of the oil-in-water emulsion was estimated at25 °C using a GBX 3S tensiometer with an accuracy of 0.01 mN/m. AplatinumduNuoy ringwasusedasaprobeand standardizedwithmilli-Qwater. The reported values of surface tensionwere the average of at leastthree measurements. Three membrane supports of each compositionwere tested for all the experiments and an average value was reported.

3. Experimental

3.1. Preparation of feed solutions for microfiltration experiments

Five different concentrations (50, 75, 100, 150 and 200 ppm) of oil-in-water emulsions were prepared using a sonication bath (Make:Elmasonic; Model: S30H). Stable oil-in-water emulsions were obtainedby operating the sonication bath for 15–25 h at a temperature of 25 °Cand no surfactantswas added because of the natural surfactants presentin the crude oil itself yielded a highly stable emulsion. The completion ofthe emulsification process can be evidenced by the disappearance ofoily layer on the water surface. Stability of the emulsion was furtherconfirmed bymeasuring the droplet size distribution and its absorbancewavelength (236 nm)at regular intervals. Thepreparedemulsionswerestable for 7–8 days. After that, a thin oil film on the water surface wasformed indicating that the oil droplets were coalesced and formedunstable emulsion. So all the separation performance experimentswerecarried out within 1–2 days and analyzed immediately to avoid anyexperimental errors. The standard deviation of absorbance of eachprepared samples varies within ±0.102.

100 200 300 400 500 600 700 800 90080

90

100

Wei

gh

t (%

)

Temperature (oC)

(a)

80

90

100(b)

80

90

100(c)

Fig. 1. Thermogravimetric analysis of the membrane supports with and without binder.(a) Support-I; (b) 3G support; (c) 6G support (dashed line denotes with binder, solidline denotes without binder).

106 P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

BSA samples were prepared by dissolving an appropriate amountof BSA in Millipore water. The pH of the freshly prepared solution wasfound to be 7.0. Effect of pH on the rejection of BSA was studied byaltering the solution pH using HCl and NaOH solutions. The preparedBSA solution was utilized within 6 h to minimize the aggregation ordenaturation of the proteins during storage.

4. Result and discussions

Particle size distribution of individual rawmaterials is analyzed andits distributions reported here is the average of two runs (Supplemen-tary material 1). The volume weighted mean diameter of the raw

0

300

600

900

KK

Q

KK

C CC C

F

FP

P

PP PQ

(a)

0

300

600

900

R

(b)

0 30 600

400

800

1200

Inte

nsi

ty (

Arb

itar

y U

nit

s)

R

(c)

0

2θ (Degree)

Fig. 2. XRD patterns of the membrane supports before and after sintering. (a) Support-I; (b)denotes after sintering). A—anorthite; C—calcium carbonate; CaO—calcium oxide; F—feldsp

materials is found to be in the range of 1 to 10 μm with a span valueranging between 1.8 to 3 indicating similar width of size distribution,which results in uniformmixing of the rawmaterial [36,37]. The particlesize distribution (Supplementary material 2) of the 3 different supportmixtures (raw materials used for preparing support-I, 3G and 6Gsupports) is found to be in the range of 0.3 to 90 μm with an averageparticle size varying between 7.5 and 8.6 μm. This indicates uniformmixing of rawmaterials to accomplish a good macroporous membranesupport.

Thermogravimetric curves (supplementary material 3) of theindividual raw materials indicate that there is a weight loss up to820 °C and a negligible weight loss after 820 °C indicating that themembrane support must be sintered above 850 °C. In our earlier work[35], the optimized sintering temperaturewas found tobe950 °C.Henceall the membrane supports were sintered at 950 °C. The binder presentin the membrane support is completely removed at 720 °C for all themembrane supports and the binder burnout mechanism follows thesame trend as shown in Fig. 1. During sintering, the calcium carbonatepresent in the raw material is converted into CaO and CO2 that causesthepores apart from theparticle–particle sintering. Theporosity and thetortuosity of themembrane supportsmust depend on the path taken bythe CO2 and the other volatile matters evolved during sintering. For the3G and 6G supports, the quantity of CaCO3 is less compared to that ofsupport-I that might have an effect on the pore formation and porosity.

The XRD patterns of the membrane supports before and aftersintering are presented in Fig. 2. In general, sintering produces a seriesof reactions or phase transformations that leads to the formation ofnew phases. These results in the disappearance and shift in theposition of the XRD peaks. Although many phase transformationsoccur during sintering of the membrane supports, the main phasetransformation of the prepared membrane supports is the conversionof kaolin to mullite. This is evidenced by the disappearance ofthe peaks at 2θ value of 8.8, 9.6 and 12.6°. A rutile phase of TiO2 isobserved for the 3G and 6G supports before and after sintering. The

0

100

200

300

400

MM

Q

FW

CaOCaO

(a)

0

100

200

300

400

R

(b)

30 600

200

400

600

2θ (Degree)

R

(c)

3G support; (c) 6G support (left hand side denotes before sintering and right hand sidear; K—kaolin, M—mullite; P—pyrophyllite; Q—quartz; R—rutile; W—wollastonite.

107P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

background noise in the XRD pattern of the sintered supports sug-gests that there may be an existence of amorphous silica [38,39]. Thisamorphous silica reacts with the CaO to form new phases such aswollastonite (CaSiO3) and anorthite (CaO.Al2O3.2SiO2) that is alsoidentified in the XRD analysis. Quartz remains as the most abundantphase for all the three membrane supports. Few unidentified weakpeaks are also observed after sintering, which is attributed to othercalcium aluminates or silicate. Crystallite size of the membrane sup-ports before and after sintering was calculated using Scherrer's for-mula for the maximum intensity peak at 2θ=26°

dXRD =Kλ

β:cosθð1Þ

where K is the shape constant (0.9), λis thewave length (0.154056 nm)of CuKα radiation, βis the full width at half maximum (radian) and θ isthe diffraction angle (degrees). In general, a decrease in the crystallitesize indicates a higher degree of particle interlockingwhichmay lead todensification of the membrane support [39]. The decrease in thecrystallite size is found to be 8, 20 and 23% for the support-I, 3G and 6Gsupports, respectively. 3G and 6G supports are more densified than thesupport-I suggesting that there may be a decrease in the pore size andpore size distribution, which have a significant effect on permeability.

Surfacemorphologyof the sinteredmembrane supports is presentedin Fig. 3. The pore size distribution and average pore diameter of thesintered membrane support are estimated from SEM micrographs(see Fig. 3) using ImageJ software (open source software provided byNational Institute of Health (NIH), weblink: http://rsbweb.nih.gov/ij/

(c)

(a)

Nu

mb

er o

f P

ore

s (%

)

Fig. 3. Microstructure and pore size distribution analysis of the membrane supports. (a)estimated using ImageJ open software.

download.html) for pores having circularity greater than 0.8. Tominimize the errors of image analysis, six SEMmicrographs (randomlyselected sections of 3 membrane supports for each composition) areanalyzed. Average pore diameter is calculated using the followingequation

Davg =∑n

i=1nidi

∑n

i=1ni

ð2Þ

where ni is the number of pores with a pore diameter di. The averagepore diameter of the support-I, 3G and 6G supports is found to be 1.01,1.06 and 0.97 μm, respectively.

Porosity of the membrane supports is estimated with ImageJsoftware after thresholding (threshold the pores only) the 8-bit imageusing “Analyze Particles” feature of this software. Since the pores areinterconnected during thresholding, the porosity of the membranesupports ismeasuredmanually by choosing three rectangular sections(area) of three SEM images of similar magnification. The porosity iscalculated by dividing the porous area (assuming cylindrical pores) bythe total area. Themanually obtained porosity valuematches verywellwith the thresholdmethod of analysis. Total porosity of themembranesupports is also determined by Archimedes' method using the fol-lowing equation

ε =MW −MD

MW −MAð3Þ

(b)

0 1 2 3 4

0

5

10

15

20

25

Pore Size (µm)

Support-I 3G Support 6G Support

Support-I; (b) 3G support; (c) 6G support; (d) pore size distribution of the supports

108 P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

whereMD is themass of the dry support,MW is themass of the supportwith pores filled with water (pores are filled with water undervacuum), MA is the mass of the water saturated support measured inwater (A refers to Archimedes) and ε is the porosity of the membranesupports.

The porosity obtained by the Archimedes' method and SEManalysis is presented in Fig. 4. The porosity of the membrane supportestimated by Archimedes' method is found to be 44, 38 and 36% forthe support-I, 3G and 6G supports, respectively. High porosity values(50, 44 and 39% for the support-I, 3G and 6G supports, respectively)observed by SEM image analysis are due to the difficulty in controllingthe threshold of the pores of the membrane supports. The decrease inthe porosity of the 3G and 6G supports is due to the decrease in theamount of the porosifier (i.e. CaCO3). Similar types of results werereported in the literature for the increase in porosity with an increasein the amount of the porosifier [15,18,19].

The flexural strength (three points bending) of the membranesupports is calculated by the following equation

σ =3Fl2bt2

ð4Þ

where σ is the flexural strength (MPa), Fis the load at the fracturepoint, l is the span length, b is the width of the sample and t is thethickness of the sample. Flexural strength for the support-I, 3G and 6Gsupports is found to be 28, 31 and 33 MPa, respectively. Increase in theflexural strength of the 3G and 6G supports is due to the decrease inthe porosity of the membrane supports by the addition of TiO2. On theother hand, the removal of CaCO3 from the raw materials decreasesthe porosity of the membrane support thereby increasing the flexuralstrength of the 3G and 6G membrane supports.

Water permeation studies were carried out using a homemadefiltration setup [40]. Prior to permeation studies, water was passedthrough the fresh supports at higher pressure until a constant fluxwasobtained. The above procedurewas done to remove any loose particles(particles which are not removed during ultrasonication) present inthe pore path. All the experiments were carried out by filling 250 ml ofwater in the top compartment. After discarding the first 50 ml ofwaterat a fixed pressure, the time required to collect next 50 ml was noteddown to calculate the water flux at that applied pressure using thefollowing equation

Jw =Q

AΔTð5Þ

Support-I 3G support 6G support0

10

20

30

40

50

Po

rosi

ty (

%)

Membrane support

SEM image analysis Archimedes' method

Fig. 4. Porosity of the membrane supports.

where Jw is the pure water flux (m3 m−2 s−1), Q is the volume ofwater permeated (m3), A is the effective membrane area (m2) and ΔTis the sampling time (s).

Water flux of the membrane supports with applied pressure ispresented in Fig. 5. The linear dependence of water flux on pressuredrop across the membrane supports indicates Poiseuille flow throughthe pores. Average pore radius of the membrane support is estimatedby using the deduced form of Hagen–Poiseuille (H–P) equation byassuming cylindrical pores.

rm =

ffiffiffiffiffiffiffiffiffiffiffiffiffiffi8μΔxJεΔP

rð6Þ

where rm is the mean pore radius (μm), μ is the viscosity (kPa S), Δx isthe thickness (μm), J is the permeate flux rate (μm/s), ε is the porosity,

ΔP is the transmembrane pressure (kPa), andJΔP

is the permeability

(μm/s kPa) determined from the slope of the linear relationshipbetween the pure water flux (J) and transmembrane pressure (ΔP).

The average pore diameter calculated from the water permeationdata using H–P equation is found to be 0.98, 0.93 and 0.83 μm for thesupport-I, 3G and 6G supports, respectively. In general, the asym-metric ceramic membrane supports are more tortuous and the poresize of the ceramic membrane supports calculated using H–P equationis always lower than the actual pore size. So, the pore diameterestimated from the H–P equation is considered for further calculationsbecause it gives the average along total permeation pathway throughwhich a solute/solution travels.

Pure solvent permeability of the membrane supports is also testedusing the same permeation setup. Polar (alcohols) and non-polar(alkanes) aliphatic solvents are chosen to study the interaction of themembrane supports with the solvents (hydrophobic/hydrophilic)as well as the resistance of the prepared supports in these solvents.Permeation experiments are performed in the following sequence:starting with methanol, ethanol, propanol, butanol, acetone, toluene,pentane, hexane and ended with heptane. While changing thesolvents from one to other, the cell is loaded and flushed with 50 mlof the next solvent. The physical properties of the solvents used in thiswork are presented in Table 2. For all themembrane supports, the fluxincreases linearly with the applied pressure for all solvents (Fig. 6)indicating that the pressure difference is the only driving force for thepermeation of solvents. The linear fit for the flux versus appliedpressure plot is greater than 0.9, irrespective of the types of thesolvents. This reveals that the transport mechanism obeys viscous

0 50 100 150 200 250 300 3500

2

4

6

8

10

12

14

16

18

Pu

re W

ater

Flu

x x

104 (

m/s

)

Applied pressure (kPa)

Support-I 3G Support 6G Support

Fig. 5. Pure water flux of the membrane supports.

Table 2Physical properties of the solvents used in the permeation experiments.

Solvent Molar volume(cm3/mol)

Viscosity(cP)

Density(g/cm3)

Dielectricconstant

Surface energymN/m

Kineticdiameter (nm)

Molar volume/Viscosity(cm3/mol cP)

Methanol 40.7 0.59 0.7917 32.6 22.6 0.38 69.0Ethanol 58.5 1.20 0.7893 24.3 22.3 0.44 48.8Propanol 76.9 2.25 0.7854 20.1 23.8 0.48 34.3Butanol 91.5 2.95 0.810 17.8 24.6 0.50 31.0Water 18.0 1.02 0.998 80.3 72.0 0.27 17.7Acetone 74.0 0.30 0.792 20.7 23.3 0.63 247.0Toluene 25.4 0.56 0.8669 2.38 28.0 0.59 184.4Pentane 116.2 0.234 0.626 1.80 16.0 0.43 496.0Hexane 131.6 0.32 0.659 1.90 17.9 0.45 411.3Heptane 146.2 0.40 0.684 1.93 19.7 0.47 366.0

109P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

flow model given by Darcy. From Fig. 6, one can clearly seen that thesolvent flux increases with decreasing viscosity of the solvent indi-cating that the viscosity is a main factor that controls the transport ofsolvent through the membrane supports. A good correlation betweenthe solvent permeability and inverse of solvent viscosity (except foracetone and toluene) validates the above comments (see Fig. 7). It isobserved that the nonpolar solvents (hexane, toluene) show higherpermeability than the polar solvents. This confirms that the inter-action of the solvents with membrane supports is also one of the

0 50 100 150 200 2500

10

20

30

40

50

60

70

80

90

Applied Pressure (kPa)

Water Methanol Ethanol Propanol Butanol Acetone Toluene Pentane Hexane Heptane

(a)

0 50 100 150 200 250 300 350 4000

5

10

15

20

25

30

Applied Pressure (kPa)

Water Methanol Ethanol Propanol Butanol Acetone Toluene Pentane Hexane Heptane

(c)

Flu

x x

104 (

m/s

)F

lux

x 10

4 (m

/s)

Fig. 6. Permeabilities of various solvents through the three different m

factors that limit the permeation. If the viscosity of the solvent is theonly controlling factor then the flux of hexane and acetone as well asmethanol and toluene would be identical because of identicalviscosities. This implies that the support-solvent interactions musthave an influence on solvent permeability in addition to viscosity[41,42]. In general, solvent interactions can be expected to varywith changes in solvent properties such as surface tension, di-electricconstant and molecular size [41–43]. In addition to the aboveparameters, the polarity of the solvent also plays a vital role because

0 50 100 150 200 250 300 350 4000

5

10

15

20

25

30

35

Applied Pressure (kPa)

Water Methanol Ethanol Propanol Butanol Acetone Toluene Pentane Hexane Heptane

(b)

Methanol Ethanol Propanol Butanol Acetone Toluene Water Pentane Hexane Heptane0

5

10

15

20

25

30

35

Solvents

Support-I 3G Support 6G Support

So

lven

t P

erm

eab

ility

x 1

06 (m

/s k

Pa)

Flu

x x

104 (

m/s

)

embrane supports. (a) Support-I; (b) 3G support; (c) 6G support.

0.01 0.1 1 10 100

0

2

4

6

8

10

Vo

lum

e (%

)

Droplet Size (µm)

Oil Concentration 50 ppm 75 ppm 100 ppm 150 ppm 200 ppm

Fig. 8. Droplet size distribution of the oil-in-water emulsion.

0 1 2 3 4 50

10

20

30

H

Hx

A

Pn

M

T

W

EPB

1/Viscosity (1/cP)

(a)0

5

10Hx

A

PnH

M

T

W

EPB

(b)0

5

10

A

T

Pn

HxH

W

BP

E

M

(c)S

olv

ent

Per

mea

bili

ty x

106 (

m/s

kP

a)

Fig. 7. Variationof thesolventpermeabilities to its viscosityof thethreedifferentmembranesupports. (a) Support-I; (b) 3G support; (c) 6G support. B—butanol; P—propanol;E—ethanol; W—water; A—acetone; M—methanol; H—heptane; Hx—hexane; T—toluene;Pn—pentane.

110 P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

it is strongly related with surface tension [43]. Increased solventpolarity results in decreased flux through these membrane supports.Based on these results, it is clear that the solvent flux of themembranesupports is mainly influenced by applied pressure, solvent viscosityand interactions of the solvents with membrane supports. Similarobservations were also reported in the literature for α-alumina andother types of membranes [42–45].

Performance of the membrane supports for the separation of oil-in-water emulsion was tested with the permeation cell filled with100 ml of the feed solution. At a fixed pressure, the first 20 ml of thecollected permeate was discarded and the time required to collect thenext 10 ml of permeate was noted down to calculate the permeateflux at that pressure. The observed rejection of the membrane supportwas calculated by the following equation

R =Cf − Cp

Cf× 100 ð7Þ

where R is the observed rejection (%), Cf is the concentration of thefeed solution (ppm) and Cp is the concentration of the permeatesolution (ppm). After each experimental run, the membrane supportswere cleaned with a detergent solution to obtain the same water fluxof the membrane supports. The variation in pure water flux of themembrane supports before and after cleaning was found to benegligible (b±2%).

The performance of the membrane supports for rejecting the oil-in-water emulsion was performed at five different concentrations (50. 75,100, 150 and 200 ppm) and applied pressure ranging between 69 and345 kPa. The oil droplet size distribution of the feed solution havingdifferent concentration obtained using particle size analyzer ispresented in Fig. 8. The oil droplet diameter is ranging between 0.05and 4 μmwith a volumemedian diameter (d0.5) ranging between 0.716and 0.743 μm for oil concentrations of 50, 75 and 100 ppm solutions. Avolumemedian diameter (d0.5) ranging between 1.183 and 1.211 μm isobserved for 150 and 200 ppm solutions. The volume median diameterof the oil droplet is higher than the pore diameter of the membranesupports suggesting a greater possibility for the rejection of oil droplets.The rejection and permeate flux of the membrane supports for oil-in-water emulsion system are shown in Fig. 9. In general, permeabilitydepends on the diameter of the pores and properties of the membranesupports. Therefore, larger support pore sizes provided higher perme-ability. From Fig. 9, it is clear that the rejection of oil droplets increaseswith an increase in the feed concentration and decreases with anincrease in the applied pressure. The increased rejection obtained at

higher concentration for all the membrane supports is due to theincrease in the oil droplet size and droplet density. All the membranesupports showed ~70–99% rejection and the maximum rejection isobtained with the 6G support when compared to other supports(support-I and 3G support). However, one can see the flux is moreaffected if the concentration of oil-in-water emulsion is increased. Fig. 9clearly reveals that the higher feed oil concentration corresponds togreater flux reduction and rejection. Higher concentration of the oilleads to coalescence of the oil droplets forming a bigger droplet thatresult in a higher rejection. The coalesced oil adheres on the surface ofthemembrane supports which causes fouling and this result in reducedflux. Similar types of results were reported in the literature[24,25,27,30,32]. At higher pressures, the oil droplet deforms and passesthrough the small pores resulting in decreased rejection. Sometimes therejection of oil is found to be higher at higher pressures. Thismay be dueto the formation of large oil droplets. The reduction in the flux is dueto the pore blocking mechanism of oil with the membrane supports.To understand the pore blocking mechanism, studies were carried outfor the oil concentrations of 100 and 200 ppm at 69 kPa for all themembrane supports and the permeate was collected for 30 mincontinuously.

In general, the resistance to permeate flow in microfiltrationoccurs due to two major factors. The first factor is due to the porespartially occupied by small particles/molecules, which is related tomembrane properties. The second one is due to the formation of afouling layer on the membrane surface, which is related to theoperating conditions such as transmembrane pressure difference andfeed concentration [24,26,34]. The membrane fouling in dead-endfiltration at a constant pressure has been generally explained by theHermia's model. The reformatted form of Hermia's model is given bythe following formula [25,34].

J = J0 1 + k 2− nð Þ AJ0ð Þ2−nth i1= n−2ð Þ ð8Þ

where J0 and J represent the permeate flux at t=0 and t= t,respectively. Theparameter,n, represents different foulingmechanisms.The value of n corresponds to 1.5, 1.0 and 0 in the above equationrepresents the standard pore fouling, the intermediate pore fouling andthe cake filtration models, respectively. The obtained experimentalpermeate flux was fitted with the above models (linearized form). Theplot of flux versus time and the obtained parameter values are reportedin the Supplementary materials 4 and 5, respectively. The regressioncoefficients of the experimental data (see Supplementary material4) indicate that the cake filtration is the dominating mechanism for

0 50 100 150 200 250 300 350 4000

10

20

30

40

50

Applied Pressure (kPa)

(a)

0 50 100 150 200 250 300 350 400 450

70

80

90

100

Rej

ecti

on

(%

)

70

80

90

100

Rej

ecti

on

(%

)

70

80

90

100

Rej

ecti

on

(%

)

Applied Pressure (kPa)

(d)

0 50 100 150 200 250 300 350 4000

2

4

6

8

10

12

14

16

Applied Pressure (kPa)

(b)

0 50 100 150 200 250 300 350 400

Applied Pressure (kPa)

(e)

0 50 100 150 200 250 300 350 4000

1

2

3

4

5

6

7

8

Applied Pressure (kPa)

(c)

0 50 100 150 200 250 300 350 400

Applied Pressure (kPa)

(f)

Per

mea

te F

lux

x 10

5 (m

/s)

Per

mea

te F

lux

x 10

5 (m

/s)

Per

mea

te F

lux

x 10

5 (m

/s)

Fig. 9. Permeate flux and rejection performance of the supports for the oil-in-water emulsion system. (a), (b) and (c) show permeate flux of support-I, 3G and 6G supports.(d) (e) and (f) show the rejection performance of support-I, 3G and 6G supports. 50 ppm, 75 ppm, 100 ppm, 150 ppm, 200 ppm.

111P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

fouling. However, it is very difficult to identify the region at which itswitches from one dominant fouling mechanism to another.

In general, hydrophobic membranes have a greater tendency tofoul than that of the hydrophilic membranes [24,27,31]. However, onecan reduce the fouling by operating at a pressure lesser than thecapillary pressure of oil droplets [24]. The surface tension of the oil-in-water emulsion is obtained using tensiometer (ranging between 54.25and 57.25 mN/m) and the estimated capillary pressure of oil dropletsis ranging between 110 and 130 kPa. The above results indicate that

the employment of applied pressure greater than 130 kPa is probablyunwise for the separation of oil-in-water emulsion in the dead-endexperiments and this may lead to poor flux due to pore blocking.

The separation potential of the membrane supports was studiedusing BSA protein solution and the influence of the operating con-ditions such as applied pressure, pH and feed concentration was alsoinvestigated. The effect of pressure on the separation of BSA wasperformed at the feed concentration of 500 ppm by applying pressureranging between 69 and 345 kPa. The rejection performance of BSA

112 P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

with respect to applied pressure, concentration and pH on themembrane supports is presented in Fig. 10. The observed rejection isfound to be decreased with an increase in the applied pressure. BSA isa globular prolate ellipsoid of dimensions 140×40×40 nm andhaving a molecular weight of 67 kDa. The approximate size of theBSA is found to be 11 nm in diameter, which is calculated using anequation given in the literatures [4,23]. Since the pore size of themembrane supports is greater than the size of BSA molecule, theincrease in the applied pressure leads to more permeation of BSA

0 50 100 150 200 250 300 350 4000

5

10

15

20

25

30

35

Applied Pressure (kPa)

Support-I 3G Support 6G Support

(a)

0 500 1000 1500 2000 2500 30000

5

10

15

20

25

30

35

40

Concentration (ppm)

support-I 3G support 6G support

(c)

3 4 5 6 7 80

5

10

15

20

25

30

35

pH

Support-I 3G support 6G support

(e)

Per

mea

te F

lux

x 10

5 (m

/s)

Per

mea

te F

lux

x 10

5 (m

/s)

Per

mea

te F

lux

x 10

5 (m

/s)

Fig. 10. Effect of applied pressure, concentration and pH on BSA separation (rejecti(b) concentration=500 ppm, pH=7; (c) and (d) pressure=207 kPa, pH=7; (e) and (f) c

molecule to the permeate side, which results in decreased rejection athigher pressures. The influence of feed concentration on the sep-aration performance was carried out for the concentrations rangingbetween 100 and 3000 ppm at 207 kPa. The observed rejection isfound to reduce with an increase in BSA concentration (Fig.10). Also,the rejection remained almost constant for concentrations above500 ppm for all the membrane supports. The observed rejection isvery low and doesn't show any further rejection above 500 ppm ofBSA solution.

0 50 100 150 200 250 300 3500

2

4

6

8

10

12

14

16

Applied Pressure (kPa)

Support-I 3G Support 6G Support

(b)

0 500 1000 1500 2000 2500 30000

5

10

15

20

25

30

35

40

Ob

serv

ed R

ejec

tio

n (

%)

Concentration (ppm)

Support-I 3G Support 6G Support

(d)

3 4 5 6 7 80

5

10

15

20

25

30

35

40

45

Ob

serv

ed R

ejec

tio

n (

%)

pH

Support-I 3G support 6G support

(f)

Ob

serv

ed R

ejec

tio

n (

%)

on and flux) through the membrane supports. Experimental conditions: (a) andoncentration=100 ppm, pressure=207 kPa.

Vo

lum

e ad

sorb

ed (

cc g

-1 a

t S

TP

)

0.0 0.2 0.4 0.6 0.8 1.00

5

10

15

20

0 20 40 60 80 100 120 140 160

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

Po

re V

olu

me

(cc/

g.n

m)

Pore Diameter (nm)

Support-I 3G Support 6G Support

Relative Pressure (P/P0)

(a) Support-I(b) 3G Support(c) 6G Support

(a)

(c)

(b)

Fig. 11. Nitrogen adsorption/desorption isotherms of the membrane supports. Inset shows the pore size distribution of the membrane supports.

113P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

The charge of the protein as well as the membrane can be alteredby changing the pH of the solution to create either an electrostaticattraction or repulsion between the membrane and BSA. Since theisoelectric point (IEP) of BSA is 4.9 (BSA is uncharged at pH of 4.9),BSA is positively charged when the solution pH is less than 4.9 (i.e.,pHb IEP) and negatively charged when pH is greater than 4.9 (i.e.pHN IEP). To obtain IEP of the membrane supports, the point of zerocharge (PZC) of the membrane supports is first determined accordingto the procedure reported elsewhere [46] and is found to be 8.86, 7.84and 7.59 for support-I, 3G and 6G supports, respectively. The PZC isthe same as the isoelectric point (IEP) if there is no adsorption of otherions than the potential determining H+/OH− at the surface [47–49].The obtained IEP values of the membrane supports confirm that themembrane is nearly positively charged at the pH studied in this work(Fig.10). It is clearly seen from Fig. 10 that the change in pH doesn'taffect the permeate flux and rejection significantly. This indicatesthat the separation is not dominated by the electrostatic interactionbetween BSA and membrane supports. However, a considerableamount of BSA rejection is obtained for all the membrane supports,which may be due to the presence of nanopores. To confirm this, BETisotherm was carried out for each membrane supports and thepore size distribution is reported in Fig. 11. It clearly reveals thepresence of nanopores in the membrane support, which might havean effect on BSA rejection. All the membrane supports showed atype-II adsorption/desorption curve having H1 hysteresis suggestingthe presence of nanopores and open porosity with a cylindricalgeometry [50,51]. It is to be noted that there is no change in poregeometry of the membrane supports (see Fig. 11) due to the addition/removal of TiO2/CaCO3. A higher rejection of BSA solution is obtainedfor the 3G and 6G supports than the support-I cannot be explainedwith confidence. However, the possible reasonmay be the presence ofmore number of nanosized pores present in the 3G and 6G supportsthan the support-I, which leads to an increased rejection.

5. Conclusion

The outcome of the results showed that the porosity and pore sizeof the support were decreased with the removal of TiO2. The porosityof the support-I, 3G and 6G supports was found to be 44, 38 and 36%with an average pore diameter of 0.98, 0.93 and 0.83 μm, respectively.

Flexural strength of the membrane support increases with theaddition of TiO2 and a maximum flexural strength of 33 MPa wasobtained for 6G support. Solvent permeation experiments confirmedthat all the membrane supports showed higher permeability for nonpolar solvents than the polar solvents. All the membrane supportsshowed higher rejection at lower pressure and higher feed concen-tration. However, it is advisable to operate the membrane supportwith lower concentration and applied pressure lower than 130 kPa toreduce fouling. The attempted bovine serum albumin (BSA) separa-tion using the prepared membrane supports gave a maximumobserved rejection of 40% for lower concentrations (100 ppm). Theobtained result suggests that themembrane support could be used formicrofiltration and ultrafiltration application.

Acknowledgment

Wewould like to thank the Center for Nanotechnology and CentralInstrument Facility, IITGuwahati for helping to perform XRD and SEManalysis, respectively.

Appendix A. Supplementary data

Supplementary data to this article can be found online at doi:10.1016/j.desal.2011.05.065.

References

[1] D. Bae, D. Cheong, K. Han, S. Choi, Fabrication and microstructure of A12O3/TiO2

composite membranes with ultrafine pores, Ceramics International 24 (1998)25–30.

[2] J.M. Benito, A. Conesa, F. Rubio, M.A. Rodriguez, Preparation and characterizationof tubular ceramic membranes for treatment of oil emulsions, Journal of theEuropean Ceramic Society 25 (2005) 1895–1903.

[3] G. Pugazhenthi, S. Sachan, N. Kishore, A. Kumar, Separation of chromium (VI)using modified ultrafiltration charged carbon membrane and its mathematicalmodeling, Journal of Membrane Science 254 (2005) 229–239.

[4] X. Ding, Y. Fan, N. Xu, A new route for the fabrication of TiO2 ultrafiltrationmembranes with suspension derived from a wet chemical synthesis, Journal ofMembrane Science 270 (2006) 179–186.

[5] J. Luyten, J. Cooymans, W. Adriansen, R. Leysen, Alternative process routes forbetter ceramic supports, Industrial Ceramics 20 (2000) 29–31.

[6] C. Falamaki, M.S. Afarani, A. Aghaie, Initial sintering stage pore growthmechanismapplied to the manufacture of ceramic membrane supports, Journal of theEuropean Ceramic Society 24 (2004) 2285–2292.

114 P. Monash, G. Pugazhenthi / Desalination 279 (2011) 104–114

[7] W. Zeng, L. Goa, L. Gui, J. Guo, Sintering kinetics of α-Al2O3, Ceramic International25 (1999) 723–726.

[8] Y.H. Wang, Y. Zhang, X. Liu, G.Y. Meng, Microstructure control of ceramicmembrane support from corundum-rutile powder mixture, Powder Technology168 (2006) 125–133.

[9] H. Firouzghalb, C. Falamaki, Fabrication of asymmetric alumina membranes: I.Effect of SrO addition on thermal stabilization of transition aluminas, MaterialsScience and Engineering B 166 (2010) 163–169.

[10] S. Rakib, M. Sghyar, M. Rafiq, A. Larbot, L. Cot, New porous ceramics for tangentialfiltration, Separation and Purification Technology 25 (2001) 385–390.

[11] M.R. Weir, E. Rutinduka, C. Detellier, C.Y. Feng, Q. Wang, T. Matsuura, R. LeVanmao, Fabrication, characterization and preliminary testing of all-inorganicultrafiltration membranes composed entirely of a naturally occurring sepioliteclay mineral, Journal of Membrane Science 182 (2001) 41–50.

[12] A. Potdar, A. Shukla, A. Kumar, Effect of gas phase modification of analcime zeolitecomposite membrane on separation of surfactant by ultrafiltration, Journal ofMembrane Science 210 (2002) 209–225.

[13] K. Khider, D.E. Akretche, A. Larbot, Purification of water effluent from a milkfactory by ultrafiltration using algerian clay support, Desalination 167 (2004)147–151.

[14] A. Belouatek, N. Benderdouche, A. Addou, A. Ouagued, N. Bettahar, Preparation ofinorganic supports for liquid waste treatment, Microporous and MesoporousMaterials 85 (2005) 163–168.

[15] F. Bouzerara, A. Harabi, S. Achour, A. Larbot, Porous ceramic supports formembranes prepared from kaolin and doloma mixtures, Journal of the EuropeanCeramic Society 26 (2006) 1663–1671.

[16] N. Saffaj, M. Persin, S.A. Younsi, A. Albizane, M. Cretin, A. Larbot, Elaboration andcharacterization of microfiltration and ultrafiltration membranes deposited onraw support prepared from natural Moroccan clay: application to filtration ofsolution containing dyes and salts, Applied Clay Science 31 (2006) 110–119.

[17] S. Khemakhem, A. Larbot, R.B. Amara, Study of performances of ceramicmicrofiltration membrane from tunisian clay applied to cuttlefish effluentstreatment, Desalination 200 (2006) 307–309.

[18] C. Falamaki, M. Naimi, A. Aghaie, Dual behavior of CaCO3 as a porosifier andsintering aid in the manufacture of alumina membrane/catalyst supports, Journalof the European Ceramic Society 24 (2004) 3195–3201.

[19] B. Boudaira, A. Harabia, F. Bouzerara, S. Condom, Preparation and characterizationof microfiltration membranes and their supports using kaolin (DD2) and CaCO3,Desalination and Water Treatment 9 (2009) 142–148.

[20] N. Saffaj, H. Loukili, S.A. Younssi, A. Albizane, M. Bouhria, M. Persin, A. Larbot,Filtration of solution containing heavy metals and dyes by means of ultrafiltrationmembranes deposited on support made of Moroccan clay, Desalination 168(2004) 301–306.

[21] A. Ezzati, E. Gorouhi, T. Mohammadi, Separation of water in oil emulsions usingmicrofiltration, Desalination 185 (2005) 371–382.

[22] Y. Erdem,M. Ciftcioglu, S. Harsa, Separation of whey components by using ceramiccomposite membranes, Desalination 189 (2006) 87–91.

[23] N.O. Becht, D.J. Malik, E.S. Tarleton, Evaluation and comparison of proteinultrafiltration test results: dead-end stirred cell compared with a cross-flowsystem, Separation and Purification Technology 62 (2008) 228–239.

[24] P. Srijaroonrat, E. Julien, Y. Aurelle, Unstable secondary oil/water emulsiontreatment using ultrafiltration: fouling control by backflushing, Journal ofMembrane Science 159 (1999) 11–20.

[25] T.C. Arnot, R.W. Field, A.B. Koltuniewicz, Cross-flow and dead-end microfiltrationof oily-water emulsions part II. Mechanisms andmodelling of flux decline, Journalof Membrane Science 169 (2000) 1–15.

[26] I.W. Cumming, R.G. Holdich, I.D. Smith, The rejection of oil by microfiltration of astabilized kerosene/water emulsion, Journal of Membrane Science 169 (2000)147–155.

[27] T. Mohammadi, A. Pak, M. Karbassian, M. Golshan, Effect of operating conditionson microfiltration of an oil–water emulsion by a kaolin membrane, Desalination168 (2004) 201–205.

[28] F.L. Hua, Y.F. Tsang, Y.J. Wang, S.Y. Chan, H. Chuand, H.N. Sin, Performance study ofceramic microfiltration membrane for oily wastewater treatment, ChemicalEngineering Journal 128 (2007) 169–175.

[29] M. Ebrahimi, K.S. Ashaghi, L. Engel, D. Willershausen, P. Mund, P. Bolduan, P.Czermak, Characterization and application of different ceramic membranes forthe oil-field produced water treatment, Desalination 245 (2009) 533–540.

[30] B.K. Nandi, R. Uppaluri, M.K. Purkait, Treatment of oily waste water using low-costceramic membrane: flux decline mechanism and economic feasibility, SeparationScience and Technology 44 (2009) 2840–2869.

[31] J. Zhou, Q. Chang, Y. Wang, J. Wang, G. Meng, Separation of stable oil–wateremulsion by the hydrophilic nano-sized ZrO2 modified Al2O3 microfiltrationmembrane, Separation and Purification Technology 75 (2010) 243–248.

[32] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Ultrafiltration of stable oil-in-wateremulsion by polysulfone membrane, Journal of Membrane Science 325 (2008)427–437.

[33] R. Mallada, M. Menendez, Inorganic Membranes: Synthesis, Characterization andApplications, Elsevier Science, Amsterdam, The Netherlands, 2008.

[34] Y. Pan, W. Wang, T. Wang, P. Yao, Fabrication of carbon membrane andmicrofiltration of oil-in-water emulsion: an investigation on fouling mechanisms,Separation and Purification Technology 57 (2007) 388–393.

[35] P. Monash, G. Pugazhenthi, Development of ceramic supports derived from lowcost raw materials for membrane applications and its optimization based onsintering temperature, International Journal of Applied Ceramic Technology8 (2011) 227–238.

[36] A. Sarkar, R. Rano, K.K. Mishra, I.N. Sinha, Particle size distribution profile of someIndian fly ash — a comparative study to assess their possible uses, Fuel ProcessingTechnology 86 (2005) 1221–1238.

[37] A. Aimable, M.T. Buscaglia, V. Buscaglia, P. Bowen, Polymer-assisted precipitationof ZnO nanoparticles with narrow particle size distribution, Journal of theEuropean Ceramic Society 30 (2010) 591–598.

[38] A. Gualtieri, M. Bellotto, G. Artioli, S.M. Clark, Kinetic study of the kaolinite –

mullite reaction sequence. Part II: mullite formation, Physics and Chemistry ofMinerals 22 (1995) 215–222.

[39] G. Cultrone, C.R. Navarro, E. Sebastian, O. Cazalla, M.J. Torre, Carbonate and silicatephase reactions during ceramic firing, European Journal of Mineralogy 13 (2001)621–634.

[40] P. Monash, A. Majhi, G. Pugazhenthi, Separation of BSA using γ-Al2O3-claycomposite ultrafiltration membrane, Journal of Chemical Technology andBiotechnology 85 (2010) 545–554.

[41] D. Bhanushali, S. Kloos, C. Kurth, D. Bhattacharyya, Performance of solvent-resistant membranes for non-aqueous systems: solvent permeation results andmodeling, Journal of Membrane Science 189 (2001) 1–21.

[42] J. Chen, J. Li, X. Zhan, X. Han, C. Chen, Effect of PEG additives on properties andmorphologies of polyetherimide membranes prepared by phase inversion,Frontiers of Chemical Engineering in China. 4 (2010) 300–306.

[43] D.R. Machado, D. Hasson, R. Semiat, Effect of solvent properties on permeate flowthrough nanofiltration membranes. Part I: investigation of parameters affectingsolvent flux, Journal of Membrane Science 163 (1999) 93–102.

[44] C. Guizard, A. Ayral, A. Julbe, Potentiality of organic solvents filtration withceramic membranes. A comparison with polymer membranes, Desalination 147(2002) 275–280.

[45] S.R. Chowdhury, R. Schmuhl, K. Keizer, J.E. Elshof, D.H.A. Blank, Pore size andsurface chemistry effects on the transport of hydrophobic and hydrophilicsolvents through mesoporous γ-alumina and Silica MCM-48, Journal ofMembrane Science 225 (2003) 177–186.

[46] S.Wang, G.Q. Lu, Effects of acidic treatments on the pore and surface properties of Nicatalyst supported on activated carbon, Carbon 36 (1998) 283–292.

[47] G.D. Parfitt, Surface chemistry of oxides, Pure and Applied Chemistry 48 (1976)415–418.

[48] R.M.T. Sanchez, E.L. Tavani, Temperature effects on the point of zero charge andisoelectric point of a red soil rich in kaolinite and ironminerals, Journal of ThermalAnalysis and Calorimetry 41 (1994) 1129–1139.

[49] M. Alvarez-Silvaa, A. Uribe-Salasb, M. Mirnezami, J.A. Finch, The point of zerocharge of phyllosilicate minerals using the Mular–Roberts titration technique,Minerals Engineering 23 (2010) 383–389.

[50] R.J.R. Uhlhorn, M.H.B.J. Huis in't Veld, K. Keizer, A.J. Burggraaf, Synthesis of ceramicmembranes. Part I synthesis of non-supported and supported γ-aluminamembranes without defects, Journal of Materials Science 27 (1992) 527–537.

[51] E.R. Leite, J.A. Cerri, E. Longo, J.A. Varela, Sintering of undoped SnO2, Ceramica 49(2003) 87–91.