Dm

Ra

b

a

ARRAA

KSMC

1

tttbsaduildwr

C

h2l

Journal of Asian Ceramic Societies 4 (2016) 243–251

Contents lists available at ScienceDirect

Journal of Asian Ceramic Societies

HOSTED BY

j ourna l ho me page: www.elsev ier .com/ loca te / jascer

evelopment of hydrophobic clay–alumina based capillaryembrane for desalination of brine by membrane distillation

akhi Dasa, Kartik Sondhib, Swachchha Majumdara, Sandeep Sarkara,∗

Ceramic Membrane Division, CSIR-Central Glass and Ceramic Research Institute, Kolkata 700032, IndiaChemical Engineering Department, Jadavpur University, Kolkata, India

r t i c l e i n f o

rticle history:eceived 14 January 2016eceived in revised form 11 April 2016ccepted 15 April 2016vailable online 13 May 2016

eywords:urface modificationembrane distillation

eramic hydrophobic

a b s t r a c t

Clay–alumina compositions of 0, 20, 40 and 55 weight percent (wt%) clay and rest alumina were main-tained in porous support preparation by extrusion followed by sintering at 1300 ◦C for 2.5 h to obtain3 mm/2 mm (outer diameter/inner diameter) capillary. 1H,1H,2H,2H-perfluorodecyltriethoxysilane(97%) (C8) was used to modify the capillary surface of all compositions without any intermediate mem-brane layer to impart hydrophobic characteristics and compared in terms of contact angle produced bythe capillaries with water and liquid entry pressure (LEPw). FTIR analysis showed that the hydrophilicsurface of the capillary membranes was efficiently modified by the proposed grafting method. Capillarywith 55 wt% clay produced a pore size of 1.43 micron and was considered as an ideal candidate for graftingwith C8 polymer to impart surface hydrophobicity. The contact angle and LEPw value obtained for thismodified membrane (C-55-M) were 145◦ and 1 bar, respectively. The modified capillary membrane was

applied for desalination of brine by air gap membrane distillation (AGMD) at a feed pressure of 0.85 bar.Maximum flux obtained for C-55-M membrane was 98.66 L/m2 day at a temperature difference of 60 ◦Cwith salt rejection of 99.96%. Mass transfer coefficient of C-55-M was 16 × 10−3 mm/s at feed temperatureof 70 ◦C.

© 2016 The Ceramic Society of Japan and the Korean Ceramic Society. Production and hosting byElsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/

. Introduction

The ocean represents an endless water resource. Despite that,here is lack of potable water, which raises the need for sus-ainable technologies to produce fresh potable water. Consideringhat, attention has been given on desalination of sea water andrackish water to produce fresh water by various techniques,uch as reverse osmosis (RO), electro-dialysis, distillation, etc. ROnd electro-dialysis are membrane-based separation processes foresalination of brine [1,2]. RO is one of the most effective and pop-lar membrane-based techniques for desalination, but the process

s very energy intensive [3] and economically not viable, particu-arly in the underdeveloped and developing countries. Membraneistillation (MD) is very low energy intensive distillation process,hich has advantages over conventional desalination process [4]. It

equires less surface area per unit volume than conventional distil-

∗ Corresponding author. Tel.: +91 9432849210.E-mail address: sandeepsarkar123@gmail.com (S. Sarkar).Peer review under responsibility of The Ceramic Society of Japan and the Korean

eramic Society.

ttp://dx.doi.org/10.1016/j.jascer.2016.04.004187-0764 © 2016 The Ceramic Society of Japan and the Korean Ceramic Society. Producti

icense (http://creativecommons.org/licenses/by-nc-nd/4.0/).

licenses/by-nc-nd/4.0/).

lation [5] and can be clubbed with renewable energy source owingto low energy consuming process [4].

MD is a thermal, vapour driven transportation process throughhydrophobic organic/inorganic membrane, as shown in Fig. 1. Thedriving force of MD is created by the difference of vapour pressure,resulting from the temperature difference between the feed andpermeate side [6–10]. Brine is heated to generate vapour pressure,which creates a partial pressure difference across the membrane.A hydrophobic membrane is permeable only to vapour and notto liquids [11]. Hot water evaporates through the pores of thehydrophobic membrane, as shown in Fig. 1, leaving behind thebrine (liquid) as a retentate, which is impermeable through thepores of the hydrophobic membrane [11].

The permeating vapour is then condensed to produce freshwater [12–14]. The technique is highly efficient in salt rejectionand the rejection rates are around 99–100%. It is an efficient tech-nique for treating highly concentrated brine, which is of particularinterest, comparing with RO, where the osmotic pressure increases

with the salt concentration [15].

Generally, all the ongoing MD processes are based on com-mercially available hydrophobic polymeric membrane [16,17] butceramic membrane has advantage over polymeric membrane due

on and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND

244 R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251

Permeate in vapour phase

Cooling Jacket

Conden sed Permea te

Porous

Feed

T1

T2

Coolin g water

T1

T2

Feed water temp.

Cooling water temp.

tltoMltmswhtp

oabchiatioimotarti

ia[vstsptglohmtAwt

Table 1Formulation of clay alumina powder in different membranes.

Sl. No. Membrane ID Clay (wt%) Alumina (wt%)

1 C-0 0 1002 C-20 20 80

per minute (rpm) for 24 h to remove the free water from samplebefore sintering it at 1300 ◦C for 2.5 h as per sintering schedule illus-trated in Fig. 2, to have a final capillary diameter of 3 mm OD and

Membrane

Fig. 1. Principal of air gap membrane distillation.

o its intrinsic properties, such as high mechanical strength and longife [18]. Also polymeric membranes are subjected to fouling morehan ceramic membranes. Thus ceramic membranes have numer-us advantages over polymeric membranes but cannot be used forD process as such, since they are hydrophilic in nature. Another

imitation of ceramic membrane is that they have high membranehickness, which increases the membrane resistance and thereby

aking it not a suitable candidate for the MD process [19]. Theurface of the ceramic membrane needs to be modified by graftingith a suitable polymer to impart hydrophobicity [20]. To impartydrophobic characteristics on the hydrophilic ceramic substrates,he metal oxide substrates are modified by various hydrophobicolymers, as reported in literature [4,21–25].

During grafting process, the hydroxyl groups (–OH) of the metalxide surface of ceramic membrane reacts with polymer solutionnd forms a stable covalent bond [26] and behave like hydropho-ic in nature. Thus, ceramic membranes become hydrophobic andan be used for MD process. Another important characteristic of aydrophobic membrane required for MD process is the resistance

t offers to liquid (water) to enter its pores. This property is char-cterised by pressure required by liquid to enter its pores and isermed as liquid water entry pressure (LEPw) measurement. It ismportant to maintain a required LEPw for effective MD. Pore sizef a hydrophobic membrane plays a predominant role in maintain-ng high LEPw. The relation between pore size of a hydrophobic

embrane and the LEPw is inversely proportional consideringther similar factors. Mulder [27] and Drioli et al. [28] suggestedhat a membrane with pore size in the range of 0.2–0.3 micronnd 0.2–1 micron respectively is ideal for such applications. Buteducing the pore size will also increase the membrane resistance,hereby reducing the membrane efficiency. An optimum pore sizes an important parameter for proper MD application.

Most of the ceramic membranes used for surface modification tompart hydrophobic character for MD process are of commerciallyvailable expensive membrane made of zirconia [29] and titania30] over macro-porous alumina support. These membranes havearious intermediate layer of different pore size in between theupport and the top membrane layer in order to successively reducehe pore size. The top membrane layer made of zirconia has a poreize of 50 nm [15], titania has 5 nm [31] and � alumina has 5 nmore size membrane layer over macro-porous � alumina supportube [21]. The support tube size used for surface modification, ineneral, is of 10 mm outer diameter and 1.5 mm thickness with aength of 120 mm [32] and 150 mm [15]. These supports are madef high purity alumina with high processing cost [32] as it sinters atigh temperature (1700 ◦C) [33]. Moreover, preparation of multipleicro-porous membrane layer over such support tube increases

he membrane thickness, along with the overall membrane cost.lso the reduction of pore size and increase in membrane thicknessill increase the membrane resistance, which eventually reduce

he membrane efficiency.

3 C-40 40 604 C-55 55 45

This study involves surface modification of the indigenouslyprepared macro porous clay alumina based membrane withoutany intermediate layer to impart hydrophobic characteristic. Themacro porous clay alumina based membrane was prepared in capil-lary configuration with less than 0.5 mm wall thickness in an effortto reduce the membrane resistance and thereby increase its effi-ciency. The membrane was applied for desalination of brine by airgap membrane distillation (AGMD) process. The results were com-pared with derived model and the published results of differentmodified ceramic membrane for overall MD performance.

2. Materials and methods

2.1. Materials

Alpha alumina (99% purity) with a mean particle size of7 microns was purchased from Hindalco, India, and kaolinite claywith a mean particle size of 10 micron was procured locally fromKolkata, India. Methocel, used as organic binder for preparationof porous ceramic membrane, was purchased from Dow Chemi-cal, USA. 1H,1H,2H,2H-perfluorodecyltriethoxysilane (97%), whichhas a chemical formula of C8F17C2H4Si(OC2H5)3(C8) was used asgrafting agent (Sigma Aldrich, USA), ethanol (C2H5OH) and sodiumchloride (NaCl) were purchased from Merck, Germany.

2.2. Preparation of macro-porous ceramic capillary membrane

The capillary membranes were prepared from clay and aluminaceramic pastes with different compositions similar to the previouswork [34]. Alumina was partially substituted by natural mineralclay and used as basic raw material. Clay and alumina were mixedwith different weight percent (Table 1) along with binders (metho-cel) and distilled water (18 M� cm at 25 ◦C). In each composition,96 wt% of the mixed raw clay–alumina powder was mixed with4 wt% methocel and then 20–25 ml water was added to the totalmixture of 100 g to obtain an extrudable paste.

The pastes were then extruded in a plunger type extrusionmachine through a capillary dye to prepare green tube of 3.5 mmouter diameter (OD) and 0.5 mm thickness. The extruded tubeswere dried at room temperature in a roller drier at 10 revolution

Fig. 2. Firing schedule of green clay alumina capillary.

R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251 245

F3

0t

2

dhmfl

iwaapiptaAoh

2

i(

ig. 3. Chemical structure of fluoroalkylsilane (C8) molecule (a) 2D structure, (b)D structure.

.45 mm thickness with various pore characteristics depending onhe composition as described in Section 5.1.3.

.3. Surface modification of ceramic capillary membrane

Fluoroalkylsilane (C8) was considered as a very successful can-idate for modifying the metal oxide surface [4,22–25] to impartydrophobic characteristic by grafting process. The general for-ula of FAS is R-C2H4-Si(R1)R3, where R represents the chain of

uorocarbon and R1 as methyl ethoxy or chlorine group (Fig. 3a).The membranes were cleaned by acetone for 10 min and dried

n oven at a temperature of 110 ◦C for 12 h before grafting themith C8. A C8 solution was prepared by mixing C8 with ethanol

t a concentration of 10−2 mol of C8 per litre of ethanol in inerttmosphere because a poly-condensation reaction of 1H,1H,2H,2H-erfluorodecyltriethoxysilane occurs due to presence of moisture

n air [35]. Four capillary membranes of 150 mm length were com-letely immersed and soaked in 45 ml C8 solution taken in a glassube (50 ml culture tube) and gently rolled in an eccentric rollert a speed of 30 rpm for 75 h to carry out the grafting process.fter grafting, the membranes were dried at 105 ◦C for 12 h in anven. The process was repeated 3 times to obtain a non-wettingydrophobic membrane.

.4. Characterisation methods

The porosity and pore size of the fired membranes (shownn Table 1) were measured by mercury intrusion porosimetryMicromeritics, AutoPore IV 9500).

Fig. 5. (a) Schematic diagram of membrane

Fig. 4. Single capillary module of ceramic hydrophobic membrane.

Fourier transform infrared spectroscopy (FTIR) analysis (Perkin-Elmer Spectrum 200 instrument) was used to determine thepresence of per-fluorinated groups on membrane surface aftergrafting. The analysis was done with the wavelength range of400–1400 cm−1.

Contact angle (CA) measurement has been performed on mem-branes to evaluate its surface hydrophobicity. CA measurement ofthe sample was performed by sessile drop method using Kruss(Germany) apparatus at room temperature. All CA readings weretaken 15 min after 0.5 ml water droplet was placed on the mem-brane surface.

The efficiency of hydrophobic membranes was determined bymeasuring the pressure at which water penetrates through themembrane pores (LEPw). The experiments were carried out bysubjecting the membrane tube side with water at a constant par-ticular pressure for at least 2 h and checked for any water dropletappearing at the membrane surface (shell side) before increasingthe pressure further. The process is continued until the first waterdroplet appeared in the membrane shell side surface. LEPw wasmeasured by cross-flow filtration using a homemade pilot withapplied N2 pressure as described in Mohammadi and Safavi [12].

2.5. Membrane distillation

Air gap membrane distillation (AGMD) was carried out withthe prepared ceramic hydrophobic capillary membrane of 150 mmlength. The MD was performed at different feed temperatureand temperature difference (�T) across the membrane. Feedsolutions for MD have been prepared by using distilled water(18 M� cm) and salt (NaCl) similar to the sea-water salt concen-tration (0.5 mol/L), which corresponded to the total dissolved solid(TDS) of 1912.476 ppm (parts per million). Permeate flux was cal-culated by weighing the volume of liquid permeate that comeswithin a fixed time interval during experiment. Permeate flux cal-

culation was started when the MD process reached at steady state,i.e. after 30 min from experiment initiation. Rejection of salt andother impurities was analysed by measuring total dissolved solid(TDS) in feed and in permeate. The TDS rejection (RNaCl) during the

distillation set-up, (b) actual set-up.

246 R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251

metry

Md

R

icaf

cfbcimlc

3

ot(

wokbm

p

U

stwrfl

Fig. 6. Flow geo

D process was calculated according to Eq. (1), where Cp and Cfenote the TDS in permeate and feed solution [24,29].

NaCl =(

1 − Cp

Cf

)× 100 (1)

AGMD was performed with different feed temperatures, start-ng from 40 ◦C to a maximum of 70 ◦C. Ceramic hydrophobic singleapillary module (Fig. 4) with 6.78 × 10−2 m2 effective surface areand 2 mm air gap has been indigenously designed and fabricatedor MD process along with a MD experimental set-up (Fig. 5).

Hot brine will be pumped to the MD module, which housed theeramic hydrophobic capillary membrane from the feed tank by theeed pump at a predetermined flow rate. The module inner wall iseing kept at a temperature lower than the feed temperature byirculating cold water in the module jacket from the chiller. Thiss to ensure that the temperature of the air gap in between the

embrane surface (shell side) and the inner module wall becomesower than the feed temperature, for condensing the vapour, whichomes out of the membrane surface in the module.

. Model development and derivation

The flow geometry of MD module was illustrated in Fig. 6, basedn which models were developed. The overall convective heatransfer coefficient for the outer wall of membrane is given by Eq.2),

1U

= 1hmern

+ d1

d2hfeed+ 1

hcooling water+ x

k(2)

Where U is overall convective heat transfer coefficient for outerall of membrane and hmem is convective heat transfer coefficient

f the inner wall of membrane. Here, d2, d1, hcooling water, x, and are the outer diameter of membrane, inner diameter of mem-rane, heat transfer coefficient of the cooling water, thickness of theembrane, and thermal conductivity of membrane, respectively.Substituting the value of U from Eq. (2), the mass of vapour

ermeating from the membrane can be calculated from Eq. (3).

AdT = mCp(Tboiling − Tf (in)) + �m′ (3)

Where m, Cp, Tboiling, Tf(in), �, and m′ are the mass of the feedtream, specific heat capacity of the feed, boiling temperature of

he feed, temperature of the inlet feed stream, latent heat of theater vapour, and mass of vapour permeating from the membrane,

espectively. Solving Eq. (3), value of m′ was obtained. The permeateux has been calculated from the value of m′.

of MD module.

The heat flux along the surface of membrane is represented byEq. (4) [36].

Q = mCp(Tf (out) − Tf (in)) (4)

Where, Q is the total heat flux along distance z and Tf(out) is thetemperature of the outlet stream from the membrane.

For the calculation of the amount of heat transfer along the radialdirection (from the cooling jacket surface to the air gap interface inMD module), we consider a cylindrical coordinates system for thetubular membrane which has a unidirectional flow along the radialdirection, as shown in Eq. (5).

∂T

∂t+ vr

∂T

∂r+ v�

r

∂T

∂�+ vz

∂T

∂z

= K

[1r

∂∂r

(r∂T

∂r

)+ 1

r2

∂2T

∂�2+ ∂2

T

∂z2

]+ hmem

�0cp(5)

The other components of the equation from the cylindrical coor-dinate system (Fig. 7) have been cancelled out as the variation alongthe � and z are not observed and also the velocity along any direc-tion has not been considered. As there is no internal heat sourcewithin the system, the term hmem/(�0cp) = 0. Therefore for heat bal-ance, the equation in cylindrical coordinate system can be derivedas Eq. (6) [37].

dT

dt= K

r

(d

dr

(r

dT

dr

)+ d2T

dz2

)(6)

Simplifying it for z, as thickness of the membrane is very smallconsidered to length of the membrane. Thus, we have an averagingcomponent for the temperature observed in the membrane, whichis denoted by T .

The final form, by making the time derivative Eq. (7), has beenobtained.

dT̄

dt= K

(2R

(T̄ − Tsv) + d2T̄

dz2

)(7)

In the above equation, r represents the direction in radial posi-tion as shown in Fig. 6.

The boundary conditions have been given as B.C. 1, 2, 3 andare in accordance with the temperature variation which should beobserved in the membrane.

B.C. 1: t = 0, T = Tf(in)

B.C. 2: r = R1, T = T̄B.C. 3: at r = R2, T = Tsv

R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251 247

geom

Tfws

b

v

sp

m[

�

dAmRat

4

S(N

D

smt

Fig. 7. Representation of axial

Tsv is the temperature of the air gap interface in MD module.here is a negligible temperature difference between Tf(in) and Tf(out)or our system (as the membrane length was small and flow rateas 0.492 L/min), so the heat transfer along z direction is not con-

idered here.The velocity variation along the radial direction (vr) can be given

y Eq. (8).

r = vF − �PL

�(

ln(

R22−R2

1R−R2

1

)) (8)

Variation of velocity along the radial direction depends on pres-ure drop (�P) between any two points (R1 and R2) along the radialosition of the membrane, where � is the viscosity of the feed.

For the calculation of NaCl rejection percentage, we consider theass balance equation along the radial direction based on Eq. (9)

37].(dVr

dt+ dVr

dr

)= dP

dr− 1

r

drVr

dr(9)

Boundary conditions for Eq. (9) are B.C. 4, 5, 6.

B.C. 4: t = 0, � = �0, v = vF,B.C. 5: r = R1, � = �, v = vF,B.C. 6: r = R2, � = �, v = vr,

In the above equation, �0 is the density of feed stream, � is theensity of the permeating vapour, and vF is the feed stream velocity.fter solving Eqs. (8) and (9) simultaneously, a two dimensionalatrix is obtained for Eq. (9) by taking R1 as the inner radius and

2 as the outer radius of the membrane. Concentration at radius R1nd R2 was calculated, and from the difference in concentration,he rejection percentage was obtained.

. Mass transfer coefficient

The mass transfer coefficients have been calculated using theherwood number. The Wilke Chang equation for dilute solutionsEq. (10)) [38] was used first for calculation of diffusivity (D) in theaCl solution.

= 7.40 × 10−8 × (ϕM)1/2 × T

V0.6(10)

Where ϕ is the association parameter of solvent, which in thistudy has been considered to be water, the value is 2.6 [39], M is theolecular weight of the solvent and has been considered as 18, T is

he temperature in K at which the experiment is being conducted

etry considered for the study.

(feed temperature), is the viscosity of the solution at the particulartemperature.

The molar volume (V) was calculated using Eq. (11).

V = M

�(11)

Here M is the mass of the NaCl solution and � is the density of thesolution.

This diffusivity (D) was derived as in the method described bySaltzman et al. [39]. Values were substituted in Eq. (12) to evaluatethe Sherwood number (Sh).

Sh = 0.646 ×(

�dv

)0.5

×(

�D

)0.33(12)

Where d is the pore diameter, v is considered to be the flux inthis case.

Based on the above, the mass transfer coefficient was calculatedby Eq. (13).

K = Sh × D

L(13)

Where L is the length of the membrane. The final mass transfercoefficient values are presented as in Table 3.

Sh = 0.646 ×(

�dv

)0.5

×(

�D

)0.333

5. Results and discussion

5.1. Pore characteristics of membrane and their effect on surfacemodification

5.1.1. Pore characteristics of membraneFig. 8 revealed that the average pore diameter and porosity grad-

ually reduce with the increase in clay content in the membranecomposition.

The silica present in clay (Al2O3–2SiO2–2H2O) acts as a lowmelting phase during the sintering process and starts reacting withits own structural Al2O3 at a temperature lower than 1000 ◦C toform mullite (2Al2O3–SiO2) [34]. Stoichiometrically the excess sil-ica present in clay does not react with the alumina mixed externally[40]. Instead, it transforms into glassy phase and stays in betweenthe voids, which eventually reduces the pore diameter and porosity[34].

5.1.2. FTIR analysisCapillary membranes were grafted with C8 solution keeping

all the parameters similar as discussed earlier in Section 2.3. The

248 R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251

C-0 C-20 C-40 C-551.0

1.5

2.0

2.5

3.0

Por

osi

ty (

%)

Por

e di

a (m

icro

met

er)

(Pore dia)

30

32

34

36

38

40

(Porosity)

Fig. 8. Pore diameter and porosity graph for different formulations of membrane.

12001400 10 00400600800

no n Grafted C-0

1CFx

}

Si-CH2CH

2C

xF

2x+1

Grafted C-0

FaIr

iagtCata

400600800100012001400

Wavelength (cm- 1)

n on g raft ed C-40

2 1

Si-O-S i

{

Si-CH2CH2CxF2x+1

C F2

{ gra fted C-40

ting multiple silanol bond with the polymer and ceramic surfacethereby imparting better hydrophobicity to the membrane. Thishas been confirmed by gradual increase in CA values with decrease

Wavelength(cm-1)

Fig. 9. FTIR spectra of C-0.

TIR spectrum of both grafted and non-grafted C-0, C-20, C-40nd C-55 membranes is represented in Figs. 9–12, respectively.n the spectrum, x-axis represents wavelength in cm−1 and y-axisepresents percentage transmission.

The FTIR spectra of C-0 and C-20 grafted membrane at zone 1n Figs. 9 and 10, respectively revealed two peaks at 1203 cm−1

nd 1205 cm−1. Peak 1203 cm−1 attributes to Si-CH2CH2CxF2x+1roup [31] and 1205 cm−1 attributes to CFx group [23], wherehere is no transmission peak near 1200 cm−1 for non-grafted

-0 and C-20. For grafted C-40 and C-55 membranes, two zonesre shown in Figs. 11 and 12. In these figures, zone 1 has tworansmission peaks at 1203 and 1205 cm−1. Zone 2 has two peakst 1115 and 1120 cm−1. Transmission bands 1115 cm−1 attributes

1Si-CH2CH2CxF2x +1

CF2

{ grafted C-20

non grafted C-20

140 0 120 0 100 0 800 60 0 40 0

Waveleng th(cm-1)

Fig. 10. FTIR analysis of C-20.

Fig. 11. FTIR analysis of C-40.

to Si–O–Si group and 1120 cm−1 attributes to Si-CH2CH2CxF2x+1group. The non-grafted spectrum of C-40 shows no peak near1200 and 1115 cm−1. In non-grafted C-55 a transmission bandappeared at 1175 cm−1, which may be due to the presence ofmullite as also observed by Saikia and Parthasarathy [41]. In all thegrafted membranes, the presence of Si-CH2CH2CxF2x+1 group is acommon, which has formed due to the chemical reaction betweenSi(OC2H5)3 and the surface –OH groups of ceramic membrane.This confirms the anchoring of hydrophobic polymer (C8) with themembrane surface (Fig. 13).

5.1.3. Effect of membrane pore characteristics and porosity ontheir surface modification

During grafting process, the polymer solution (C8) reacts withthe hydroxyl groups present on metal oxide surface and constructa Si–O covalent bond with the –OH groups (Fig. 14) as described byKujawa et al. [25].

Membrane containing smaller pore size corresponds to a densersurface and lesser void space on its surface. Availability of moresurfaces implies the increase in number of available –OH group onthe membrane surface for bonding. This may encourage construc-

400600800100012001400

21Si-CH 2 CH 2CxF2x+ 1

S-O-iS i

C F2

Wavelength(cm-1 )

n on gr a fted C -5 5

}}

g ra fted C-55

Fig. 12. FTIR spectra of C-55 (C-55-M).

R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251 249

115 120 125 130 135 140 145

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

3.0po

re s

ize(

mic

ron)

CA(degree )

po re size

34

35

36

37

38

39

C-0

C-20

C-40C-55

po rosity

poro

sity(

%)

Fig. 13. Variation of contact angle with different membranes pore size and porosity.

F

ic

5

bbe

�

ra

ewaf

5

bwcssiwb

1.

0 .2

0 .4

0 .6

0 .8

1 .0

LE

P w(b

ar)

.4 1 .6 1 .8P o r

2.0 2 .2 2 .4r e s iz e (m ic r om et er

Ex p er im e nt T he or i tic al

2.6 2 .8 3

r )

al

3 .0

Fig. 15. Variation of LEPw with different membranes pore size.

0

20

40

60

80

100

Flux

(l/m

2 .day

)

20 3 0

Tem p

E xpe rim ental S imula ted

40

perature d ifferenc e(

50 60

(0C)

ig. 14. Schematic of grafting reaction between FAS molecule and –OH group.

n membrane pore size (Fig. 13). Similar phenomenon was alsoonfirmed by Pinheiro et al. [42].

.1.4. LEPw measurement of the modified capillary membraneIt was revealed that LEPw increases with the decreases in mem-

rane pore size. Based on the Laplase (Cantor) Equation, the relationetween LEPw and pore size of a hydrophobic membrane may bexpressed by Eq. (14) [43].

P = 2�L

rp,maxcos � (14)

Where �P is the LEPw, �L is the surface tension of the liquid,p,max is the largest pore size of the membrane and � is the contactngle.

Fig. 15 represents the comparison between theoretical andxperimental value of LEPw. A variation in LEPw value was observedith higher clay composition modified membrane. Considering the

bove modified C55 capillary membrane (C-55-M) was selected forurther MD application.

.2. Membrane distillation

Performance of developed C-55-M membrane was evaluatedased on permeate flux value and percentage salt rejection. Fluxas calculated in terms of L/m2 day. Salt rejection percentage cal-

ulation has been discussed in Section 2.5. The experimental andimulated fluxes with respect to temperature difference at a con-

tant pressure of 0.85 bar and flow rate of 0.492 L/min are illustratedn Fig. 16. Both the experimental and simulated flux values increase

ith increasing temperature difference, which was also observedy Kujawa et al. [44]. In practical application the MD process runs

Fig. 16. Relation between MD flux and temperature difference (Tf(in) − Tf(out)).

continuously so the process needs a membrane with low foulingcharacteristic. The experimental value of flux is within ±5% devia-tion which attributes that the membrane has no cracks or foulingduring the MD process.

Rejection rate is a function of total grafting time and tempera-ture difference [35]. In this work, grafting time was 75 h for C-55-Mmembrane and was not varied. Hence the exhibited rejection rateis only a function of temperature difference. A very small deviationfrom experimental results was observed with the simulated results(Fig. 17).

Fig. 17. MD flux vs. rejection rate.

250 R. Das et al. / Journal of Asian Ceramic Societies 4 (2016) 243–251

Table 2Comparison of C-55-M with published results.

Membrane used Pore size/MWCO Temp difference (◦C) Flux (L/m2 day) Method used for calculation of rejection Rejection (%)

Zr50 membrane [15](2009)

50 nm 75 65 – 99.885 84 99.595 113 99.8

Ti-t-C6 membrane [35](2013)

300 KD 65 13.44 By ion chromatography 92.375 46.008 99.585 57.744 99.8

Clay alumina macroporous membrane(C-55-M) (2015)

1.43 �m 20 5.4864 By TDS metre 99.140 28.612 99.950 50.13 99.9560 98.66 99.96

Table 3Mass transfer coefficient and diffusivity data for C-55-M membrane for 150 mm length.

Sl. No. Flux (L/m2 day) Temperature (◦C) Density (kg/m3) Specific volume(m3 kg)

Diffusivity (m2/s) Mass transfercoefficient (mm/s)

1 5.4864 40 8.584 2.096 23.5 × 10−5 20.38 × 10−4

2 28.6124 50 17.168 1.048 35.7 × 10−5 68.73 × 10−4

3 50.139 60 21.46 0.838 40.8 × 10−5 10.31 × 10−3

5p

wtf9ttComd

5

ad2caRih(

6

mgsmcoaswm

[

[

[[[[

[16] M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel, Desalination, 1, (4)435–449 (2005).

4 98.667 70 25.752

.3. MD performance comparison of C-55-M membrane withublished result

The performance of C-55-M membrane has been comparedith the modified zirconia [15] and titania [35] membrane. In

erms of flux and percentage rejection the maximum flux obtainedor zirconia membrane was 113 L/m2 day with rejection rate of9.8% and temperature difference of 95 ◦C. For titania membrane,he flux was 57.7 L/m2 day with rejection rate of 99.8% at 85 ◦Cemperature difference whereas the maximum flux obtained for-55-M membrane was 98.66 L/m2 day at a temperature differencef 60 ◦C with a rejection rate of 99.96% (Table 2). This enhance-ent in flux of C-55-M membrane with relative low temperature

ifference is may be due to the presence of larger pore size.

.4. Result of mass transfer coefficient

Mass transfer coefficient increases with temperature differencend flux value of C-55-M membrane. Mass transfer coefficient isirectly proportional with diffusivity and Sh as discussed in Section.5. Diffusivity depends on temperature; therefore, mass transferoefficient will change according to variation of temperature. Sh is

function of Reynolds number (Re) and Schmidt number (Sc). Bothe and Sc depend on flux value. So value of mass transfer coefficient

ncreases with increase in flux value. C-55-M membrane shows theighest value of mass transfer coefficient at a temperature 70 ◦CTable 3).

. Conclusion

Indigenously developed clay alumina based C-55 capillaryembrane (C-55-M) was developed for surface modification by

rafting with C8 polymer. Grafting of polymer on the membraneurface was successfully carried out directly on the micro-porousembrane with 1.43 micron pore size without any intermediate

oating layer. FTIR analysis revealed the siloxane bond presentn the grafted membrane surface. A gradual increase in contact

ngle values was observed with decrease in membrane poreize. It was also observed that LEPw of the modified membraneas increased with the decrease in membrane pore size. C-55-Membrane has a contact angle of 145◦ and 1 bar of LEPw. MD has

[[

[

0.698 45.5 × 10−5 16.03 × 10−3

been carried out at a feed pressure of 0.85 bar. The maximumflux obtained for C-55-M membrane was 98.66 L/m2 day at atemperature difference of 60 ◦C with a salt as well as the tap waterimpurities rejection of 99.96%. C-55-M membrane has the masstransfer coefficient of 16.03 × 10−3 mm/s at the feed temperatureof 70 ◦C. The experimental flux value is within ±5% deviation fromthe theoretical values, which attributes that the membrane hasno cracks or fouling during the MD process and can be applied forcontinuous MD process.

Acknowledgements

The work has been fully funded by DST first track project GAP0341 and partially funded by CSIR 12th five year plan project CSC0115.

References

[1] S.T. Hsu, K.T. Cheng and J.S. Chiou, Desalination, 143, 279–287 (2002).[2] E. Brauns, V. Van Hoof, C. Dotremont, H. De Wever, P. Lens, E. Van Hoof, G.

Thomas, B. Molenberghs and D. Demey, Desalination, 170, 123–136 (2004).[3] R. Dashtpour and S.N. Al-Zubaidy, Int. J. Environ. Sci. Dev., 3, (4) 340–345 (2012).[4] L. Gazagnes, S. Cerneaux, M. Persin, E. Prouzet and A. Larbot, Desalination, 217,

260–266 (2007).[5] L.M. Camacho, L. Dumee, J. Zhang, d.J. Li, M. Duke, J. Gomez and S. Gray, Water,

5, 94–196 (2013).[6] C.A. Rivier, M.C. Garcia-Payo, I.W. Marison and U. Von Stockar, J. Membr. Sci.,

201, 1–16 (2002).[7] R.W. Schofield, A.G. Fane and C.J.D. Fell, J. Membr. Sci., 33, 299–313 (1987).[8] R.W. Schofield, A.G. Fane and C.J.D. Fell, J. Membr. Sci., 53, 173–184 (1990).[9] M. Tomaszewska, Desalination, 104, 1–11 (1996).10] M. Tomaszewska, M. Gryta and A.W. Morawski, J. Membr. Sci., 166, 149–156

(2002).11] M. Khayet, Desalination and Water Resources, Encyclopedia of Life Support

Systems, Eoless Publishers (2010).12] T. Mohammadi and M. Safavi, Desalination, 249, (1) 83–89 (2009).13] N. Tang, Q. Jia, H. Zhang, J. Li and S. Cao, Desalination, 256, (1–3) 27–36 (2010).14] M. Khayet, Adv. Colloid Interface Sci., 164, (1–2) 56–88 (2011).15] S. Cerneaux, I. Struzynska, W.M. Kujawski, M. Persin and A. Larbot, J. Membr.

Sci., 337, 55–60 (2009).

17] M. Tomaszewska, Polish J. Environ. Stud., 9, (1) 27–36 (2000).18] K. Guerra and J. Pellegrino, Ceramic membranes for water treatment applica-

tions, Reclamation managing water in west, Research and Development Office,U.S. Department of interior, Bureau of Reclamation (2013).

19] M. Tomaszewska, Polish J. Environ. Stud., 9, (1) 27–36 (2000).

Ceram

[

[

[

[

[

[

[

[

[[

[[

[[

[

[

[

[

[[

[[

R. Das et al. / Journal of Asian

20] W. Kujawski, S. Krajewska, M. Kujawski, L. Gazagnes, A. Larbot and M. Persin,Desalination, 205, (1–3) 75–86 (2007).

21] S.A. Younssi, C. Kiefer, A. Larbot, M. Persin and J. Sarrazin, J. Membr. Sci., 1439,27–36 (1998).

22] S.R. Krejewski, W. Kujawski, F. Dijoux, C. Picard and A. Larbot, Colloids Surf. A:Physicochem. Eng. Aspects, 243, 43–47 (2004).

23] C. Picard, A. Larbot, E. Tronel-peyroz and R. Berjoan, Solid State Sci., 6, 605–612(2004).

24] B. Paul, W.N. Martens and R.L. Frost, J. Colloid Interface Sci., 360, 132–138(2011).

25] J. Kujawa, S. Cerneaux and W. Kujawski, Colloids Surf. A: Physicochem. Eng.Aspects, 447, 14–22 (2014).

26] J. Kujawa, S. Cerneaux, S. Koter and W. Kujawski, ACS Appl. Mater. Interfaces,6, 14223–14230 (2014).

27] M. Mulder, Basic Principal of Membrane Technology, 2nd ed., Kluwer AcademicPublisher (1996), pp. 280–415.

28] E. Drioli, V. Calabro and Y. Wu, Pure Appl. Chem., 58, 1657–1662 (1986).

29] S.R. Krajewski, W. Kujawski, M. Bukowska, C. Picard and A. Larbot, J. Membr.

Sci., 281, 253–259 (2006).30] L.M. Diez and F.J.F. Diaz, Desalination, 139, 373–379 (2001).31] C. Picard, A. Larbot, F. Guida-Pietrasanta, B. Boutevin and A. Ratsimihety, Sep.

Purif. Technol., 25, 65–69 (2001).

[

[[

ic Societies 4 (2016) 243–251 251

32] H.R. Davis, Appl. Sci. Res., 34, (2–3) 127–143 (1978).33] I. Voigt, G. Fischer, P. Puhlfur, M. Schleifenheimer and M. Stahn, Sep. Purif.

Technol., 32, 87–91 (2003).34] S. Sarkar, S. Bandyopadhyay, A. Larbot and S. Cerneaux, J. Membr. Sci., 392–393,

130–136 (2012).35] J. Kujawa, W. Kujawski, S. Koter, K. Jarzynka, A. Rozicka, K. Bajda, S.

Cernaux, M. Persin and A. Larbot, Desalin. Water Treat., 51, 1352–1361(2013).

36] J. Li, Chemosphere, (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.12.006.

37] R.B. Bird, W.E. Stewart and E.N. Lightfoot, Transport Phenomena, 2nd ed., WileyPublishers (2002).

38] C.R. Wilke and P. Chang, AIChE J., (2) 264–270 (1955).39] E.S. Saltzman, D.B. King, K. Holmen and C. Leck, J. Geophys. Res., 98, (C9)

16481–16486 (1993).40] C.C. Osawa and C.A. Bertran, J. Braz. Chem. Soc., 16, (2) 1678–4790 (2005).41] B.J. Saikia and G. Parthasarathy, J. Mod. Phys., 1, 206–210 (2010).

42] A.F.M. Pinheiro, D. Hoogendoorn, A. Nijmeijer and L. Winnubst, J. Membr. Sci.,

463, 24–32 (2014).43] A.M. Alkiaibi and N. Lior, Desalination, 171, 111–131 (2004).44] J. Kujawa and W. Kujawski, Chemical Papers, (2015), http://dx.doi.org/10.1515/

chempap-2015-0155.