Em

FOU

a

ARRAA

KAMSCC

1

toiCdstta[cmiafl

B

0d

Journal of Membrane Science 345 (2009) 53–58

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

ffects of antiscalants to mitigate membrane scaling by direct contactembrane distillation

ei He, Kamalesh K. Sirkar ∗, Jack Gilron1

tto H. York Department of Chemical, Biological and Pharmaceutical Engineering, Center for Membrane Technologies, New Jersey Institute of Technology,niversity Heights, Newark, NJ 07102-1982, United States

r t i c l e i n f o

rticle history:eceived 17 April 2009eceived in revised form 28 July 2009ccepted 8 August 2009vailable online 15 August 2009

eywords:ntiscalantembrane distillation

a b s t r a c t

The effects of antiscalants on mitigating the potential for membrane scaling by calcite and gypsum,respectively, were investigated during the direct contact membrane distillation process (DCMD) imple-mented with porous hydrophobic polypropylene (PP) hollow fibers having a porous fluorosilicone coatingon the fiber outside surface. The surface tension and the membrane breakthrough pressure were testedfor different kinds of antiscalants. At room temperature, antiscalant solutions behave like tap water.Based on this result, DCMD scaling experiments with CaSO4 or CaCO3 as a scaling salt were conducted.The supersaturation indices of the scaling salts used correspond to sea water concentrated 5 times forCaSO4 (∼75 ◦C) or half of the maximum saturation index (SI) reached during the concentration of sea

◦

calingaCO3

aSO4

water to 10 times for CaCO3 (∼73 C). The results show that antiscalants K752 and GHR could dramat-ically extend the induction period for the nucleation of gypsum and calcite, respectively; further theyslow down the precipitation rate of crystals, even at a dosage of only 0.6 mg/L. By comparison, a largeramount of antiscalant could further slow down the precipitation and also extend the induction period forboth calcite and gypsum systems. There was no sign of any drop in the water vapor flux nor any increasein the distillate conductivity. Concentrates or reject streams from reverse osmosis desalination processes

ay th

containing antiscalants m

. Introduction

Like other thermal desalination processes, membrane distilla-ion (MD) could achieve much higher water recovery than reversesmosis (RO), given its lack of sensitivity to osmotic pressures. Dur-ng such a membrane distillation process, the mineral salts, e.g.,aSO4 and CaCO3 may potentially cause severe membrane scalingue to their inverse temperature-solubility behavior [1–3]. Indeeduch scaling is a major limiting factor to recovery in conventionalhermal desalination processes. Any method to solve such a poten-ial problem will be very helpful in terms of reducing downtimend cost. While it has been shown in our previous communications4,5] that membrane distillation is more resistant to scaling thanonventional thermal processes, there is still room for improve-

ent. As reported in the literature [6–9], antiscalants are efficient in

nhibiting scaling from deposits not only during the RO process butlso during higher temperature processes, for example, multistageash (MSF) desalination, multieffect desalination (MED) and also

∗ Corresponding author. Tel.: +1 973 596 8447; fax: +1 973 642 4854.E-mail addresses: sirkar@adm.njit.edu, kamalesh.k.sirkar@njit.edu (K.K. Sirkar).

1 Zuckerberg Institute for Water Research, on leave from Ben-Gurion University,eer-Sheva, Israel 84105 (09/01/05–08/30/06).

376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2009.08.021

erefore be conveniently concentrated further by DCMD.© 2009 Elsevier B.V. All rights reserved.

associated heat exchange processes. The application of antiscalantscould potentially help prevent scaling in membrane distillation aswell, if there is no problem of pore wetting caused by antiscalants.To our knowledge, no investigations of this application have beenreported.

Commonly used antiscalants include condensed polyphos-phates, organophosphonates, and polyelectrolytes. Effective poly-electrolyte inhibitors are mostly polycarboxylic acids, e.g.,polyacrylic acid, polymethacrylic acid and polymaleic acid. Theantiscalant technique is cost effective: in many cases scaling can besuppressed with less than 10 mg/L of the antiscalant due to physicalmechanisms rather than chemical mechanisms [10]. Specifically,antiscalants can (1) force a crystal morphology selective nucle-ation, (2) add diffusion and adsorption resistances retarding thecrystal growth velocity and (3) change the crystal surface proper-ties and therefore its agglomeration tendency [11]. Because of theirreported effects in modifying the surface energies of crystals, thepossible impact of these antiscalants on the surface energy of thehydrophobic MD membranes should be evaluated.

We have recently initiated a series of studies on desalination byDCMD. These studies carried out in laboratory at two scales [12,13]as well as in pilot plant scale [14] employed porous hydrophobicpolypropylene hollow fiber membranes having a porous fluorosil-icone coating on the outside surface exposed to the hot brine.

5 brane

Imbtbimhbctpb

iccopagPmtFn

tmmtdt

2

S0T∼pCcstwt

TI

4 F. He et al. / Journal of Mem

n the present study, experiments were first conducted to deter-ine whether the microporous hydrophobic PP membrane would

e wetted by antiscalant solutions at possible working concen-rations by measuring solution surface tension and membranereakthrough pressure. Polar liquids, such as water, have strong

ntermolecular interactions and thus high surface tension. Organicolecules which are mostly apolar or have low polarity usually

ave a small value of surface tension. As reported [15], a hydropho-ic porous membrane could easily become wetted by aqueous feedsontaining organics. Breakthrough pressure is a critical penetra-ion pressure, above which the liquid penetrates the membraneores [16]. Membrane distillation operation can be difficult if thereakthrough pressure magnitude is low.

Next, the effects of different antiscalant solutions on mitigat-ng potential membrane scaling in DCMD were investigated andompared. In each case, the induction period and the calcium pre-ipitation rate of gypsum or calcite were measured via the changef calcium concentration with time. Since scaling caused by calciterecipitation could be easily inhibited by the addition of a dilutecid, e.g., HCl [5], the performance of antiscalants on mitigatingypsum scaling was given greater weight for the MD experiments.rior to DCMD scaling experiments, five different kinds of com-ercial antiscalants were compared in a non-membrane system, so

hat the best antiscalant to inhibit the gypsum scaling was known.or both scaling salts, the effects of the antiscalant dosage on theirucleation were investigated.

As reported in literature [17], partial pore wetting will decreasehe water vapor flux. The entrance of a saline solution into the

embrane pores [18] resulting from pore wetting by an antiscalantay cause an increase in the distillate conductivity. Here we used

he parameters of water vapor flux and distillate conductivity toetermine whether there was any problem with antiscalants onhe performance of DCMD through pore wetting.

. Experimental procedure

The antiscalants studied in this paper are listed in Table 1.urface tensions of antiscalant solutions having concentrations of.6, 20, 50 and 70 mg/L were tested with the Kruss K-8 Surfaceensiometer (with Du Nouy Ring) (KRÜSS USA, Matthews, NC) at23 ◦C. For breakthrough pressure tests, a piece of a 25 �m thickorous hydrophobic polypropylene flat membrane (Celgard 2400,elgard, Charlotte, NC) was used in a test cell. The gas N2 from a

ylinder was used to push the antiscalant solution inside a pres-ure vessel to flow over one side of the membrane. The outlet ofhe test cell on this side of membrane was blocked. Breakthroughas checked by whether water came out from the other side of

he test cell. Within 24 h if no breakthrough happened, the pres-

able 1nformation on antiscalantsa studied.

Antiscalants Chemical name Weighthan

K797b Water 50Acrylic terpolymer/Solids 50

K752b Polyacrylic acid 47Water 37Sodium polyacrylate 16

GHRc Aqueous solution of a nitrogen containingorgano-phosphorus compound

N/A

GLFc Aqueous solution of an organo-phosphorus compound N/AGSIc Synergistic blend of antiscalants based on neutralised

carboxylic and phosphonic acidsN/A

a Provided by manufacturers.b From Noveon Inc. (Cleveland, OH).c From Genesys International LTD. (Minneapolis, MN).

Science 345 (2009) 53–58

sure measured by a test gauge was increased to a higher value till997.6 kPa (130 psig) was reached. Breakthrough pressures of anti-scalant solutions were tested for different concentrations in thefollowing order, e.g., 0.6, 2, 4, 8, 14, 20 mg/L. Before testing anotherseries of antiscalant solutions, the test cell and the flat membranewere thoroughly washed/flushed with DI water and dried in air.

The following procedure was adopted for CaSO4 scaling stud-ies with antiscalants without any membrane. At room temperatureNa2SO4 solution was added into the mixed solution of CaCl2 solu-tion and antiscalants. The mixed solutions were stirred for 20 min(Corning magnetic stirrer, Lowell, MA) before being heated up tothe MD operating temperature under a fixed heating procedureand stirring rate. The time to achieve a full turbid white solutionwas recorded for each of the tested mixed solution.

The experimental apparatus developed to study the effects ofantiscalants on mitigating scaling during the DCMD process wassimilar to that employed in our previous scaling studies [4,5] exceptthat the filter holder was removed. The properties of the mem-brane modules employed are shown in Table 2. The hot brine andthe cold distillate were pumped respectively through the shell sideand lumen side of the hollow fiber DCMD membrane module. Ineach experiment the inlet temperatures of both brine and distillatewere kept constant. To keep the feed concentration stable, the liq-uid level was maintained between two level probes. By recordingthe amount of make-up water for each refill (noted as mm, mg) andthe time difference between two continuous make-ups (noted ast, min), the water vapor production rate could be calculated frommv = mm/t. Alternatively it could be calculated from the local slopeof the total make-up water weight vs. time.

Supersaturated solutions of calcium sulfate (or calcium car-bonate) were prepared by mixing equimolar solutions of CaCl2and Na2SO4 (or NaHCO3). The antiscalant solutions were dosedand mixed with the supersaturated Ca solution for 20 min beforethe feed solution was heated up. The methods of sample-taking,sample-preparation, atomic absorption (AA) analysis for [Ca] in thebrine tank and Ion Chromatograph (IC) analysis for the anion in thedistillate tank were as described in [4,5].

3. Results and discussion

3.1. Surface tension of antiscalant solutions

For all of the tested antiscalant solutions having concentra-

tions varying from 0.6 to 70 mg/L, the value of surface tension wasfound to be 71.5 mN/m, which is very close to that of tap water(71.8 mN/m). It means that, at the experimental concentrations,the solutions of antiscalants, K752, K797, GHR, GLF and GSI show awater-like polar property and negligible surfactant effects.

t % less pH Recommended conditions

Dosage (mg/L) CaCO3 CaSO4

2.4–3.0 N/A N/A N/A

2.2–3.0 N/A N/A N/A

1.8–2.0 0.4–0.8 Best choice Effective

9.8–10.2 2–4 Best choice Effective9.8–10.2 2–5 Effective Effective

F. He et al. / Journal of Membrane Science 345 (2009) 53–58 55

Table 2Representative transport data for different membrane modules.

MXFR #37a MXFR #46a MXFR #47a

Permeance of N2, cm3(STP)/s-cm2-cm Hgb 0.18 0.21 0.20Water vapor flux, kg/m2-hc, at Tb,i = 75 ◦C and Td,i = 20 ◦C 11 12 12

a The fibers in the hollow fiber membrane modules containing porous polypropylene hollow fiber membranes (MEMBRANA, Wuppertal, Germany) were coated on the O.D.with silicone fluoropolymer by Applied Membrane Technology (AMT) Inc. (Minnetonka, MN) using plasma polymerization. No. of layers × fibers per layer = 13 × 29; effectivemembrane surface area (based on I.D.) = 250 cm2; effective cross-sectional area for shell side liquid flow (as shown in [4]) = 4.31 cm2; fiber O.D./I.D. (�m) = 630/150; effectivefi .

N2 inemper

t

3

Gptpicbcpa

3p

prf

l7SAmls

pmtoctcacttl

TE

Q

rate of CO2 generation : rCO2 = rCaCO3

rate of CO2 removal =∫

module area

JCO2 dA

ber length = 6.4 cm; maximum pore size = 0.60 �m; membrane porosity = ∼60–80%b Experimental conditions: temperature: 23 ◦C; atmospheric pressure: 76 cm Hg;c Experimental conditions: shell side: 0.07 mol/L NaCl solution at 75 ◦C (inlet t

emperature) at 138 mL/min of volumetric flow rate.

.2. Breakthrough pressure through porous PP membrane

For each of the antiscalants studied, K752, K797, GHR, GLF andSI, there was no breakthrough of the solution through the flatorous hydrophobic polypropylene membrane for all samples con-aining antiscalants in the range of 0.6–20 mg/L, even when the testressure was increased to 997.6 kPa (130 psig). The hollow fibers

n the membrane modules employed for the DCMD process wereoated with fluorosilicone, which is significantly more hydropho-ic than polypropylene and therefore has a much lower value ofritical surface tension. They should exhibit a higher breakthroughressure for a given pore size, since the antiscalant solutions showwater-like hydrophilic property.

.3. Effects of concentration of antiscalant GHR on the inductioneriod for CaCO3 nucleation during DCMD

Before any membrane module was tested for scaling, its DCMDerformance was studied using hot brine and cold distillate. Theesults are provided in Table 2. With this background we will nowocus on results from scaling experiments.

The experimental conditions for this group of experiments areisted in Table 3. Here the value of SI for calcite was around 10.7 (at5 ◦C), which corresponds among others to half of the maximumI reached during the concentration of sea water to 10 times [5].ccording to the manufacturer (Table 1), GHR might work well toitigate calcite scaling and the recommended dosage is relatively

ow; so GHR was studied for its performance to mitigate calcitecaling during DCMD.

The effects of dosage of GHR on the induction period and therecipitation rate for calcite are shown in Fig. 1a. For the experi-ent without any addition of GHR, the induction period was less

han 9 min; it was extended to 143 and 225 min at a concentrationf [GHR] = 0.6 and 3.0 mg/L, respectively. It is obvious that the pre-ipitation for calcite was significantly inhibited by the addition ofhe antiscalant solution. At the end of the experiment, calcium con-entration dropped to 68% at 285 min for the experiment withoutny antiscalant. However, there is less than 5% of crystallization of

alcite at the end of the experiments (around 250 min) for the otherwo experiments for calcite. Therefore, the effect of antiscalant GHRo mitigate the scaling of calcite at SI = 10.7 is significant even at aow dosage of 0.6 mg/L.

able 3xperimental conditions for CaCO3 as a scaling salta.

GenesysHR (mg/L)

Tb,i (◦C) Tb,i (◦C) Total time(min)

Module # mv

(mg/min)

3.0 71.5 21.4 256 37 3.40.6 71.5 21.8 245 37 3.40 73.2 21.3 285 37 3.3

a [Ca]0 = 2.325 mmol/L, SI(Calcite) = 10.7 (at 75 ◦C), Qb,i = 465 mL/min,d,i = 100 mL/min.

let: tube side; N2 outlet: shell side.ature) at 465 mL/min of volumetric flow rate; tube side: DI water at 20 ◦C (inlet

Fig. 1b shows the change of pH values with time in the brinefeed solution. In the case of no antiscalant, the pH value droppedwithin the first 70 min. This reflects the fact that during this periodthe precipitation rate for calcite, which releases CO2 (see Eq. (1)),was faster than the desorption rate of CO2:

2HCO−3 + Ca2+Heat−→H2O + CO2 ↑ + CaCO3 ↓ (1)

Fig. 1. Effects of antiscalant Genesys HR on the precipitation rates of calcite at aninitial SICal of 10.7: (a) calcium concentration in percentage of the initial value vs.time; (b) pH vs. time.

56 F. He et al. / Journal of Membrane Science 345 (2009) 53–58

Table 4Effects of different antiscalants on CaSO4 nucleation.

Tested antiscalants Concentration ofantiscalants (mg/L)

[Ca2+]0 = [SO42−]0 (mmol/L) Additional time after 50 ◦C

was reached (min)Temperature of the fullturbid white solution (◦C)

Without antiscalant 0 48 +3.5 60GSI 6 48 +6 75GHR 6 48 +11 75K797 6 48 +12.5 75GLF 6 48 +14.5 75

Acaadfp

3f

cmbosostcwoifK5b3slftubBtp

iis

TE

Q

was used to reduce the induction period, the local slope which isproportional to the permeation rate (mv) is constant with time.The constancy of flux is supported by the constancy of the brinetemperature drop and the distillate temperature rise during eachexperiment [4]. For example, for the case with a maximum dose of

GLF 11 44K752 6 48

fterwards, the pH value increased with time when the rate ofalcite precipitation slowed down relative to CO2 desorption rate,s recently reported [5]. In contrast, during experiments involvingntiscalant treatment, the pH kept on increasing and no apparentrop in pH was observed. The reason is that the induction periodor the nucleation of calcite is long, and that the rate of calciumrecipitation is relatively low.

.4. Effects of antiscalant concentration on the induction periodor CaSO4 nucleation during DCMD

As proved by experiments during the DCMD process, the pre-ipitation rate for CaSO4 is faster than the process without anyembrane due to the concentration polarization on the mem-

rane surface [4]. Therefore, to evaluate the relative effectivenessf the antiscalants in the absence of a membrane (to prevent pos-ible membrane damage from suspended crystal solids), a seriesf experiments were conducted at much higher gypsum super-aturations (with SIGyp ≥ 2.8) than in the DCMD experiments. Oneypical experiment was conducted as follows: a supersaturatedalcium sulfate solution of 48 mmol/L at 75 ◦C (SI(gypsum) = 2.83,hich corresponds to sea water concentrated 8 times), was tested

n 5 different kinds of antiscalants dosed at 6 mg/L. After heat-ng quickly up to 50 ◦C for 4 min, turbidity appeared in theollowing order: without antiscalant, GSI, GHR, K797, GLF and752 as shown in Table 4. When raising the dose of GLF bymg/L and lowering the initial concentration of calcium sulfatey 4 mmol/L the complete turbidity of the solution was observed2 min earlier than that in the above K752 experiment. The resultshow that when K752 is added as an antiscalant, the crystal-ization of gypsum is significantly slowed down. For the otherour antiscalants, there was not much of a difference betweenheir performances. It is believed that the main advantage insing polyacrylates (in K752) lies in their excellent thermal sta-ility compared with polyphosphates and other copolymers [6].ased on these results, the effects of K752 and GHR on reducinghe induction period during DCMD were investigated and com-

ared.

The DCMD experimental conditions for K752 and GHR are listedn Table 5. An initial SI for gypsum was around 1.4 (at 75 ◦C) whichs close to the value for sea water concentrated 5 times [5]. Fig. 2hows the effects of dosage of the antiscalant GHR on the precipi-

able 5xperimental conditions for CaSO4 as a scaling salta.

Antiscalants mg/L Tb,i (◦C) Tb,i (◦C) Total time(min)

Module # mv

(mg/min)

GHR 3.0 75.5 22.3 255 37 4.5GHR 0.6 75.7 23.5 241 37 4.6– 0 74.0 21.5 240 47 4.9K752 0.6 72.0 21.8 413 46 4.9

a [Ca]0 = 24 mmol/L, SI(Gyp) = 1.39 (at 75 ◦C), SI (Anhy) = 2.32 (at 75 ◦C),b,i = 465 mL/min, Qd,i = 138 mL/min.

+42 75+74 75

tation rate of gypsum during DCMD process. Without any additionof GHR, the induction period was 40 min. When the concentra-tion of GHR was increased to 3 mg/L, the induction period wasextended to 2 h. The precipitation rate for gypsum was also sloweddown at a higher concentration of GHR. For example, at around260 min, the amount of calcium ion in solution dropped to 97.5%and 95.5% at [GHR] = 3.0 and 0.6 mg/L, respectively. In compari-son, without the addition of antiscalant the calcium concentrationdecreased to 86%. The different effects of the antiscalants K752 andGHR on the calcium ion concentration in the solution during theDCMD process are also shown in Fig. 2. At an antiscalant dosage of0.6 mg/L, the induction period was extended from 50 min to over100 min when K752 was added instead of GHR. Furthermore, whenK752 was used, the precipitation rate of gypsum was dramaticallyslowed. Within 7 h, the amount of precipitation (of [Ca]) was nomore than 3 %, although the concentration of K752 was as low as0.6 mg/L.

3.5. Effects of antiscalants on water vapor flux during the DCMDscaling experiments

Fig. 3a shows the effect of antiscalants GHR and K752 on thetotal weight loss of the make-up pure water vs. time, namely,the water vapor production rate, during DCMD when mixed solu-tions of CaCl2 and NaHCO3 or CaCl2 and Na2SO4 were used asscaling salts. For each experiment in which either GHR or K752

Fig. 2. Effects of antiscalants GHR and K752 on the precipitation rates of calciumsulfate at an initial SIGyp of 1.4.

F. He et al. / Journal of Membrane Science 345 (2009) 53–58 57

Fig. 3. (a) The change of the total make-up weight loss with time when an antis-cTdp

asiwsogo3flTbdpf

3t

i

Fig. 4. The distillate conductivities vs. time when NaCl solution (72 mmol/L), anti-

alant Genesys HR or K752 was used to reduce CaSO4 or CaCO3 scaling during DCMD.he slope of each line represents the water vapor production rate. Experimental con-itions are listed in Tables 3 and 5. (b) Effects of antiscalant GHR (3.0 mg/L) on theerformance of DCMD when CaSO4 (SIGyp = 1.4) was used as a scaling salt.

ntiscalant, i.e., [GHR] = 3.0 mg/L, the temperature drops on bothides of the membrane module during the calcium sulfate scal-ng experiment appeared stable, as shown in Fig. 3b, although the

ater vapor flux shown in Fig. 3a appears to fluctuate a little. Thepecific values of water vapor production rate based on the slopef the line in Fig. 3 are listed in Tables 3 and 5. For the series ofypsum experiments, the water vapor fluxes were close to eachther when the concentrations of antiscalant GHR were 0.6 and.0 mg/L. For the series of calcite experiments, the water vaporuxes were also constant independent of the dosage of antiscalant.hese results prove that the water vapor flux cannot be affectedy the dosing of antiscalant GHR or K752 during the membraneistillation process. We may conclude that there is no problem ofartial pore wetting caused by the dosed antiscalant into the brineeed.

.6. Effects of antiscalants on the distillate conductivity duringhe DCMD scaling experiments

During any of the experiments listed in Tables 3 and 5, increasen the distillate conductivity was never observed. Fig. 4 com-

scalant K752 (0.6 mg/L) and the mixed solution of CaCl2 (24 mmol/L), Na2SO4

(24 mmol/L) and K752 (0.6 mg/L) were used respectively as the hot feed (72–76 ◦C)during DCMD process.

pares the distillate conductivities with time during DCMD whenthree different feed solutions were used: NaCl (72 mmol/L); K752(0.6 mg/L); the mixed solution of CaSO4 (24 mmol/L) and K752(0.6 mg/L), respectively. It shows that for the three experiments,the distillate conductivity decreases in a similar manner with time.At the end of 7 h of experiments, the average distillate conductiv-ity was no more than 3 �S/cm. Similar results were obtained forother experiments, where GHR at a concentration of 0.6 or 3 mg/Lwas used to reduce either calcite or gypsum scaling during DCMD.As measured by ion chromatography, no diffusion of anions acrossthe membrane pores is observed. The results prove that there is noproblem of distillate contamination caused by partial or completewetting of the membrane micropores by the antiscalants K752 andGHR, when they are used to reduce calcite or gypsum scaling duringDCMD.

4. Concluding remarks

(1) The addition of antiscalants prolongs substantially the induc-tion period of CaSO4 and CaCO3, even at dosage levels as low as0.6 mg/L.

(2) Antiscalant solutions do not cause any pore-wetting problemunder our experimental conditions during DCMD.

(3) Antiscalant K752, a polyacrylic acid and sodium polyacrylatebased compound, is more effective in inhibiting CaSO4 scaling,compared with other antiscalants tested, i.e., K797, GHR, GLFand GSI.

(4) A larger amount of antiscalant helps to inhibit scaling of gypsumin a tested range of SI(Gyp) from 1.39 to 2.83 and calcite testedat a SI(calcite) of 10.7.

Acknowledgements

The authors are grateful for the financial support from the Office

of Naval Research (Contract No. N000140510803). Jack Gilron spenthis sabbatical at NJIT during 09/05–08/06. Noveon Inc. (Cleveland,OH) and Genesys International LTD (Minneapolis, MN) provided theantiscalants.

58 F. He et al. / Journal of Membrane

Nomenclature

DCMD direct contact membrane distillationmm weight loss of make-up pure water, gmv distillate production rateMD membrane distillationSI saturation indexTb,i brine inlet temperature to DCMDTd,i distillate inlet temperature to DCMDQ brine inlet flow rate, mL/min

R

[

[

[

[

[

[

b,iQd,i distillate inlet flow rate, mL/min

eferences

[1] K. Karakulski, M. Gryta, Water demineralisation by NF/MD integrated pro-cesses, Desalination 177 (2005) 109–119.

[2] M. Gryta, Water purification by membrane distillation process, Separation Sci-ence and Technology 41 (2006) 1789–1798.

[3] K. Karakulski, M. Gryta, A. Morawski, Membrane processes used for potablewater quality improvement, Desalination 145 (2002) 315–319.

[4] F. He, J. Gilron, H. Lee, L. Song, K.K. Sirkar, Potential for scaling by sparinglysoluble salts in crossflow DCMD, Journal of Membrane Science 311 (2008)68–80.

[5] F. He, K.K. Sirkar, J. Gilron, Studies on scaling of membranes in desalinationby direct contact membrane distillation-2: CaCO3 and mixed CaCO3/CaSO4

systems, Chemical Engineering Science 64 (2009) 1844–1859.

[

[

[

Science 345 (2009) 53–58

[6] Z. Amjad, Calcium sulfate dihydrate (gypsum) scale formation on heatexchanger surfaces: the influence of scale inhibitors, Journal of Colloid InterfaceScience 123 (1988) 523–536.

[7] S.B. Ahmed, M. Tlili, M.B. Amor, H.B. Bacha, B. Elleuch, Calcium sulphate scaleprevention in a desalination unit using the SMCEC technique, Desalination 167(2004) 311.

[8] P. Budhiraja, A.A. Fares, Studies of scale formation and optimization of anti-scalant dosing in multi-effect thermal desalination units, Desalination 220(2008) 313–325.

[9] A. Rahardianto, W.-Y. Shih, R.-W. Lee, Y. Cohen, Diagnostic characterization ofgypsum scale formation and control in RO membrane desalination of brackishwater, Journal of Membrane Science 279 (2006) 655–668.

10] L.A. Perez, Mechanism of Calcium Phosphate Scale Formation and Inhibition inCooling Systems, Kluwer Academic Press, New York, 1998.

11] M. Gloede, T. Melin, Physical aspects of membrane scaling, Desalination 224(2008) 71–75.

12] L. Song, B. Li, K.K. Sirkar, J. Gilron, Direct contact membrane distillation-baseddesalination: novel membranes, devices, larger-scale studies and a model, I&ECResearch 46 (2007) 2307–2323.

13] B. Li, K.K. Sirkar, Novel membrane and device for direct contact membranedistillation-based desalination process, Industrial & Engineering ChemistryResearch 43 (2004) 5300–5309.

14] L. Song, Z. Ma, X. Liao, P.B. Kosaraju, J.R. Irish, K.K. Sirkar, Pilot plant studies ofnovel membranes and devices for direct contact membrane distillation-baseddistillation, Journal of Membrane Science 323 (2008) 257–270.

15] A.C.M. Franken, J.A. Nolten, M.H.V. Mulder, D. Bargeman, C.A. Smolders, Wettingcriteria for the applicability of membrane distillation, Journal of MembraneScience 33 (1987) 315–328.

16] V. Calabro, E. Drioli, F. Matera, Membrane distillation in the textile wastewatertreatment, Desalination 83 (1991) 209–224.

17] M. Gryta, Long-term performance of membrane distillation process, Journal ofMembrane Science 265 (2005) 153–159.

18] M. Gryta, The fermentation process integrated with membrane distillation,Separation and Purification Technology 24 (2001) 283–296.