R

Ss

Aa

b

c

a

ARRAA

KRSACR

C

0d

Journal of Membrane Science 383 (2011) 1– 16

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

jo u rn al hom epa ge: www.elsev ier .com/ locate /memsci

eview

cale formation and control in high pressure membrane water treatmentystems: A review

lice Antonya,∗, Jor How Lowa, Stephen Grayb, Amy E. Childressc, Pierre Le-Clecha, Greg Lesliea

UNESCO Centre for Membrane Science and Technology, The University of New South Wales, Sydney 2052, AustraliaInstitute of Sustainability and Innovation, Victoria University, PO Box 14428, Victoria 8001, AustraliaDepartment of Civil and Environmental Engineering, University of Nevada, Reno, NV 89557-0258, United States

r t i c l e i n f o

rticle history:eceived 8 June 2011eceived in revised form 16 August 2011ccepted 27 August 2011vailable online 2 September 2011

eywords:everse osmosiscale formationntiscalantsalcium carbonate scaleeal time monitoring

a b s t r a c t

The nucleation and growth of inorganic scale on membrane surfaces is a problem shared by operatorsof reverse osmosis systems. While suite of indicators exists to predict scale formation based on purecompound solubility, it is apparent the onset of crystal nucleation and the properties of precipitatedscale in reverse osmosis systems operating on natural waters differs from what would be expected incontrolled conditions. Managing scale formation relies on understanding the chemistry of the scale,judicious system design, appropriate chemical application and early detection. The following reviewconsiders the mechanism of scale formation and the properties of alkaline, non-alkaline and silica basedscales that are encountered when reverse osmosis is used in desalination, brackish water and wastewaterrecycling applications. Management of scale formation can be achieved at the design stage by the inclusionof unit processes to the scale forming constituents or by the application of antiscalants that delay the onsetof nucleation. A range of conventional and emerging analytical techniques, including direct observationand spectroscopic methods have evolved to detect scale formation in real time. It is apparent, however,

the type of scale encountered, limits of system recovery, process configurations and decisions on chemicalselection are site specific. In particular, in wastewater reclamation, the high reporting incidence of calciumphosphate scale suggests that more research is needed on the mechanisms, control strategies for thisscale. The paper highlights the need for molecular level understanding and various factors including theneed for greater transparency of antiscalant formulations, studies on natural waters and complex waterchemistry.

© 2011 Elsevier B.V. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22. Scale forming mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33. Factors affecting scale formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34. Types of scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

4.1. Calcium carbonate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.2. Calcium sulphate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54.3. Calcium phosphate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64.4. Other non-alkaline scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.5. Silicate scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

5. Prediction of scaling tendency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75.1. Langelier saturation index (LSI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85.2. Stiff and Davis saturation index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

5.3. Supersaturation index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.4. Molar ratio concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6. Scale control techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +61 2 9385 5373; fax: +61 2 9385 5966.E-mail address: alicea@unsw.edu.au (A. Antony).

376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2011.08.054

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 A. Antony et al. / Journal of Membrane Science 383 (2011) 1– 16

6.1. Altering feed water characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.1.1. Coagulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.1.2. Ion-exchange softening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.1.3. Acidification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

6.2. Optimization of operating parameters and system design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.2.1. Limiting product recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96.2.2. Feed flow reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106.2.3. Intermediate chemical demineralization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106.2.4. Rotation filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

6.3. Antiscalant addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107. Non destructive scale monitoring techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

7.1. Permeate flux decline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117.2. Ultrasonic time-domain reflectometry analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127.3. Visual observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127.4. Electrical impedance spectroscopy analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

8. Concentrate disposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139. Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 . . . . . .

1

awrprrSi[ccwdwd((Hp

easwdi[i

swioctbe

afcc

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. Introduction

In the 21st century, recognition of a potential water short-ge and the unpredictable impact of global warming on overallater scarcity posits that the first and second decades should be

eferred to as the “water crisis decades” [1]. This shortage can beartially attributed to global population growth, limited naturalesources, and increased industrial activities [2]. Consequently, toesolve the water scarcity problem in many regions in the Unitedtates [3,4], drought-prone zones such as the Middle East [5,6], andn areas where freshwater supplies are scarce, such as Singapore7], seawater is no longer merely a marginal water resource but aommercial option for securing water supplies. Desalination appli-ations are not limited to seawater, but are also applied to brackishater, river water and wastewater [8]. Based on the IDA worldwideesalting inventory report, high pressure membrane desalination,ith nanofiltration (NF) and reverse osmosis (RO), is the prevalentesalination operation for various feed types and accounts for 55%RO – 51%; NF – 4%) of the total water produced by desalination44.1 Mm3/day) [9]. The largest seawater RO (SWRO) plant is theadera plant located in the Mediterranean coast of Israel with aroduction capacity of 127 GL/yr.

Even though thermal and membrane desalination processesqually share the desalination production capacity, RO has emergeds the leader in future desalination installations [10]. High pres-ure membrane operations constitute 38, 87 and 79% of the totalater production from seawater, brackish water and wastewateresalination processes, respectively [9]. Brackish water RO (BWRO)

s gaining more attention due to the low cost compared to SWRO10]. As per the water desalination report, the membrane markets estimated to healthily grow at an annual rate of 16% [11].

The unique property of RO membranes to reject inorganicpecies, while passing relatively pure water has led to itsidespread use in the treatment and reclamation of high-salinity

nland water sources [12]. Achieving high product recovery (ratiof the product volume to the feed volume) and minimizing pro-ess cost is a major challenge in RO operation. Studies have shownhat water product recovery for inland water reclamation by mem-rane desalination has to be sufficiently high, i.e. ≥70–80% to beconomically feasible [12,13].

Membrane fouling is a serious problem, resulting in perme-

te flux decline or increased transmembrane pressure. Membraneouling was, by and large, viewed as an accumulation of rejectedonstituents on the membrane surface [14–17]. Hoek et al. [18]omprehensively defined RO membrane fouling as comprising

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

two components: external/surface and internal fouling. There arethree distinct mechanisms of external fouling, and fouling can beattributed to one or more of them: (i) heterogeneous crystallizationof mineral salts on the membrane, scale formation (ii) accumula-tion of rejected solids on the membrane, cake formation and (iii)colonization of potential microorganisms, i.e. biofouling. Externalfouling is influenced by operating conditions like solution chem-istry, temperature, nature of membrane and module geometry.Internal fouling includes any possible changes to the membranestructure such as membrane degradation and compaction. Externalfouling is generally reversible and manageable by chemical clean-ing, whereas internal fouling is usually irreversible [18]. Surfacefouling is a complex phenomenon and could be a combination ofcolloidal, organic, inorganic and biofouling. The nature of foulingis strongly dependent on feed water source. For example, seawa-ter sources are characterized by high total dissolved solids (TDS,typically ranging from 18,000 to 45,000 mg/L), particulates and col-loidal contaminants whereas brackish waters have relatively lowTDS (typically ranging from 1000 to 10,000 mg/L), organic carbonand colloidal contaminants.

Fouling in high pressure membrane systems can be classifiedinto four broad categories, colloidal/particulate fouling due to theaccumulation of colloidal/particulate matters, organic fouling asa result of deposition of organic macromolecules, inorganic foul-ing which is precipitation of inorganic salts and biofouling due tomicroorganisms [19]. Inorganic fouling or scaling is the formationof hard mineral deposits on the membrane surface as the feed waterbecomes supersaturated by inorganic salts [20]. Inorganic foulingor scaling is referred as precipitation or crystallization fouling [21]and the term scale refers to adherent inorganic deposits formedin place [22]. In high pressure membrane operations, the rela-tive concentration of dissolved salts are concentrated 4–10 times,depending on the operating recovery and rejection efficiencies [23].This causes sparingly soluble inorganic mineral ions like calcium,magnesium, carbonate, sulphate, phosphate and silica to increaseresulting in concentration polarization at the membrane surface.As a result, the concentration of salts may exceed their solubilitylimit and the salts may crystallize onto the membrane surface. Scaleformation has always been a serious limitation in designing andoperating RO systems since scaling causes flux decline, membranedegradation, loss of production and elevated operating costs. Pre-

diction of scale formation tendency of the feed and implementingsuitable scale prevention measures are two essential managementprotocols for the successful operation of RO. Scale formation can bemitigated by appropriate pretreatment measures upstream of RO.

A. Antony et al. / Journal of Membra

fsvgred

2

bcatdiuocfsmac

ssattsatfaip

uti

J

wmp

separations is determined by the operating conditions such as flux

Fig. 1. Schematic illustration of scale formation schemes.

The present article reviews various aspects of inorganic scaleormation, control and monitoring in high pressure membraneystems with emphasis on the role of chemical additives to pre-ent scale formation. It is apparent that there are significantap in understanding of the problem and considerable effort isequired to improve performance in this case. Some of the knowl-dge gaps and specific areas needing further research are alsoiscussed.

. Scale forming mechanisms

Scale formation comprises complex phenomenon involvingoth crystallization and transport mechanisms. Thermodynami-ally, crystallization or precipitation becomes feasible when thectivity of ions in solution is above their saturation limit andhe solution is supersaturated. In addition to supersaturation con-itions, kinetics of precipitation should also be considered as it

s a key determinant of the severity of scaling. When supersat-ration exceeds a critical value, nucleation of scale formationn particle surfaces induces growth of crystals, and low con-entration of nucleation sites slows crystallization kinetics. Scaleormation was reported to occur by two crystallization pathways,urface crystallization and bulk crystallization [24–29]. Scaling inembrane systems is a combination of these two extreme mech-

nisms and is affected by membrane morphology and processonditions.

Surface crystallization occurs due to the lateral growth of thecale deposit on the membrane surface, resulting in flux decline andurface blockage. Bulk crystallization arises when crystal particlesre formed in the bulk phase through homogeneous crystalliza-ion and may deposit on membrane surfaces as sediments/particleso form a cake layer that leads to flux decline. In addition, super-aturated scale forming conditions leads to scale growth andgglomeration. This is due to the random collision of ions with par-icles and secondary crystallization occurs on the surface of theseoreign bodies present in the bulk phase [28,30]. Simultaneous bulknd surface crystallization may also occur for high recovery operat-ng conditions. A schematic representation of these crystallizationrocesses is illustrated in Fig. 1.

Scale formation in high pressure systems has been describedsing a membrane in series model which incorporated crystalliza-ion kinetics [25,27,31]. The approach begins with estimating fluxn absence of fouling using, expressed as:

W = �P − �

�Rm(1)

here Jw is the pure water flux estimated from resistance-in-seriesodel, �P is the applied pressure, � is osmotic pressure, � is the

ermeate viscosity and Rm is the membrane resistance.

ne Science 383 (2011) 1– 16 3

The model is expanded to include permeate flux decline dueto cake layer formation through bulk crystallization and porousdeposit layer formation using:

Jb = �P − �

�(Rm + Rc)(2)

where Jb is the permeate flux estimated from the cake filtrationmodel and Rc is the resistance due to the cake layer formation. Rc

can be calculated using filtration theory from by

Rc = ˛mc

At(3)

where is the specific cake resistance per unit weight of cake andmc is the total accumulated weight of the precipitated scale and At

is the total membrane area.Conversely, scale formation by surface crystallization on the

membrane surface occurs by lateral crystal growth and the filtra-tion surface considered to completely block filtration. Assumingthat the crystals on the surface are impermeable and cake formationis absent, the flux decline can be expressed as:

Js = �P − �

�Rm× At − Ab

At(4)

where Js is the permeate flux estimated from the surface blockagemodel, At is the total membrane surface, Ab is the membrane areaoccupied by surface crystals and (At − Ab) is the free membrane sur-face, uncovered by surface crystals. Assuming that the thickness ofcrystal layer is constant, Ab is the product of area covered per unitmass and weight of scale formed [24].

At high operating recoveries, as the bulk phase becomes super-saturated, both cake formation and surface blockage may occursimultaneously, and the resulting permeate flux can be representedas given below:

Jt = �P − �

�(Rm + Rc)× At − Ab

At(5)

where Jt is the permeate flux estimated by combining the cakefiltration and the surface blockage models.

Even though the retenate solutions are usually supersaturated,scale formation on the membrane is predominantly governed bythe availability of sufficient nucleation sites. Therefore, inductiontime �, the time to induce formation of detectable nucleation crys-tals, is crucial for scale formation of RO membrane.

Bulk crystallization can occur simultaneously with surface crys-tallization but it results in loosely adherent particulate deposits.

3. Factors affecting scale formation

Various operating conditions such as pH [32], temperature[33,34], operating pressure [26], permeation rate [35], flow veloc-ity, [26] and presence of other salts or metal ions [36,37] caninfluence scale formation. Importantly, concentration polarizationplays a vital role in scale formation in high pressure membrane sys-tems [25,26,38,39], as it leads to elevated salt concentrations nearthe surface of the membrane where particles may deposit.

Concentration polarization is a phenomenon that takes placewhen separation occurs at the membrane surface. Solvent passesthrough the membrane and rejected solutes accumulate at thesurface of the membrane [18]. Thus, the concentration of thesalts may become supersaturated at the membrane surface, eventhough the bulk concentration may still remain unsaturated orunder-saturated. The extent of concentration polarization in RO

and water recovery, solution chemistry, temperature, membraneproperties, and module geometry. Concentration polarizationshould be considered when modelling RO systems [40,41], as a

4 embrane Science 383 (2011) 1– 16

sd

piai[appfbcrfi[

4

cpsdcsnactaipaScomc

4

dThbacfsceCna

dachts1

Fig. 2. Crystallization mechanism and transformation of CaCO3 [51], originally

A. Antony et al. / Journal of M

tudy by Lee et al. demonstrated that surface crystallization isirectly influenced by concentration polarization [26].

At low operating pressure, the initial flux and the concentrationolarization remain low, and the degree of surface crystallization

s also low. At high cross flow velocity, the fluid motion increasest the membrane surface and the degree of concentration polar-zation is reduced resulting in a slow rate of surface crystallization25]. Surface crystallization is favored at low crossflow velocitiesnd high operating pressures, while bulk crystallization is morerevalent at intermediate crossflow velocities and high operatingressures [25], because, crossflow enhanced back diffusion awayrom the membrane increases mainly from the centre of the mem-rane. In batch filtration both under stirred [42] and unstirred [26]onditions, surface crystallization was dominant due to the highatio of concentration polarization. On the other hand, in crossflowltration, both surface and bulk crystallizations were responsible26,42].

. Types of scale

Types of scale encountered on RO membranes can be broadlyharacterized as alkaline, non-alkaline and silica based. While it isossible to predict scale based on solubility data, it is often neces-ary to examine a fouled membrane. Common scale encountereduring this assessment includes calcium carbonate (CaCO3), cal-ium sulphate (CaSO4·xH2O), barium sulphate (BaSO4), strontiumulphate (SrSO4), silicates, calcium phosphate (Ca3PO4) and alumi-osilicates [43–47]. It should be noted that membrane autopsiesre performed as a means of problem solving by identifying theause for the loss in membrane performance. It is the only wayo identify the types and extent of fouling in a membrane modulet a pilot or full scale operation. However, the incidence of scal-ng in these cases may not be representative of the actual scalingotential of the feed, since scale deposition in these cases usu-lly occurs after establishing various scale mitigation measures.imilarly the exact crystallographic polymorphs of the scale oftenannot be acquired due to co-precipitation of more than one typef inorganic scale or co-precipitation with organic, colloidal andicrobial foulants, leading to largely amorphous, sludge-like, or

arbonaceous deposits [18,43,48,49].

.1. Calcium carbonate

Calcium carbonate is an alkaline scale formed due to the break-own of bicarbonate ion, and is one of the most common scale.he degree of scaling primarily depends on the level of calciumardness and bicarbonate alkalinity of the feed water. The rate oficarbonate breakdown increases with increase in pH and temper-ture and is also affected by the TDS content. Although calciumarbonate is a common scale with all feed types, its tendency toorm is generally predictable and controllable. For example, in atatistical analysis examining 150 membrane modules, calciumarbonate was problematic in only 4% of cases as only 6 out of 150lements had >30% surface coverage of calcium carbonate scale.alcium carbonate scaling is only a significant issue for a minorumber of reverse osmosis plants, with these plants generally oper-ting with poor or non-existent shutdown flushing procedures [45].

Calcium carbonate is reported to exist in six forms, three anhy-rous crystalline polymorphic forms, calcite, aragonite and vateritend three hydrated forms, amorphous calcium carbonate (ACC),alcium carbonate monohydrate (MCC), and calcium carbonate

exahydrate (CCH) [50–53]. In these studies, a single calcite crys-al shows well-developed rhombohedral morphology, with sharp,traight edges. The average calcite particle size is usually about0 �m. The aragonite clusters are in the form of agglomerates with

adapted from [54]. (a) Change in log IAP with time at 25 ◦C and (b) change inpolymorph abundance.

outward oriented needles, apparently emerging from a commonbase crystal. Vaterite is usually spherical in shape with diametersranging from 0.05 to 5 �m.

Among the three crystalline polymorphs, calcite is the mostthermodynamically stable and vaterite the least stable form[50,52]. The actual crystallization of CaCO3 starts from thermo-dynamically unstable ACC, and crystallizes spontaneously whenpresent in aqueous solution at ambient temperatures. This pro-cess is rapid and governed by factors such as temperature, pHand concentration, leading to the formation of calcite, aragoniteor vaterite. The mechanism of crystallization and transformationof CaCO3 through the morphological changes were proposed to becontrolled by the ion activity product (IAP) [51,54,55]. The solubil-ity of CaCO3 is governed by the reaction CaCO3 = Ca2+ + CO3

2− andIAP = [CA+] [CO3

2−]. Sawada [54] deduced a time sequence throughthe change in the logarithm of IAP as a function of time at 25 ◦C,shown in Fig. 2a and change in the polymorphic abundance of theprecipitates of the suspension, shown in Fig. 2b.

In the log IAP vs. time plot, three regions are seen, unstable (I),metastable (II) and stable (III). After the precipitation of ACC inthe unstable region (I), lasting for less than 5 min, ACC graduallytransforms to calcite and vaterite, and the IAP value is controlledby the solubility of ACC. In the metastable region (II), the log IAPvalue remained almost constant and the transformation gradu-ally moved from vaterite to calcite. The metastable region is quitelengthy, typically extends up to 3 h. The transformation to calciteleads to the stable region (III) and no further change in log IAP orpolymorphic abundance was observed. From these results, the twotransformations, ACC to vaterite and vaterite to calcite, were pro-posed to proceed by a recrystallization mechanism and that wasconfirmed by scanning electron micrographs and size distributionmeasurements [54].

Various studies [50,56–59] suggest that temperature and pH areimportant factors controlling the formation of the final crystallinephase. ACC will transform to calcite via vaterite at low tempera-tures (<30 ◦C) and to aragonite via vaterite at higher temperatures(≥40 ◦C) [54,60]. Although calcite presents the greatest thermody-namic stability under ambient conditions, the thermodynamicallyless stable aragonite and/or vaterite phase may be stabilized under

certain temperature conditions or in the presence of other ions orinhibitors. Coexistence of magnesium ions in solutions supersatu-rated with CaCO3 hinder the formation of vaterite and may favor the

A. Antony et al. / Journal of Membrane Science 383 (2011) 1– 16 5

Fig. 3. SEM images of the CaCO scaling on TFC-FR membranes under controlled conditions (bench scale) after 4 h (a and b) and 14 h (c and d), operated at 10.2 bar, 8.2 pHa

pnf–iefocsto

tsits

4

atmtfo[m

3

nd LSIb – 1.1 [50].

recipitation of aragonite [54,57,61]. For example, surface exami-ation by scanning electron microscopy (SEM) of CaCO3 formation

rom tap water on a TFC-FR membrane operated at 10.2 bar, pH 8.2 and bulk LSI (LSIb) – 1.1 is shown in Fig. 3a–d. Magnesium

s commonly present in water sources together with calcium andxerts a profound inhibitory effect on CaCO3 precipitation evidentrom literature [61–66]. After 4 h filtration, small crystals of arag-nite were largely observed with the beginning of growing calciterystals. After 14 h, calcite crystals were seen all over the membraneurface along with aragonite. Formation of vaterite is not seen onhe membrane surface due to the presence of a substantial amountf magnesium present in the feed water.

SEM examination coupled with energy dispersive X-ray spec-roscopy (EDS) analysis of RO membrane from papermill industry,everely fouled with CaCO3 was shown is presented in Fig. 4 [67]. Its evident that the CaCO3 crystal morphology is different betweenhe scale obtained under controlled conditions using syntheticolutions and natural waters.

.2. Calcium sulphate

Calcium sulphate is the most common scale among the non-lkaline scales. Since cleaning sulphate scale is relatively difficulto alkaline scales in water treatment plants, the best practice for

anaging calcium sulphate scale is to operate the RO system belowhe saturation level. CaSO4 precipitates in three crystallographic

orms: gypsum or calcium sulphate dihydrate form, plaster of Parisr calcium sulphate hemihydrates and calcium sulphate anhydrite26]. At ambient temperatures of 20 ◦C, gypsum is the most com-

on form [26]. Various literature [26,68–70] and experimental

investigations [24,71] have established that gypsum occurs as twoprimary morphologies: needles and platelets, with monoclinic andprismatic structures (Fig. 5).

At room temperature, Lash and Burns [73] found that theneedle morphology was produced from solutions whose initial con-centrations were greater than 0.4 M in CaSO4 while the plateletmorphology was produced from solutions whose initial CaSO4 con-centration was less than 0.25 M. However, Christoffersen et al. [69]found the opposite effect, with low concentrations (0.3 M CaSO4)favoring the needle-like morphology, whereas high reagent con-centrations (0.725 M CaSO4) produce smoother surfaces.

The crystal morphology of CaSO4 was seen to be dependenton the supersaturating ratio and crystallization kinetics. Needlelike crystals were noticed to be generated under conditions oflow supersaturation ratio (defined as the ratio of the ionic activ-ity product to the solubility product), <2.27, dominated by surfacecrystallization and characterized by a lengthy induction time priorto nucleation [68]. On the other hand, plate-like morphology isseen at a supersaturation ratio of 10.86, dominated by bulk crys-tallization. A progressive crystal growth pattern of CaSO4 wasdemonstrated in a bench scale plate and frame RO system by Shihet al. [12] using model feed solution of saturation index 0.06 withrespect to gypsum. During crystal growth on the membrane sur-face with CaSO4 model solutions set at a concentration factor of8, it was observed that at the membrane channel entry, surfacegypsum crystals were at their initial growth stage formed as pri-

marily needle-like structures. With increasing axial distance fromthe entry, surface gypsum crystallization results in crystal struc-tures that transition from the needle and plate-like structures, inthe submicron size range, to partial rosettes to complete rosette

6 A. Antony et al. / Journal of Membrane Science 383 (2011) 1– 16

on obs

sdobg

wotfottC

4

pwohphp

Fig. 4. SEM-EDS images of CaCO3 scale formati

tructures in the millimeter size range. It was also found that theegree of scaling increased toward the channel exit as the levelf concentration polarization increased along the membrane. Thisehavior can be represented by the well-accepted kinetic model forypsum crystallization [74–77]:

dm

dt= k(Cm − CS)n (6)

here k is the crystal growth rate constant, Cm is the concentrationf gypsum at the membrane surface; Cs is the solubility of CaSO4 athe experimental conditions and n is a kinetic order, which variesrom a value of 1 for diffusion-controlled crystallization to a valuef 2 for surface reaction process. From the above equation and givenhat CP increases axially along the membrane, it can be inferred thathe rate of surface scale formation should increase, with increasingm at the tail end of the membrane stage.

.3. Calcium phosphate

Calcium phosphate scaling in RO plants started to be a majorroblem as RO technology was extended to the treatment ofastewater streams [78–80]. Since phosphate can form a variety

f polymeric ions, many forms of phosphate salts are possible;

owever, the majority of calcium phosphate deposits are amor-hous in nature [78]. The crystalline forms of calcium phosphate areydroxyapatite and fluoroapatite, which are calcium orthophos-hate with varying amounts of OH−, Cl− and F−. Formation of

Fig. 5. Needle-like (left) and plate type

erved on RO membrane from a papermill [67].

hydroxyapatite occurs through autocatalytic solution mediatedtransformation of highly unstable amorphous calcium phosphate,hydrolysing rapidly to more stable hydroxyapatites microcrys-tals in water [81]. The lifetime of metastable amorphous phaseis reported to be strongly governed by the exact solution envi-ronment, with the presence of other additive molecules and ions,pH, ionic strength and temperature affecting the lifetimes [82].The transformation of amorphous to crystalline hydroxyapatite,which can be described by a first-order rate law is only a func-tion of pH of the mediating solution at constant temperature. Also,the transformation depends upon the conditions which regulateboth the dissolution of the amorphous calcium phosphate andrenucleation of hydroxyapatite crystals [83]. Crystalline calciumphosphate apatite is less soluble in neutral and alkaline condi-tions and preferentially dissolves in acid pH [81,84]. Phosphatesof aluminium and iron are, however, less soluble at moderatelyalkaline conditions. A calcium phosphate stability index based onthe phosphate concentration, temperature and pH is developedby Kubo et al. [85] that helps to predict the threshold level ofscaling [86].

Unlike CaCO3 scale, there are no suitable antiscalants to mitigatecalcium phosphate scaling on membranes [79,87]. However, Qinet al. achieved effective scale control through pH adjustment [88].

In general, antiscalant addition has less impact in mitigation of thisscale compared with pH adjustment approaches. More specifically,controlling the feed water pH at 6.4 can dramatically minimize therisk of calcium phosphate precipitation [80].

(right) morphology of CaSO4 [72].

A. Antony et al. / Journal of Membrane Science 383 (2011) 1– 16 7

ne au

tipioefsbu

fprtssic

mcupf

4

taeipbetc

lammacm

Fig. 6. SEM-EDS images of RO membrane surface in membra

Ning and Troyer describe the importance of the state in whichhe calcium phosphate arrives at the RO system, in order to min-mize phosphate deposition [87]. After pretreatment, if calciumhosphate is in its dissolved form, i.e. calcium and orthophosphate

ons, then crystallization of calcium phosphate is more likely toccur on the membrane surface and antiscalants are less likely to beffective. On the other hand, if calcium phosphate is in its colloidalorm, colloidal deposits are more likely to form on the membraneurface, instead of crystallization. In these cases, antiscalants wille ineffective at their typical dosage levels and therefore anticoag-lants or dispersants should be added.

Ning and Troyer also demonstrated the presence of colloidalouling of RO membranes arising from nanoparticles of calciumhosphate passing through MF/UF pretreatment in a wastewatereclamation plant [87]. Membrane autopsy results established thathe RO deposit contained 20% organic matter and 80% inorganicalts, which was primarily calcium phosphate. Calcium phosphatecaling was also reported in similar membrane examination stud-es, co-precipitated along with other organic and/or inorganiconstituents [43,45,89].

Thus, the risk of calcium phosphate scaling can be effectivelyanaged by (i) reducing the concentration of orthophosphate, cal-

ium, aluminium, iron and fluoride during pretreatment, (ii) these of dispersants when calcium phosphate is in the form of nano-articles in the feed and (iii) maintaining a low pH for the ROeed [90].

.4. Other non-alkaline scale

Scales including calcium fluoride, barium sulphate and stron-ium sulphate are rarely observed as deposits on RO elements, andre more likely the result of feed water conditions. If present, how-ver, the formation of barium sulphate scale can be problematicn membrane systems due to its low solubility, i.e. 2.33 mg/L inure water. Unfortunately, early detection of barium sulphate cane difficult, and when un-noticed, has the potential to form adher-nt scale [91,92]. Presence of barium sulphate scale seen duringhe examination of a fouled RO module is shown in Fig. 6 [67], therystal shape was consistent with earlier studies [47,78,93].

Calcium fluoride and strontium sulphate can be similarly chal-enging if present in significant concentrations and may also formdherent scales. Adherent layers are challenging to conventionalembrane cleaning practices, necessitating complicated cleaning

ethods like use of ethers and concentrated acids, which dam-

ge the membrane. Antiscalants are effective for barium sulphate,alcium fluoride and strontium sulphate, providing an effectiveitigation technique for scaling control.

topsy, revealing the formation of barium sulphate scale [67].

4.5. Silicate scale

The final form of scale that is important for RO, one is silicascaling. Silica is abundant in nature due to its high concentra-tion in the Earth’s crust. Dissolved silica is present in natural feedwaters either as crystalline or amorphous form. The mechanismof formation of amorphous silica deposits is widely accepted asthe polymerization of silica monomers at high concentrations,and when formed, silica colloids can foul the membrane surface[94–96]. The formation of silica scale is dependent on the pH ofthe silica concentration in solution. The general form of silica islow molecular weight meta silicic acid, (H2SiO3)n, which is a weakacid and remains undissociated below neutral pH. Silicate scalingon the membrane surface occurs when supersaturated silicic acidpolymerizes to form insoluble colloidal or gel-like silica. Above neu-tral pH, silicic acid dissociates to form silicate anions, (SiO3

2−)n,which then form insoluble silicates upon reaction with metal ionslike calcium, iron, manganese and aluminium. Silica is considereda common foulant and is invariably associated with aluminium[45]. Aluminium silicate deposits found on membrane surfaces arepresent in the form of silt, clay and other combinations such asdiatom skeletons; some of the commonly seen forms of silica com-pounds include: clay (Al2O3SiO2·xH2O), mullite (3Al2O3·2SiO2),andalucite (Al2OSiO4) and feldspar (KAlSi3O8) [45]. Silica scalingis likely to occur, even below saturation levels, due to the catal-ysis effects of iron and aluminium ions [45,97–99]. Presence ofsilica detected during the examination of fouled Ro membranesfrom a papermill and groundwater treatment plant were shownin Fig. 7 [67].

Silica scale formation can be mitigated by (i) maintaining thealuminium and iron levels below 0.05 mg/L, regardless of silicaconcentration, (ii) establishing appropriate pretreatment measuresto remove colloidal silica and silicates, and (iii) acidification offeed water and preventative acid cleaning [90]. In the absenceof other trivalent metal cations, scaling tendency is calculatedfrom the silica concentration, temperature, pH and total alkalinity[100].

5. Prediction of scaling tendency

While an understanding of the kinetics and mechanisms of crys-tal growth are important for developing strategies for prevention,the ability to predict scale occurrence from feed water characteris-

tics is vital in practical terms. There are a number of widely appliedmethods for estimating the scale forming potential of feed water;interestingly many of them were originally developed to assess thescaling or corrosive tendency of metallic structures.

8 A. Antony et al. / Journal of Membrane Science 383 (2011) 1– 16

nes op

5

acs

L

wt[C

dpdoiChwmt[l

5

iod

Fig. 7. SEM-EDS images showing silica fouling on RO membra

.1. Langelier saturation index (LSI)

LSI is the most common scale prediction tool for calcium carbon-te scale. This equilibrium model is derived from the theoreticaloncept of saturation and provides an indicator of the degree ofaturation of water with respect to CaCO3. LSI is calculated as:

SI = pH − pHs (7)

here pH is the measured water pH and pHs is the pH at satura-ion in calcite or CaCO3, pHs = (9.3 + A + B) − (C + D) where, A = (Log10TDS] − 1)/10, B = −13.12 × Log10 (temperature in ◦C + 273) + 34.55,

= Log10 [Ca2+ as CaCO3] − 0.4 and D = Log10 [alkalinity as CaCO3].Negative LSI indicates no potential to scale and the water will

issolve CaCO3; positive LSI shows that scale is likely and CaCO3recipitation may occur; LSI’s that are close to zero signify bor-erline scale potential. Water quality or changes in temperaturer evaporation may change the index. LSI is purely an equilibriumndex and deals only with the thermodynamic driving forces foraCO3 scale formation and growth. LSI provides no indication ofow much scale or CaCO3 will actually precipitate or the rate athich it occurs. Standard practice for the calculation and adjust-ent of LSI in RO systems for feed streams containing TDS less

han 10 g/L is also available for the water treatment community101]. However, this index cannot be applied for solutions of veryow salinities below an ionic strength of 0.1 mol/L B.

.2. Stiff and Davis saturation index

Another modified version of LSI, i.e. Stiff and Davis saturationndex (S&DSI), takes into account high levels of dissolved solidsn the solubility of CaCO3 at a given temperature. This index waseveloped for seawater system. Eq. (7) expresses the effects of pH,

erated with (a) papermill effluent and (b) ground water [67].

calcium, total alkalinity, dissolved solids and temperature as theyrelate to the solubility of CaCO3:

S&DSI = pH − pHs (8)

where pH is the measured water pH and pHs = (pCa + pAlk + K).pCa and pAlk are the negative logarithms of the calcium ion andalkalinity, K is a factor that takes into account ionic strength andtemperature. If the index is less than zero, there is little CaCO3 scal-ing potential. As the index value increases, so too does the scalingpotential. In order to prevent CaCO3 scaling, the S&DSI requiresadjustment to a negative value by appropriate measures [102]. AnASTM standard is available and details the calculation and adjust-ment of S&DSI in RO systems for feed streams containing TDS ofmore than 10 g/L [103]. Pena et al. [51] identified that the specificcrystalline polymorph vaterite is the generic value of pHs used inthis index.

5.3. Supersaturation index

From a thermodynamic perspective, precipitation may occurwhen the solubility of mineral salt is exceeded. The extent of super-saturation can be expressed as the supersaturation index (S), whichexpresses the level of saturation of a solution with respect to thevarious mineral salts as Eq. (9):

Sx = IAPKsp,x

(9)

where IAP is the ion activity product and Ksp,x is the solubility prod-uct for mineral salt x [12,104,105]. The solubility product term is afunction of temperature, pressure and ionic strength. The supersat-

uration index can be used to estimate the scaling potential of CaCO3,CaSO4, BaSO4 and SiO2. A saturation index greater than unity indi-cates that the solubility limit for the mineral salt is exceeded andthis a clear warning sign for scaling possibility.

embra

tb2tte

K

Ktsaptmceot

bsuii

5

bitvawffwc(ebsaii

6

i

123

ni

6

t

A. Antony et al. / Journal of M

One example for the application of supersaturation index ishe Du Pont method, most widely applied prediction method forarium sulphate scale in RO systems using the salt solubility at5 ◦C and the solubility product, Kc [106,107]. The Kc expresseshe dynamic equilibrium between the scalant crystal ions in solu-ion and the crystal solid phase, for example the Kc for BaSO4 isxpressed as:

c = [Ba2+][SO2−4 ] (10)

c is determined graphically as a function of ionic strength forhe desired recovery [91]. The relationship between Kc and ionictrength is derived from data on the effect of individual monovalentnd divalent cations on barium solubility. Kc is compared with theroduct of the scalant ions (molarity). When the scalant concentra-ion product exceeds Kc, the solution is supersaturated and scaling

ay occur. Du Pont recommends a recovery such that the scalantoncentration product is 20% below the Kc to prevent scaling. Thisnsures a safety factor for concentration polarization which mayccur at the membrane surface, causing a higher localized concen-ration than in the bulk solution [106,107].

The major limitation of this index is that this index lacksackground information on interactions between ionic species onaturation concentrations; neglects the effects of common andncommon ions on solubility of scale forming species and the

nteractions between ionic species. Correctly, it cannot provide anynformation on precipitation kinetics.

.4. Molar ratio concept

El-Manharawy et al. [108] reported that molar ratios of brine cane used as chemical indicators for the prediction of scaling potential

n RO systems. The results are based on field observations and inves-igations of the formed scales from 33 different cases, with wideariations in TDS (ranging from 300 to 50,000 mg/L). The feed waternd chemical composition of the scale obtained from different casesere analyzed. From the molar concentrations of the brines derived

rom the chemical analysis of the feed water samples, molar ratiosor various combinations of ions were developed and correlatedith scaling indices and saturation indices. The derived molar ratios

orrelate well with the observed scale. For example, molar ratioSO4/HCO3) describes the chemical type of scale formed at differ-nt levels of chloride and calcium ions. Therefore, molar ratio maye used to predict the scale type and its potential formation in ROystems. The interrelationships among different molar species maylso be useful in understanding the relative behavior of associatedons under high pressure conditions. This method still needs fieldnvestigation and validation.

. Scale control techniques

The most common scale mitigating techniques can be groupednto three categories:

. Altering feed water characteristics,

. optimization of operating parameters and system design, and

. antiscalant addition.

Selection among these methods/techniques depends on theature of the feed water, membrane compatibility with acid or scale

nhibitor and cost.

.1. Altering feed water characteristics

The feed water quality is altered to minimize scale formingendency. This method generally involves various pretreatment

ne Science 383 (2011) 1– 16 9

options to limit the mineral concentration or to reduce the alka-linity.

6.1.1. CoagulationCoagulation is an effective pretreatment practiced to primarily

remove the particulate and colloidal matters, both prior to mediafiltration and low pressure membrane filtration. This is an effec-tive method for minimizing the dissolved silica and iron content infeed water and their corresponding scales. For example, Ma et al.[109] demonstrated an effective reduction of iron, silica and micro-bial contamination in the feed water by applying a combination ofenhanced coagulation and membranes.

6.1.2. Ion-exchange softeningSodium regenerated ion-exchange softeners are a good method

for scale control. In this method, sodium adsorbed on ion exchangeresin is exchanged for magnesium and calcium ions that are con-centrated in the RO feed water. This method is popular in systemsdesigned for operating at high recovery on feeds with considerablealkalinity. It is adopted for inland desalting application such as coalseam gas water, where, brine disposal is difficult. The equations forwater softening are as follows [110]:

Ca2+ + 2NaZ → 2Na+ + CaZ2 (11)

Mg2+ + 2NaZ → 2Na+ + MgZ2 (12)

When all the sodium ions have been replaced by calcium andmagnesium, the resin (NaZ) must be regenerated with a brinesolution. Ion-exchange softening may eliminate the need for con-tinuous feed of either acid or antiscalant. However, the significantcapital and operational expense involved with using these soft-eners makes other alternatives more attractive. In addition, brineregenerate discharge can be an issue [110].

High efficiency reverse osmosis process (HEROTM) is anenhanced membrane system process from GE Water and ProcessTechnologies which uses an ion exchange softening method [111].This method involves three steps, (i) cation exchange to minimizethe average hardness to <0.1 mg/L, (ii) degasification of dissolvedcarbon dioxide to less than 10 mg/L and (iii) increasing the feed pH.

6.1.3. AcidificationAcidification involves reduction of pH of the feed water to 5–7

and increasing the solubility of alkaline scale, especially CaCO3which is a potential scalant in all feed water types [112]. The solu-bility of CaCO3 depends on pH as:

Ca2+ + HCO3− ↔ H+ + CaCO3 (13)

Addition of H+ in the form of acid shifts the equilibrium to theleft and maintains the calcium carbonate in the dissolved form.Generally, sulphuric acid or hydrochloric acid is used for pH adjust-ment, however, hydrochloric is preferred due to the potential forsulphuric acid to form sulphate scales, e.g. CaSO4, BaSO4 and SrSO4.Acidification is also effective for controlling calcium phosphatescale.

6.2. Optimization of operating parameters and system design

The risk of scaling can be minimized by making specific changesto the system design or its operating parameters, to keep the scaleforming mineral concentration below the critical threshold limit orby slowing the kinetics of scale formation.

6.2.1. Limiting product recoveryOptimal product recovery is predominantly decided by the sol-

ubility of sparingly soluble salts and varies depending on the feedwater quality. For example, typical operating recovery for SWRO

1 embra

rbrfmvrtmt

6

tfctcBsb

6

wisbeiCchs

6

tTRiat

6

mucomqitilcloosr

ta

0 A. Antony et al. / Journal of M

anges from 35 to 45% and for BWRO from 75 to 90% [10], dictatedy the osmotic pressure of the concentrate stream. As the productecovery increases, the solute concentration at the membrane sur-ace amplifies the concentration polarization effect, increasing the

embrane scaling potential. Therefore, the simplest means of pre-enting scale formation in RO systems is by operating at productecoveries sufficiently low that the reject stream is not concen-rated enough to form inorganic scales. On the other hand, this

ethod has economic implications for the operation efficiency ofhe plant.

.2.2. Feed flow reversalHigh operating recoveries with minimal or no chemical addi-

ives have been proposed by Lauer as a feed flow reversal techniqueor RO process [113]. The principle of this technique is to signifi-antly reduce the elapsed nucleation time by periodically switchinghe feed entrance and concentrate exit positions of the RO pro-ess at times less than the induction time for the scale formation.y reversing the flow before the system achieves induction time,upersaturated brine at the exit port is replaced with unsaturatedrine and vice versa [114,115].

.2.3. Intermediate chemical demineralizationThis operating technique applies scale mitigation for high

ater recovery in two-stage RO processes and involves chem-cal demineralization of the concentrate stream after the firsttage [13]. The first stage RO is operated at product recoverieselow the expected/calculated critical scaling limits; the min-ral ion concentration of the concentrate from the first stages then lowered chemically and sent to the second stage RO.hemical demineralization involves accelerated precipitation byhemical addition, solid–liquid separation and acidification, andas been proposed for CaSO4, CaCO3 and calcium phosphatecales [13,116–118].

.2.4. Rotation filtrationBy rotating of the RO module/cell, the centrifugal flow instabili-

ies help to reduce concentration polarization and scale formation.his approach has been demonstrated for CaSO4 scale control [119].otation is expected to favor bulk crystallization over surface scal-

ng, and the high shear forces reduce the ion concentration build upt the membrane surface. Application of this concept in field seemso be tedious and therefore not established and validated.

.3. Antiscalant addition

Antiscalants were originally developed for scale inhibition inetal structures for the cooling water boiler industry [120]. The

se of antiscalants is widespread and is an effective method forontrolling scale formation [25,28,59]. One of the major advantagesf antiscalants is the low dosage levels required (i.e. substoichio-etric amounts), which has minimal impact on the feed water

uality. The inhibition of scale formation, therefore, does notnvolve bond formation or breaking between the antiscalant andhe scale forming constituent. With antiscalant addition, scalenhibition occurs by disrupting one or more aspects of the crystal-ization process. Generally antiscalants do not eliminate the scalingonstituents or its tendency; instead they delay the onset crystal-ization (nucleation phase of crystallization) or retard the growthf mineral salt crystals (growth phase of crystallization). Additionf antiscalants increases the effective solubility limits of scalingalts and hence the economic benefit of achieving higher product

ecovery [121].

Application of antiscalants in high pressure membrane sys-ems is an important part of pretreatment needs in desalinationnd water treatment for reuse. Antiscalants work by one or more

ne Science 383 (2011) 1– 16

mechanisms, dictated by their functional groups, dosage level andcharge density [50,78,110,120,122–125]. Some functionalities ofantiscalants are:

- At the submicroscopic level of crystallization, negatively chargedgroups located on the antiscalant molecule target the positivecharges on scale nuclei and distort the ionic balance that prop-agates crystal growth. Thus, antiscalants function as a crystalmodifier and the resulting precipitate will be distorted and turnsless adherent.

- Antiscalants will adsorb to the crystal forming constituents andrepel other ions, keeping them in solution, preventing particlesfrom fixing to anionic charges present on the membrane surface.

- Substoichiometric amounts of antiscalants prevent the precipi-tation of salts once the salt has exceeded its solubility product,called threshold inhibition. The chemical inhibitors may alsoretard the clustering process of charged ions and crystal structure.

- At equimolar amounts of antiscalants and scale forming ions, anti-scalants may act as a chelating agent. Antiscalants form solublecomplex molecules with particular metal ions, inactivating theions so that they cannot react with other elements or ions toproduce precipitates or scale.

Antiscalants are predominantly polyelectrolytes with reportedoptimal molecular weights in the range of 1000–3500 g/mol[126–128]. For polymeric antiscalants, their performance is depen-dent on the functional groups of the molecules, the molecularweight and their charge density. Their effectiveness may also beinfluenced by external factors such as the temperature and pH. Thechoice of a specific antiscalant depends on the feed water quality.Generally commercial antiscalants are designed to target specificscale types. Where antiscalant performance is known to be lim-ited, it is also common to employ a combination of two or moreantiscalants to improve performance. Maintaining the correct anti-scalant dosage level is an important task and increasing dosagelevels does not necessarily minimize precipitation as the presenceof certain precipitates can alter the effectiveness of applied anti-scalant [129,130]. The optimal dosing and allowable maximumsaturation levels of scale forming ions are generally determinedby specific software programmed by suppliers. However, labora-tory tests are recommended to optimize the dosage levels prior tofull scale trials.

Hydranautics reports a general upper saturation limit (as LSI orSI) of foulants with antiscalants. These limits are based on the con-centrated stream with a normal saturation limit of 100% withoutuse of an antiscalant: for CaCO3, LSI of +2.9; for CaSO4, saturationlimit of 400% or SI – 4; for BaSO4, saturation limit of 8000% or SI –80; for SrSO4, saturation limit of 1200% or SI – 12 and for CaF2 –saturation limit of 12,000% or SI – 120 [131].

Commercially available antiscalants can be classified into threemajor categories, phosphates, phosphonates and polycarboxylates.Polyphosphates, especially sodium hexametaphosphate (NaPO3)6(SHMP) was the first antiscalant commercially available to themembrane industry [120]. The antiscalant functionality in phos-phates are due to the 0–P–(0)3 linkage, obtained from condensationreactions of orthophosphoric acid, either in linear or cyclic form[125]. Cyclic polyphosphates have the general formula of (MPO3)n,with M being a monovalent cation such as sodium or hydrogen,while linear polyphosphates have the formula of Mn+2PnO3n+1.Polyphosphates provide many functions including the sequester-ing of iron, manganese and alkali earth metals such as calciumand magnesium [124,125]. In general, sequestering of iron, man-

ganese and calcium requires a 1:1 molar ratio with polyphosphate.Polyphosphates also act as crystal modifiers for calcium and mag-nesium scales at substoichiometric amounts. In theory, SHMP doseof 500 mg/L is required to sequester 200 mg/L of calcium as CaCO3;

embra

hgihwDtiad

Haothitoibhpd

gwmlDgpfntosPattsing

dde

ts

1

2

3

4

A. Antony et al. / Journal of M

owever, a dose of only 2–4 mg/L required to modify the crystalrowth of CaCO3. Amjad [124] reported that SHMP increases thenduction time of CaSO4. The major disadvantage of SHMP is theydrolytic cleavage of the active O–P–P group to orthophosphate,hich can enhance the formation of calcium phosphate scale [120].espite this, SHMP is still used, but is more common in RO plants

hat operate on feed water temperatures below 45 ◦C [123]. SHMPs sometimes found to perform better in RO systems than newerntiscalants, which may be due to higher brine concentrations orifferent hydrodynamics in RO [125].

Phosphonates are the salts and esters of phosphonic acid,PO(OH)2 and are highly water soluble. Phosphonate antiscalantsre important in water treatment applications as they providene or more phosphonic groups which are linked via C–P bondso the organic framework. The C–P bonds are more stable atigher temperatures than the polymeric O–P–O bonding found

n polyphosphates [125]. The stability of the complexes rises ashe number of phosphonic acid groups increase. Phosphonates arebserved to perform better than polyphosphates especially for thenhibition of CaCO3, Mg(OH)2 and BaSO4 scales [54]. Also, the inhi-ition action of phosphonates towards CaCO3 was significantlyigher compared to CaSO4 scale [124,125]. Like polyphosphates,hosphonates are also suspected to form calcium phosphateeposits [43,132,133].

Polycarboxylates are characterized by functional –COOHroups. Polycarboxylate antiscalants are anionic, low moleculareight polyelectrolytes. Polycarboxylates are mostly homopoly-ers and copolymers of acrylic acid and maleic acid, but polymers

ike polymethacrylic acid and polyasparates are also available.ue to polycarboxylates’ anionic nature, these antiscalants areood chelators for multivalent cations, leading to dispersion ofrecipitates and lattice distortion. However, the anti-scalant per-ormance of polycarboxylate is dependent on the molecular weight,umber of carboxyl groups and spatial arrangement of the func-ional groups.Polyacrylates with molecular weights in the rangef 5000–6000 g/mol are the most widely used due to their highcale inhibition power and environmental compatibility [134].olyacrylates are highly efficient in preventing the nucleationnd crystallization of many scale-forming minerals via adsorp-ion of polyacrylates onto developing nuclei [134–136]. Selectinghe optimum concentration of polyacrylates is very importantince a lower concentration of polyacrylates will not effectivelynhibit scale formation, whereas a high concentration is uneco-omical and may additionally have other adverse effects such aselation [134].

Recent advances in antiscalants have seen the emergence ofendrimeric polymers which are highly branched with a three-imensional structure, some of which have been referred asnvironmentally friendly antiscalants [137,138].

In spite of effective scale inhibition action, some of the limita-ions encountered with the use of antiscalants in RO operation areummarized below [110,123,130,139–141]:

. Optimum dosing of antiscalants is essential, otherwise they canbe a foulant at high concentrations.

. Antiscalants were shown to enhance the biofouling potential inRO systems. Some antiscalants can increase biological growth upto 10 times their normal growth rate.

. Carryover of pretreatment chemicals may react with antiscalantsand form foulants or retard antiscaling efficiency. Cationic floc-culants used for pretreatment can particularly react with some

types of antiscalants forming sticky foulants.

. Polyacrylates are membrane foulant in the presence of iron andother metal ions. Similarly, HEDP lose its antiscalant efficiencyat high alkalinities and in the presence of chlorine.

ne Science 383 (2011) 1– 16 11

5. Monitoring the presence of antiscalants in the system is compli-cated, compared to examining acid dosing by pH changes.

7. Non destructive scale monitoring techniques

A comprehensive understanding of scaling at the atomic leveland the development of non destructive real time monitoringfor early detection of scale formation are vital for the successfulmanagement of membrane scaling. Some of the scale monitoringtechniques reported in the past to achieve this is discussed in thefollowing section.

7.1. Permeate flux decline

The industry practice for evaluating membrane performance isto compare permeate flow and salt passage under identical con-ditions, through standardising/normalising industrial data [142],mostly coupled with controlled automation warning systems. Onesuch example is the system reported by Saad et al. functioning as anearly-warning alarm, which benefits plant owners and end-users[143].

Various approaches have been applied at laboratory and pilotscales to identify the critical threshold limits of scaling and thus theoptimum recovery for a given feed water to assess the antiscalingefficiency of chemical additives [121,144,145]. The first approachwhere the feed is passed through the RO system and permeate iscontinuously withdrawn, concentrating the concentration of saltsin the feed. The sudden drop in membrane performance in terms ofpermeability is monitored to identify the threshold limits of scalingand the onset of scale formation. Using this approach, the concen-tration factor accelerates the scaling potential [144]. Hassen et al.adopted a once-through flow method to study scale whose solu-bility levels are low and the amount of scale forming constituentscannot be supplied from the small batch tests. In this method, scal-ing material is continuously provided in sufficient quantities to themembrane by once through flow of feed solution at a controlledsupersaturation level [145]. The scale formation is detected fromthe rate of flux decline and the mass of the scale deposited. An alter-native approach is the intermittent recycle technique, reported toenable convenient control of the scaling potential of the water flow-ing though the membrane system [121]. In this method, membranefiltration is performed with the feed representing a concentrate cor-responding to a particular product recovery level for a fixed periodin a total recycling mode (where both concentrate and permeateare returned to the feed tank). The filtration is repeated for fixedperiods of time to solutions of progressively increasing supersat-uration. Scaling potential is calculated from the flux decline dataand solution composition changes, which are due to precipitationof scalants.

Irrespective of the approach adopted, permeate flux and saltrejection curves have been widely used at the laboratory scaleto assess the antiscalant effectiveness on the scaling process[50,80,123,146,147]. A good antiscalant should retain the initialflux, extend the product recovery, and produce a salt rejection curvethat does not involve significant decline in salt rejection over time.A large fall in flux and salt rejection indicates scale formation. In atypical normalised permeate flux curve, a small initial increase influx is always seen, due to membrane conditioning such as com-paction. The limitation associated with the permeate flux declinecurve is the inability to distinguish between flow decline due toscaling and decline simply due to the increase in osmotic pressure

[148], and therefore normalisation of the flux with respect to ionicstrength is required. However, even then flux decline is insensi-tive to the on-set of scaling, so that scaling of the membrane hasoccurred once an appreciable flux decline has been observed.

1 embra

7

ssausiafabiuiotaioot

ttttsanfFmcqURs

twgsMatafdSeaprdnaptcamiit

2 A. Antony et al. / Journal of M

.2. Ultrasonic time-domain reflectometry analysis

In the ultrasonic time domain reflectometry (UTDR) technique,ound waves are used to measure the location of a moving ortationary interface and derive information on the physical char-cteristics of the media through which the waves travel. When anltrasonic wave is incident on the system interfaces, as the ultra-onic waves meet the interface between two media, the responses the energy partitioning and reflection of the incident wave. Themplitude of the reflected wave relative to the incident wave is aunction of the acoustic impedance difference between the mediand the topography of the interface, and the reflected wave cane measured using an ultrasonic transducer as echoes for every

nterface. If the fouling layer is thick enough to be measured by theltrasonic signal and it falls within the spatial resolution capabil-

ties of the system, a new echo will be detected as a consequencef the bulk solution/fouling interface. The time interval betweenhe incident wave instigation and its reflection can be interpreteds a measure of the distance between the interfaces and used tondicate the thickness of the scale formed [149]. The amplitudef the reflected wave relative to the incident wave is a functionf the acoustic impedance difference between the media and theopography of the interface [150].

Mairal et al. [149] first employed UTDR for the nondestruc-ive, in situ measurement of CaSO4 scaling in real time. Over theimescale of these experiments, due to the low axial flow rates,he scale obtained was not significant enough to produce an echohat was distinct from the initial echo, and therefore time delay ofignals were not achieved. The signal amplitudes were taken as anlternative measure for monitoring fouling. Thus, the ultrasonic sig-al amplitude decline correlated well with the development of the

ouling layer and surface coverage as the permeate flux declined.urther, comprehensive examination of inorganic fouling underore realistic operating conditions were performed to study the

hanges on the membrane surface due to scale growth and subse-uent removal during cleaning [151]. Chong et al. [152] used theTDR technique to monitor the growth of silica fouling layer inO. Also in these experiments, change in amplitude of the reflectedignal was correlated to membrane fouling.

Sanderson et al. investigated calcium carbonate scale forma-ion and removal in flat-sheet RO membrane modules [153]. UTDRas demonstrated to effectively detect fouling-layer initiation, its

rowth on the membrane and its removal, in real-time. The acousticignal response corresponded well with the flux decline behavior.ore importantly, as fouling proceeded, a second echo with greater

mplitude appeared in the time-domain, representing the state ofhe fouling layer on the membrane surface. Increase in the relativemplitude of the fouling echo results from the accumulation of theouling layer. In addition, the difference in the fouling pattern fromead end and crossflow modes were well differentiated by UTDR.imilarly, a fouling echo was obtained for CaSO4 deposition by Lit al. [154]. A polynomial relationship between flux and the rel-tive amplitude was observed under crossflow operation and therogress of the fouling echo in the time domain was noticed as aesult of an increase in the thickness of the fouling layer. The fluxecline was relative to the increases in both the density and thick-ess of the deposit. The changes in the density of a fouling layers well as the thickness were validated by a predictive modellingrogram i.e. ultrasonic reflection modelling. Zhang et al. extendedhe use of UTDR for the measurement of fouling and cleaning inommercial spiral-wound RO membrane modules [155]. Perme-tion rate, acoustic amplitude and arrival time measurements were

ade at regular intervals during clean water equilibration, foul-

ng and cleaning on two modules. During this, systematic changesn the entire acoustic spectrum as a function of module operationime were represented in terms of shift factors. Results showed a

ne Science 383 (2011) 1– 16

reasonable correlation between the permeation rate and typicalUTDR factors, amplitude and arrival time shift factors. In general,UTDR technique provides a sound basis for developing an on-linesensor for the timely detection of inorganic scale formation in com-mercial membrane modules.

7.3. Visual observation

Efforts made for the direct observation of foulant accumulationon the membrane surface and to monitor the dynamics of foulinglayer formation were critically reviewed by Chen et al. [156]. All ofthe techniques studied were applicable to low pressure membraneoperations. For plate and frame RO membrane cells, direct visualobservation and real-time monitoring of mineral surface scaling ofCaSO4 was developed by Uchymiak et al. [157], called the ex situscale observation detector (EXSOD). In this technique, membranesurface imaging under RO operation is achieved by high resolu-tion digital photography using an optical microscope with lightingarrangement to enhance the boundaries of semi-transparent crys-tals and to provide the most straightforward visualization ofmembrane fouling. The experiment consisted of rectangular highpressure RO cells, designed to operate in excess of 2.41 MPa, withdimensions 7.87 (l) × 2.54 (w) × 0.254 (channel height) cm, withan active membrane surface area of 20 cm2 and transparent win-dow of 2.5 cm diameter. High resolution membrane surface imagesof 2048 × 1536 pixels were recorded in real-time using a digitalcamera and an optical microscope (wide field 10–30×). With thesophistication of optical magnification along with the field illu-mination possibilities and digital image acquisition and analysis,capturing even subtle changes of crystal boundaries was achieved.Thus, the nucleation and growth of gypsum crystals was quantifiedin a plate and frame RO membrane channel, detected by real-timeanalysis of the recorded images and the evolution of the surfacenumber density, size of mineral crystals and the percent of surfacearea covered by scale [71,114,157]. This technique was integratedwith a brackish water RO (BWRO) pilot plant, whereby the EXSODdevice was located at the tail of the RO pilot plant and used to detectthe on-set of scaling. The use of the EXSOD device as a scaling sensorwas used to evaluate the feasibility of operating the RO system in afeed flow reversal mode for scale control [114] and used to evaluateantiscalant efficiency retarding scale nucleation and growth [158].

The materialization of direct scale detection can be applied infull scale plants to detect the precise timing of the onset of scaleformation by connecting this detector to a side-stream from thetail element of an RO spiral wound module, quantified by theobserved crystallization induction time or the threshold surfacescale coverage [114,157]. The major limitation with this techniqueis currently the determination of appropriate operating conditionsfor the EXSOD meter, such that the concentration polarization andflux conditions are similar to those in the operating RO plant. Addi-tionally, application of the EXSOD detector has focused on gypsumscaling, and further work is required to demonstrate its applicabil-ity for mixed scale types.

7.4. Electrical impedance spectroscopy analysis

Electrical impedance spectroscopy (EIS) has been proposed asa valuable tool for characterizing and analyzing the structures ofmicroporous membranes various researchers [40,159,160]. Struc-tural properties of ultrafiltration membranes were studied byCoster et al., specifically the permeabilities and thicknesses of the

layers comprising the membrane [161]. In other research by Parket al. [162] fouling of ion exchange membranes and the accumu-lation of salt crystals and other material on the membrane surfacewas studied in real time using this technique. Similarly, EIS was

A. Antony et al. / Journal of Membra

Fm

uu

ommvctnitaibfltiphfmAfci

fsmisapt

8

ftMrasc

ardt

ig. 8. Cross section of high pressure RO membrane module for in situ impedanceeasurements.

sed to differentiate between surface fouling vs. internal fouling ofltrafiltration membranes [159,160].

EIS works by injecting sinusoidal alternating currents at vari-us known frequencies into a system through the membrane andeasuring the voltage (potential difference) developed across theembrane as well as the phase difference between the current and

oltage. From this impedance, parameters such as the phase angle,onductance and capacitance can be determined [163]. Any varia-ion in these properties with frequency can be used to assess theumber and properties of electrical/fouling layers at the membrane

nterface [161,164]. Research by Kavanagh et al. [164] has shownhat changes that take place in and on the surface of membraness a result of membrane separation processes, give rise to changesn the magnitude and relative phase of the impedance of the mem-ranes. EIS measurements in this research were performed using aour cell chamber. The four electrode system overcomes the prob-ems arising from the strong, frequency, dependent impedance ofhe electric double layer at the interface between the metal currentnjecting electrodes and the aqueous environment. However, thisarticular four terminal membrane chamber cannot be operated atigh pressure and, therefore, membrane fouling needs to be per-

ormed externally in a filtration system and then the impedanceeasurements made subsequently in the four terminal chamber.

key limitation with this system is the possible disturbance of theouling layer during the transfer of the membrane and the possibleonsequent loss of the interfacial structure when the measurements made.

EIS could be used for continuous monitoring of membraneouling with appropriate instrumentation to allow in situ mea-urements. A recent development in that direction for impedanceeasurement is a robust high pressure RO system capable of in situ

mpedance measurement at high operating pressures [165]. This ishown in Fig. 8. The stainless steel cell can operate at pressuress high as 5–10 MPa. This unit incorporates two current injectinglates and two transmembrane voltage measuring electrodes. It ishus also a 4-terminal impedance measuring system.

. Concentrate disposal

A major concern associated with addition of chemical additivesor managing the scale formation is the high cost and environmen-al impacts associated with the concentrate management [166].

ostly, the concentrate stream is either chemically treated for theecovery of minerals, disposed or undergoes volume reduction. Inll cases, presence of antiscalants in the brine stream, added forcale mitigation, may cause environmental issues and may signifi-antly alter brine treatment performance, efficiency and cost.

In a two stage process of intermediate chemical deminer-

lization, antiscalant carryover from stage I is observed toetard the crystal growth in an interstage crystallization process,ecreasing the efficiency of desupersaturation [116,118]. Addi-ion of antiscalant scavengers/degraders is therefore necessary to

ne Science 383 (2011) 1– 16 13

accelerate the desupersaturation. However, addition of seed mate-rial increases the kinetics of precipitation and increases the settlingrate. One of the promising approaches for concentrate disposaland to reduce the concentrate volume is through precipitation andsolid/liquid separation. Presence of antiscalants in the concentrateaffects the efficiency and effectiveness of precipitation and the pre-cipitate particle morphology [59,167,168]. Presence of antiscalantsprevents calcium precipitation, even at high saturation indices, andcan reduce precipitate particle size which in turn presents issuesduring the filtration of the precipitated salts.

The potential environmental impacts of seawater desalina-tion projects have been discussed in detail by Lattemann et al.[169]. Generally, the levels of antiscalants are of low risk to themarine environment and aquatic life [169,170]. Potential prob-lems may occur when polyphosphates are used, as they easilyhydrolyze to orthophosphate and cause eutrophication and its sub-sequent effects. Polycarboxylic acid and phosphates are stable frombiodegradation. Acid addition does not affect the marine environ-ment as it reacts with the sea water forming harmless compounds[169].

9. Conclusion

Compared to the increasing research activities in various aspectsof membrane technology and applications, studies related to theuse of chemical additives in membrane systems are limited. Morespecifically, there is a huge research requirement for the use of anti-scalants that has not been addressed so far, not necessarily relatedto their performance alone. Some of the knowledge gaps identifiedfrom the literature review include:

- Scale formation and subsequent performance decline still remaina challenge for RO systems as evidenced from membrane autopsyreports, in spite of the availability of a suite of scale predictiontools and a range of scale mitigation measures practiced in thewater treatment industry.

- Studies on antiscalant suitability and efficiency are generallyperformed with commercially available antiscalants; given thatantiscalant formulations are proprietary in nature, it makesit hard to assess and understand the antiscaling efficiency atthe molecular level for modelling and prediction of outcomes.A greater transparency in terms of the constituents will helpmolecular level understanding of different categories of chemicaladditives to various scale types.

- Research on scale formation and control, especially, assessingantiscalant efficiency has usually been performed with individualand model solutions. Although this helps in assessing the effi-ciency for individual components, it fails to mimic complex waterchemistries, where there is possibility of co-precipitation of morethan one component of unknown proportions. It is apparent thescale usually noticed in reverse osmosis systems operating onnatural waters differs from what would be expected in controlledconditions. Studies aimed with mixed salt solution, real feed typesand incorporating molar ratio approach would help solving this.

- Similarly some of the scale assessment techniques were targetedfor specific antiscalants or scale limiting approaches for particu-lar waters. Rather, these techniques should be made industriousafter method development, by taking them to more complexwater

- Potential to form various scales are generally estimated from thedifferent scale prediction techniques and appropriate antiscalant

is selected for the given feed type, since antiscalants are mostlyselective to mitigate a specific scale type. Therefore, in achievingscale suppression/inhibition of a mixture of scales, generally a for-mulation consisting of a mixture of antiscalants is recommended

1 embra

A

aivi

R

4 A. Antony et al. / Journal of M

by the antiscalant manufacturers. Taking into account the selec-tivity of antiscalant action, the overall inhibition efficiency couldbe a summation of individual efficiencies, a synergistic or anantagonistic effect. Therefore, research is needed in identifyingthe deleterious combinations of antiscalant – feed composition.Although the software models generally provided by the anti-scalant manufacturers account for this, basic understanding atmicroscopic level should be established as this would help inoperation with complicated water mixtures like coal seam gaswaters, where the usual softwares does not help much in pre-dicting the actual scaling propensity.

cknowledgements

The authors thank Water Quality Research Australia (WQRA)nd Water Research Foundation (waterRF) for the financial supportn performing this review. The authors thank Prof. Hans Coster, Uni-ersity of Sydney and Dr. Shane Cox, UNSW for providing valuablenformation to this review.

eferences

[1] H. Mehdizadeh, Membrane desalination plants from an energy-exergy view-point, Desalination 191 (2006) 200–209.

[2] M.N.A. Hawlader, J.C. Ho, T. Chua Kok, Desalination of seawater: an experi-ment with RO membranes, Desalination 132 (2000) 275–280.

[3] R.W. Atwater, L. Palmquist, J. Onkka, The West Basin desalter project: a viablealternative, Desalination 103 (1995) 117–125.

[4] P. Gagliardo, S. Adham, R. Trussell, A. Olivieri, Water repurification via reverseosmosis, Desalination 117 (1998) 73–78.

[5] M. Abdel-Jawad, S. Al-Shammari, J. Al-Sulaimi, Non-conventional treatmentof treated municipal wastewater for reverse osmosis, Desalination 142 (2002)11–18.

[6] M.D. Afonso, J.O. Jaber, M.S. Mohsen, Brackish groundwater treatment byreverse osmosis in Jordan, Desalination 164 (2004) 157–171.

[7] K.T. Chua, M.N.A. Hawlader, A. Malek, Pretreatment of seawater: results ofpilot trials in Singapore, Desalination 159 (2003) 225–243.

[8] NWC, Emerging trends in desalination: a review, Waterlines report No. 9.,National Water commission, 2008.

[9] S. Lattemann, M.D. Kennedy, J.C. Schippers, G. Amy, Chapter 2 global desali-nation situation, in: C.E. Isabel, I.S. Andrea (Eds.), Sustainability Science andEngineering, Elsevier, 2010, pp. 7–39.

[10] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosisdesalination: water sources, technology, and today’s challenges, Water Res.43 (2009) 2317–2348.

[11] T. Pankratz, Membrane growth takes off, in: T. Pankratz (Ed.), Water Desali-nation Report, Media Analytics Ltd., Oxford, UK in Cooperation with GlobalWater Intelligence, Houston, TX, 2008.

[12] W.-Y. Shih, A. Rahardianto, R.-W. Lee, Y. Cohen, Morphometric character-ization of calcium sulfate dihydrate (gypsum) scale on reverse osmosismembranes, J. Membr. Sci. 252 (2005) 253–263.

[13] C.J. Gabelich, M.D. Williams, A. Rahardianto, J.C. Franklin, Y. Cohen,High-recovery reverse osmosis desalination using intermediate chemicaldemineralization, J. Membr. Sci. 301 (2007) 131–141.

[14] T.H. Chong, F.S. Wong, A.G. Fane, Enhanced concentration polarization byunstirred fouling layers in reverse osmosis: detection by sodium chloridetracer response technique, J. Membr. Sci. 287 (2007) 198–210.

[15] Y.A. Le Gouellec, M. Elimelech, Calcium sulfate (gypsum) scaling in nanofil-tration of agricultural drainage water, J. Membr. Sci. 205 (2002) 279–291.

[16] S. Hong, M. Elimelech, Chemical and physical aspects of natural organic mat-ter (NOM) fouling of nanofiltration membranes, J. Membr. Sci. 132 (1997)159–181.

[17] A.R. Bartman, E. Lyster, R. Rallo, P.D. Christofides, Y. Cohen, Mineral scalemonitoring for reverse osmosis desalination via real-time membrane surfaceimage analysis, Desalination 273 (2011) 64–71.

[18] E.M.V. Hoek, J. Allred, T. Knoell, B.-H. Jeong, Modeling the effects of fouling onfull-scale reverse osmosis processes, J. Membr. Sci. 314 (2008) 33–49.

[19] C.Y. Tang, T.H. Chong, A.G. Fane, Colloidal interactions and fouling of NF andRO membranes: a review, Adv. Colloid Interface Sci. 164 (2011) 126–143.

[20] T. Tran, B. Bolto, S. Gray, M. Hoang, E. Ostarcevic, An autopsy study of a fouledreverse osmosis membrane element used in a brackish water treatment plant,Water Res. 41 (2007) 3915–3923.

[21] T.R. Bott, Fouling of Heat Exchangers, Elsevier, 1995.

[22] J.C. Cowan, D.J. Weintritt, Water-formed Scale Deposits, Gulf Pub. Co., Book

Division, 1976.[23] C.A.C. van de Lisdonk, J.A.M. van Paassen, J.C. Schippers, Monitoring scaling

in nanofiltration and reverse osmosis membrane systems, Desalination 132(2000) 101–108.

ne Science 383 (2011) 1– 16

[24] J. Gilron, D. Hasson, Calcium sulphate fouling of reverse osmosis membranes:flux decline mechanism, Chem. Eng. Sci. 42 (1987) 2351–2360.

[25] S. Lee, J. Kim, C.-H. Lee, Analysis of CaSO4 scale formation mechanism invarious nanofiltration modules, J. Membr. Sci. 163 (1999) 63–74.

[26] S. Lee, C.-H. Lee, Effect of operating conditions on CaSO4 scale formation mech-anism in nanofiltration for water softening, Water Res. 34 (2000) 3854–3866.

[27] S. Lee, C.H. Lee, Scale formation in NF/RO: mechanism and control, Water Sci.Technol. 51 (2005) 267–275.

[28] A.G. Pervov, Scale formation prognosis and cleaning procedure schedules inreverse osmosis systems operation, Desalination 83 (1991) 77–118.

[29] D. Hasson, A. Drak, R. Semiat, Inception of CaSO4 scaling on RO membranesat various water recovery levels, Desalination 139 (2001) 73–81.

[30] M. Okazaki, S. Kimura, Scale formation on reverse-osmosis membranes, J.Chem. Eng. Jpn. 17 (1984) 145–151.

[31] S. Shirazi, C.-J. Lin, D. Chen, Inorganic fouling of pressure-driven membraneprocesses—a critical review, Desalination 250 (2010) 236–248.

[32] P.G. Klepetsanis, P.G. Koutsoukos, Precipitation of calcium sulfate dihydrateat constant calcium activity, J. Cryst. Growth 98 (1989) 480–486.

[33] Z. Amjad, J. Hooley, Influence of polyelectrolytes on the crystal growth ofcalcium sulfate dihydrate, J. Colloid Interface Sci. 111 (1986) 496–503.

[34] D.H. Troup, J.A. Richardson, Scale nucleation on a heat transfer surface and itsprevention, Chem. Eng. Commun. 2 (1978) 167–180.

[35] X.-L. Wang, C. Zhang, P. Ouyang, The possibility of separating saccharides froma NaCl solution by using nanofiltration in diafiltration mode, J. Membr. Sci.204 (2002) 271–281.

[36] S.K. Hamdona, R.B. Nessim, S.M. Hamza, Spontaneous precipitation of calciumsulphate dihydrate in the presence of some metal ions, Desalination 94 (1993)69–80.

[37] R. Sheikholeslami, Mixed salts—scaling limits and propensity, Desalination154 (2003) 117–127.

[38] T.H. Chong, R. Sheikholeslami, Thermodynamics and kinetics for mixed cal-cium carbonate and calcium sulfate precipitation, Chem. Eng. Sci. 56 (2001)5391–5400.

[39] P. Dydo, M. Turek, J. Ciba, Scaling analysis of nanofiltration systems fed withsaturated calcium sulfate solutions in the presence of carbonate ions, Desali-nation 159 (2003) 245–251.

[40] E.M.V. Hoek, M. Elimelech, Cake-enhanced concentration polarization: a newfouling mechanism for salt-rejecting membranes, Environ. Sci. Technol. 37(2003) 5581.

[41] E.M.V. Hoek, A.S. Kim, M. Elimelech, Influence of crossflow membrane filtergeometry and shear rate on colloidal fouling in reverse osmosis and nanofil-tration separations, Environ. Eng. Sci. 19 (2002) 357.

[42] H.-J. Oh, Y.-K. Choung, S. Lee, J.-S. Choi, T.-M. Hwang, J.H. Kim, Scale forma-tion in reverse osmosis desalination: model development, Desalination 238(2009) 333–346.

[43] F.H. Butt, F. Rahman, U. Baduruthamal, Identification of scale deposits throughmembrane autopsy, Desalination 101 (1995) 219–230.

[44] E.G. Darton, RO plant experiences with high silica waters in the Canary Islands,in: Conference on Desalination and the Environment, European DesalinationSociety and the International Water Services Association, Las Palmas, GranCanaria, 1999.

[45] E.G. Darton, M. Fazell, A statistical review of 150 membrane autopsies, in:62nd Annual International Water Conference, Pittsburgh, 2001.

[46] M. Karime, S. Bouguecha, B. Hamrouni, RO membrane autopsy of Zarzis brack-ish water desalination plant, Desalination 220 (2008) 258–266.

[47] T. Darton, U. Annunziata, F. del Vigo Pisano, S. Gallego, Membrane autopsyhelps to provide solutions to operational problems, Desalination 167 (2004)239–245.

[48] P. Xu, C. Bellona, J.E. Drewes, Fouling of nanofiltration and reverse osmosismembranes during municipal wastewater reclamation: membrane autopsyresults from pilot-scale investigations, J. Membr. Sci. 353 (2010) 111–121.

[49] F.H. Butt, F. Rahman, U. Baduruthamal, Hollow fine fiber vs. spiral-woundreverse osmosis desalination membranes. Part 2. Membrane autopsy, Desali-nation 109 (1997) 83–94.

[50] C. Tzotzi, T. Pahiadaki, S.G. Yiantsios, A.J. Karabelas, N. Andritsos, A study ofCaCO3 scale formation and inhibition in RO and NF membrane processes, J.Membr. Sci. 296 (2007) 171–184.

[51] J. Pena, B. Buil, A. Garralón, P. Gómez, M.J. Turrero, A. Escribano, G. Garralón,M.A. Gómez, The vaterite saturation index can be used as a proxy of the S&DSIin sea water desalination by reverse osmosis process, Desalination 254 (2010)75–79.

[52] A.G. Xyla, J. Mikroyannidis, P.G. Koutsoukos, The inhibition of calcium car-bonate precipitation in aqueous media by organophosphorus compounds, J.Colloid Interface Sci. 153 (1992) 537–551.

[53] H. Elfil, H. Roques, Role of hydrate phases of calcium carbonate on the scalingphenomenon, Desalination 137 (2001) 177–186.

[54] K. Sawada, The mechanism of crystallization and transformation of calciumcarbonates, Pure Appl. Chem. 69 (1997) 921–928.

[55] T. Ogino, T. Suzuki, K. Sawada, The formation and transformation mecha-nism of calcium carbonate in water, Geochim. Cosmochim. Acta 51 (1987)2757–2767.

[56] D. Chakraborty, V.K. Agarwal, S.K. Bhatia, J. Bellare, Steady-state transitionsand polymorph transformations in continuous precipitation of calcium car-bonate, Ind. Eng. Chem. Res. 33 (1994) 2187–2197.

[57] Y.B. Kitano, Morphology and structure of CaCO3 Scale layers formed underisothermal flow conditions, Chem. Soc. Jpn. 35 (1962) 1973.

embra

A. Antony et al. / Journal of M

[58] M.K.N. Andritsos, A.J. Karabelas, P.G. Koutsoukos, CaCO3 deposit formationunder isothermal conditions, Can. J. Chem. Eng. 74 (1996) 911–919.

[59] L.F. Greenlee, F. Testa, D.F. Lawler, B.D. Freeman, P. Moulin, The effect ofantiscalant addition on calcium carbonate precipitation for a simplified syn-thetic brackish water reverse osmosis concentrate, Water Res. 44 (2010)2957–2969.

[60] J.D. Rodriguez-Blanco, S. Shaw, L.G. Benning, The kinetics and mechanismsof amorphous calcium carbonate (ACC) crystallization to calcite, via vaterite,Nanoscale 3 (2011) 265–271.

[61] M.M. Reddy, G.H. Nancollas, The crystallization of calcium carbonate: IV. Theeffect of magnesium, strontium and sulfate ions, J. Cryst. Growth 35 (1976)33–38.

[62] H. Roques, A. Girou, Kinetics of the formation conditions of carbonate tartars,Water Res. 8 (1974) 907–920.

[63] J.M. Astilleros, L. Fernández-Díaz, A. Putnis, The role of magnesium in thegrowth of calcite: an AFM study, Chem. Geol. 271 (2010) 52–58.

[64] A. Stashans, G. Chamba, A new insight on the role of Mg in calcite, Int. J.Quantum Chem. 111 (2011) 2436–2443.

[65] T. Chen, A. Neville, M. Yuan, Influence of Mg2+ on CaCO3 formation–bulkprecipitation and surface deposition, Chem. Eng. Sci. 61 (2006) 5318–5327.

[66] A. Mucci, J.W. Morse, The incorporation of Mg2+ and Sr2+ into calcite over-growths: influences of growth rate and solution composition, Geochim.Cosmochim. Acta 47 (1983) 217–233.

[67] UNSW, Internal reports, Centre For Membrane Science and Technology, Mem-brane Autopsy reports.

[68] J.N.A. Lewis, S. Seewoo, S. Lacour, Prevention of Scaling in Mine Water UsingSlurry Precipitation and Recycle Reverse Osmosis (SPARRO) 15th Interna-tional Symposium on Industrial Crystallisation, 2002.

[69] M.R. Christoffersen, J. Christoffersen, M.P.C. Weijnen, G.M. Van Rosmalen,Crystal growth of calcium sulphate dihydrate at low supersaturation, J. Cryst.Growth 58 (1982) 585–595.

[70] R. Sheikholeslami, Calcium sulfate fouling-precipitation of particulate: a pro-posed composite model, Heat Transfer Eng. 21 (2000) 24.

[71] M. Uchymiak, E. Lyster, J. Glater, Y. Cohen, Kinetics of gypsum crystal growthon a reverse osmosis membrane, J. Membr. Sci. 314 (2008) 163–172.

[72] S. Seewoo, R. Van Hille, A. Lewis, Aspects of gypsum precipitation in scalingwaters, Hydrometallurgy 75 (2004) 135–146.

[73] J.E.L.a.G. Burns, Heats of crystallization of CaSO4·2H2O, Bull. Soc. Chim. Belg.93 (1984) 271–279.

[74] S.H. Najibi, H. Müller-Steinhagen, M. Jamialahmadi, Calcium sulphatescale formation during subcooled flow boiling, Chem. Eng. Sci. 52 (1997)1265–1284.

[75] W.-Y. Shih, J. Gao, A. Rahardianto, J. Glater, Y. Cohen, C.J. Gabelich, Rankingof antiscalant performance for gypsum scale suppression in the presence ofresidual aluminum, Desalination 196 (2006) 280–292.

[76] M. Brusilovsky, J. Borden, D. Hasson, Flux decline due to gypsum precipitationon RO membranes, Desalination 86 (1992) 187–222.

[77] R. Sheikholeslami, H.W.K. Ong, Kinetics and thermodynamics of calcium car-bonate and calcium sulfate at salinities up to 1.5 M, Desalination 157 (2003)217–234.

[78] S.P. Chesters, Innovations in the inhibition and cleaning of reverse osmosismembrane scaling and fouling, Desalination 238 (2009) 22–29.

[79] G. Greenberg, D. Hasson, R. Semiat, Limits of RO recovery imposed by calciumphosphate precipitation, Desalination 183 (2005) 273–288.

[80] A. Zach-Maor, R. Semiat, A. Rahardianto, Y. Cohen, S. Wilson, S.R. Gray, Diag-nostic analysis of RO desalting treated wastewater, Desalination 230 (2008)239–247.

[81] L. Wang, G.H. Nancollas, Calcium orthophosphates: crystallization and disso-lution, ChemInform 40 (2009).

[82] J.D. Termine, R.A. Peckauskas, A.S. Posner, Calcium phosphate formation invitro: II. Effects of environment on amorphous-crystalline transformation,Arch. Biochem. Biophys. 140 (1970) 318–325.

[83] A.L. Boskey, A.S. Posner, Conversion of amorphous calcium phosphate tomicrocrystalline hydroxyapatite. A pH-dependent, solution-mediated, solid-solid conversion, J. Phys. Chem. 77 (1973) 2313–2317.

[84] M. Maurer, M. Boller, Modelling of phosphorus precipitation in wastewatertreatment plants with enhanced biological phosphorus removal, Water Sci.Technol. 39 (1999) 147–163.

[85] S. Kubo, T. Takahashi, H. Morinaga, H. Ueki, Inhibition of calcium phosphatescale on heat exchanger: the relation between laboratory test results and testson heat transfer surfaces, Corrosion 79, Atlanta, 1979, p. 15.

[86] Filmtec, FilmTec Membranes Technical Manual, The Dow Chemical Company,1995.

[87] R.Y. Ning, T.L. Troyer, Colloidal fouling of RO membranes following MF/UF inthe reclamation of municipal wastewater, Desalination 208 (2007) 232–237.

[88] J.J. Qin, M.N. Wai, M.H. Oo, K.A. Kekre, H. Seah, Impact of anti-scalant on foulingof reverse osmosis membranes in reclamation of secondary effluent, WaterSci. Technol. 60 (2009) 2767–2774.

[89] Y. Zhao, L. Song, S.L. Ong, Fouling behavior and foulant characteristics ofreverse osmosis membranes for treated secondary effluent reclamation, J.Membr. Sci. 349 (2010) 65–74.

[90] D. Filmtec, FilmTec Membranes Technical Manual, The Dow Chemical Com-pany, 1995.

[91] S.F.E. Boerlage, M.D. Kennedy, G. Jan Witkamp, J. Peter van der Hoek, J.C. Schip-pers, BaSO4 solubility prediction in reverse osmosis membrane systems, J.Membr. Sci. 159 (1999) 47–59.

ne Science 383 (2011) 1– 16 15

[92] S.F.E. Boerlage, M.D. Kennedy, I. Bremere, G.J. Witkamp, J.P. Van der Hoek,J.C. Schippers, The scaling potential of barium sulphate in reverse osmosissystems, J. Membr. Sci. 197 (2002) 251–268.

[93] R.P. Schneider, L.M. Ferreira, P. Binder, J.R. Ramos, Analysis of foulant layer inall elements of an RO train, J. Membr. Sci. 261 (2005) 152–162.

[94] I. Nishida, Y. Shimada, T. Saito, Y. Okaue, T. Yokoyama, Effect of aluminum onthe deposition of silica scales in cooling water systems, J. Colloid Interface Sci.335 (2009) 18–23.

[95] J.S. Gill, Inhibition of silica–silicate deposit in industrial waters, Colloids Surf.A: Physicochem. Eng. Aspects 74 (1993) 101–106.

[96] I. Bremere, M. Kennedy, S. Mhyio, A. Jaljuli, G.-J. Witkamp, J. Schippers, Preven-tion of silica scale in membrane systems: removal of monomer and polymersilica, Desalination 132 (2000) 89–100.

[97] S.I. Graham, R.L. Reitz, C.E. Hickman, Improving reverse osmosis performancethrough periodic cleaning, Desalination 74 (1989) 113–124.

[98] M. Pontié, S. Rapenne, A. Thekkedath, J. Duchesne, V. Jacquemet, J. Leparc, H.Suty, Tools for membrane autopsies and antifouling strategies in seawaterfeeds: a review, Desalination 181 (2005) 75–90.

[99] M. Luo, Z. Wang, Complex fouling and cleaning-in-place of a reverse osmosisdesalination system, Desalination 141 (2001) 15–22.

[100] ASTM D 4993-08, Standard Practice for Calculation and Adjustment of Silica(SiO2) Scaling for Reverse Osmosis, ASTM International, 2008.

[101] ASTM, D3739-06 Standard Practice for Calculation and Adjustment of Lan-gelier Saturation Index for Reverse Osmosis, ASTM International, WestConshohocken, PA, 2003, 2010.

[102] J.G. Pickin, S.T.S. Yuen, H. Hennings, Waste management options to reducegreenhouse gas emissions from paper in Australia, Atmos. Environ. 36 (2002)741–752.

[103] ASTM, D4582-10 Standard Practice for Calculation and Adjustment of theStiff and Davis Stability Index for Reverse Osmosis, ASTM International, WestConshohocken, PA, 2003, 2010.

[104] C.A.C. van de Lisdonk, B.M. Rietman, S.G.J. Heijman, G.R. Sterk, J.C. Schip-pers, Prediction of supersaturation and monitoring of scaling in reverseosmosis and nanofiltration membrane systems, Desalination 138 (2001)259–270.

[105] M.A. Marwan, N.H. Aly, M.S. Saadawy, Simple code for the estimation of scal-ing potential, Desalination 101 (1995) 279–286.

[106] J.W.D.A.A.G. Collins, Solubility of barium and strontium sulfates in strongelectrolyte solutions, Environ. Sci. Technol. 5 (1971) 10.

[107] S. Boerlage, Scaling and Particulate Fouling in Membrane Filtration Systems,Taylor and Francis, 2001.

[108] S. El-Manharawy, A. Hafez, Molar ratios as a useful tool for prediction ofscaling potential inside RO systems, Desalination 136 (2001) 243–254.

[109] W. Ma, Y. Zhao, L. Wang, The pretreatment with enhanced coagulation and aUF membrane for seawater desalination with reverse osmosis, Desalination203 (2007) 256–259.

[110] E.H. Kelle Zeiher, B. Ho, K.D. Williams, Novel antiscalant dosing control, Desali-nation 157 (2003) 209–216.

[111] D. Mukhopadhyay, Method and apparatus for high efficiency reverse osmosisoperation. US Patent 6,537,456, 2003.

[112] N. Prihasto, Q.-F. Liu, S.-H. Kim, Pre-treatment strategies for seawater desali-nation by reverse osmosis system, Desalination 249 (2009) 308–316.

[113] G. Lauer, Conditioning process and device for producing pure water. US Patentapplication 5,690,829, December 9, 1993, 1997.

[114] M. Uchymiak, A.R. Bartman, N. Daltrophe, M. Weissman, J. Gilron, P.D.Christofides, W.J. Kaiser, Y. Cohen, Brackish water reverse osmosis (BWRO)operation in feed flow reversal mode using an ex situ scale observation detec-tor (EXSOD), J. Membr. Sci. 341 (2009) 60–66.

[115] N. Pomerantz, Y. Ladizhansky, E. Korin, M. Waisman, N. Daltrophe, J. Gilron,Prevention of scaling of reverse osmosis membranes by zeroing the elapsednucleation time. Part I. Calcium sulfate, Ind. Eng. Chem. Res. 45 (2006)2008–2016.

[116] A. Rahardianto, J. Gao, C.J. Gabelich, M.D. Williams, Y. Cohen, High recoverymembrane desalting of low-salinity brackish water: integration of acceler-ated precipitation softening with membrane RO, J. Membr. Sci. 289 (2007)123–137.

[117] C.J. Gabelich, A. Rahardianto, C.R. Northrup, T.I. Yun, Y. Cohen, Processevaluation of intermediate chemical demineralization for water recoveryenhancement in production-scale brackish water desalting, Desalination 272(2011) 36–45.

[118] A. Rahardianto, B.C. McCool, Y. Cohen, Accelerated desupersaturation ofreverse osmosis concentrate by chemically-enhanced seeded precipitation,Desalination 264 (2010) 256–267.

[119] S. Lee, R.M. Lueptow, Control of scale formation in reverse osmosis by mem-brane rotation, Desalination 155 (2003) 131–139.

[120] E.G. Darton, Membrane chemical research: centuries apart, Desalination 132(2000) 121–131.

[121] A. Drak, K. Glucina, M. Busch, D. Hasson, J.-M. Laîne, R. Semiat, Laboratory tech-nique for predicting the scaling propensity of RO feed waters, Desalination132 (2000) 233–242.

[122] M. Al-Ahmad, F.A. Aleem, Scale formation and fouling problems effect on the

performance of MSF and RO desalination plants in Saudi Arabia, Desalination93 (1993) 287–310.

[123] M. Al-Shammiri, M. Safar, M. Al-Dawas, Evaluation of two different antis-calants in real operation at the Doha research plant, Desalination 128 (2000)1–16.

1 embra

6 A. Antony et al. / Journal of M

[124] Z. Amjad, Applications of antiscalants to control calcium sulfate scaling inreverse osmosis systems, Desalination 54 (1985) 263–276.

[125] S. Ghani, N.S. Al-Deffeeri, Impacts of different antiscalant dosing rates andtheir thermal performance in Multi Stage Flash (MSF) distiller in Kuwait,Desalination 250 (2010) 463–472.

[126] K. Farahbakhsh, C. Svrcek, R.K. Guest, D.W. Smith, A review of the impactof chemical pretreatment on low-pressure water treatment membranes, J.Environ. Eng. Sci. 3 (2004) 237–253.

[127] L. Zou, P. Sanciolo, G. Leslie, Using MF-NF-RO train to produce low salt and highnutrient value recycled water for agricultural irrigation, Water Sci. Technol.(2008) 1837–1840.

[128] W.Y. Shih, J. Gao, A. Rahardianto, J. Glater, Y. Cohen, Ranking of antiscalant per-formance for gypsum scale suppression in the presence of residual aluminum,Desalination (2006).

[129] R. Semiat, I. Sutzkover, D. Hasson, Characterization of the effectiveness ofsilica anti-scalants, Desalination 159 (2003) 11–19.

[130] E.E.A. Ghafour, Enhancing RO system performance utilizing antiscalants,Desalination 153 (2003) 149–153.

[131] Hydranautics, Chemical treatment for RO and NF, Technical application bul-letin no. 111, 2003.

[132] J.S. Gill, A novel inhibitor for scale control in water desalination, Desalination124 (1999) 43–50.

[133] H.A. El Dahan, H.S. Hegazy, Gypsum scale control by phosphate ester, Desali-nation 127 (2000) 111–118.

[134] A. Yuchi, Y. Gotoh, S. Itoh, Potentiometry of effective concentration of poly-acrylate as scale inhibitor, Anal. Chim. Acta 594 (2007) 199–203.

[135] S.B. Ahmed, M.M. Tlili, M.B. Amor, Influence of a polyacrylate antiscalant ongypsum nucleation and growth, Cryst. Res. Technol. 43 (2008) 935–942.

[136] A. Jada, R. Ait Akbour, C. Jacquemet, J.M. Suau, O. Guerret, Effect of sodiumpolyacrylate molecular weight on the crystallogenesis of calcium carbonate,J. Cryst. Growth 306 (2007) 373–382.

[137] K.D. Demadis, A structure/function study of polyaminoamide dendrimersas silica scale growth inhibitors, J. Chem. Technol. Biotechnol. 80 (2005)630–640.

[138] K.D. Demadis, E. Neofotistou, E. Mavredaki, M. Tsiknakis, E.-M. Sarigiannidou,S.D. Katarachia, Inorganic foulants in membrane systems: chemical controlstrategies and the contribution of green chemistry, Desalination 179 (2005)281–295.

[139] M.-M. Kim, J. Au, A. Rahardianto, J. Glater, Y. Cohen, F.W. Gerringer, C.J.Gabelich, Impact of conventional water treatment coagulants on mineralscaling in RO desalting of brackish water, Ind. Eng. Chem. Res. 48 (2009)3126–3135.

[140] J.S. Vrouwenvelder, S.A. Manolarakis, H.R. Veenendaal, D. van der Kooij, Bio-fouling potential of chemicals used for scale control in RO and NF membranes,Desalination 132 (2000) 1–10.

[141] A. Plottu-Pecheux, B. Houssais, C. Democrate, D. Gatel, C. Parron, J. Cavard,Comparison of three antiscalants, as applied to the treatment of water fromthe River Oise, Desalination 145 (2002) 273–280.

[142] ASTM, D 4516-00 Standard Practice for Standardizing Reverse Osmosis Per-formance Data, in, ASTM International, West Conshohocken, PA, 2003, 2010.

[143] M. Amin Saad, Early discovery of RO membrane fouling and real-time mon-itoring of plant performance for optimizing cost of water, Desalination 165(2004) 183–191.

[144] F.H. Butt, F. Rahman, U. Baduruthamal, Pilot plant evaluation of advanced vs.conventional scale inhibitors for RO desalination, Desalination 103 (1995)189–198.

[145] D. Hasson, R. Semiat, D. Bramson, M. Busch, B. Limoni-Relis, Suppression ofCaCO3 scale deposition by anti-scalants, Desalination 118 (1998) 285–296.

[146] R. Semiat, I. Sutzkover, D. Hasson, Technique for evaluating silica scaling andits inhibition in RO desalting, Desalination 140 (2001) 181–193.

[147] C.J. Gabelich, K.P. Ishida, F.W. Gerringer, R. Evangelista, M. Kalyan, I.H.M. Suf-

fet, Control of residual aluminum from conventional treatment to improvereverse osmosis performance, Desalination 190 (2006) 147–160.

[148] H. Li, A.G. Fane, H.G.L. Coster, S. Vigneswaran, Direct observation of parti-cle deposition on the membrane surface during crossflow microfiltration, J.Membr. Sci. 149 (1998) 83–97.

ne Science 383 (2011) 1– 16

[149] A.P. Mairal, A.R. Greenberg, W.B. Krantz, L.J. Bond, Real-time measurementof inorganic fouling of RO desalination membranes using ultrasonic time-domain reflectometry, J. Membr. Sci. 159 (1999) 185–196.

[150] G.Y. Chai, A.R. Greenberg, W.B. Krantz, Ultrasound, gravimetric, and SEM stud-ies of inorganic fouling in spiral-wound membrane modules, Desalination 208(2007) 277–293.

[151] A.P. Mairal, A.R. Greenberg, W.B. Krantz, Investigation of membrane foulingand cleaning using ultrasonic time-domain reflectometry, Desalination 130(2000) 45–60.

[152] T.H. Chong, F.S. Wong, A.G. Fane, Fouling in reverse osmosis: detection bynon-invasive techniques, Desalination 204 (2007) 148–154.

[153] R. Sanderson, J. Li, L.J. Koen, L. Lorenzen, Ultrasonic time-domain reflectom-etry as a non-destructive instrumental visualization technique to monitorinorganic fouling and cleaning on reverse osmosis membranes, J. Membr. Sci.207 (2002) 105–117.

[154] J. Li, L.J. Koen, D.K. Hallbauer, L. Lorenzen, R.D. Sanderson, Interpretation ofcalcium sulfate deposition on reverse osmosis membranes using ultrasonicmeasurements and a simplified model, Desalination 186 (2005) 227–241.

[155] Z.X. Zhang, A.R. Greenberg, B. Krantz, Study of membrane fouling and cleaningin spiral wound modules using ultrasonic time-domain reflectometry, in: D.Bhattacharyya, D. Butterfield (Eds.), New Insights into Membrane Science andTechnology: Polymeric and Biofunctional Membranes, Elsevier, 2003.

[156] J.C. Chen, Q. Li, M. Elimelech, In situ monitoring techniques for concentra-tion polarization and fouling phenomena in membrane filtration, Adv. ColloidInterface Sci. 107 (2004) 83–108.

[157] M. Uchymiak, A. Rahardianto, E. Lyster, J. Glater, Y. Cohen, A novel RO ex situscale observation detector (EXSOD) for mineral scale characterization andearly detection, J. Membr. Sci. 291 (2007) 86–95.

[158] E. Lyster, M.-M. Kim, J. Au, Y. Cohen, A Method for evaluating antiscalantretardation of crystal nucleation and growth on RO membranes, J. Membr.Sci. 364 (2010) 122–131.

[159] L. Gaedt, T.C. Chilcott, M. Chan, T. Nantawisarakul, A.G. Fane, H.G.L. Coster,Electrical impedance spectroscopy characterisation of conducting mem-branes: II. Experimental, J. Membr. Sci. 195 (2002) 169–180.

[160] T.C. Chilcott, M. Chan, L. Gaedt, T. Nantawisarakul, A.G. Fane, H.G.L. Coster,Electrical impedance spectroscopy characterisation of conducting mem-branes: I. Theory, J. Membr. Sci. 195 (2002) 153–167.

[161] H.G.L. Coster, K.J. Kim, K. Dahlan, J.R. Smith, C.J.D. Fell, Characterisation ofultrafiltration membranes by impedance spectroscopy. I. Determination ofthe separate electrical parameters and porosity of the skin and sublayers, J.Membr. Sci. 66 (1992) 19–26.

[162] J.S. Park, T.C. Chilcott, H.G.L. Coster, S.H. Moon, Characterization of BSA-foulingof ion-exchange membrane systems using a subtraction technique for lumpeddata, J. Membr. Sci. 246 (2005) 137–144.

[163] T. Sørensen, Surface Chemistry and Electrochemistry of Membranes, M.Dekker, 1999.

[164] J.M. Kavanagh, S. Hussain, T.C. Chilcott, H.G.L. Coster, Fouling of reverse osmo-sis membranes using electrical impedance spectroscopy: Measurements andsimulations, Desalination 236 (2009) 187–193.

[165] Inphaze 2010. Inphaze Pty Ltd. [Online]. Available: www.inphaze.com.au(accessed 12.02.2011).

[166] S.J. Khan, D. Murchland, M. Rhodes, T.D. Waite, Management of concentratedwaste streams from high-pressure membrane water treatment systems, Crit.Rev. Environ. Sci. Technol. 39 (2009) 367–415.

[167] L.F. Greenlee, F. Testa, D.F. Lawler, B.D. Freeman, P. Moulin, Effect of anti-scalants on precipitation of an RO concentrate: metals precipitated andparticle characteristics for several water compositions, Water Res. 44 (2010)2672–2684.

[168] L.F. Greenlee, F. Testa, D.F. Lawler, B.D. Freeman, P. Moulin, Effect of anti-scalant degradation on salt precipitation and solid/liquid separation of ROconcentrate, J. Membr. Sci. 366 (2011) 48–61.

[169] S. Lattemann, T. Höpner, Environmental impact and impact assessment ofseawater desalination, Desalination 220 (2008) 1–15.

[170] O.A. Hamed, H.A. Al-Otaibi, Prospects of operation of MSF desalinationplants at high TBT and low antiscalant dosing rate, Desalination 256 (2010)181–189.