Desalination 344 (2014) 36–47

Contents lists available at ScienceDirect

Desalination

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

Three strategies to treat reverse osmosis brine and cation exchange spentregenerant to increase system recovery

S. Salvador Cob a,b,c,⁎, F.E. Genceli Güner d, B. Hofs a, J. van Spronsen b, G.J. Witkamp a,b,c

a KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlandsb Biotechnology Dept., Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlandsc Wetsus, Centre for Sustainable Water Technology, P.O. Box 1113, 8900 CC Leeuwarden, The Netherlandsd Process & Energy Dept., Delft University of Technology, Leeghwaterstraat 44, 2628 CA Delft, The Netherlands

H I G H L I G H T S

• Three different strategies to treat concentrate and regenerant were investigated.• With the application of EFC pure ice and pure salt were recovered.• Combination of NF, RO and EFC allows near zero waste discharge.

⁎ Corresponding author at: KWRWatercycle Research IBox 1072, 3430 BB Nieuwegein, The Netherlands. Tel.: +

E-mail address: sara.salvador.cob@kwrwater.nl (S. Sal

http://dx.doi.org/10.1016/j.desal.2014.03.0090011-9164/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 December 2013Received in revised form 4 March 2014Accepted 6 March 2014Available online 30 March 2014

Keywords:Eutectic freeze crystallizationZero liquid dischargeReverse osmosisCation exchange regenerate

Concentrate streams from industrial processes entail an important risk for the environment, as they are oftendischarged into it. Therefore, effortsmust bemade to reduce the load of contaminants discharged to the environ-ment. Eutectic freeze crystallization (EFC) is a novel technique which allows separation of salts and water fromaqueous streams. In this research, three treatment options for reverse osmosis (RO) concentrate and cationexchange (CIEX) spent regenerant are investigated.First, application of EFC to RO concentratewas investigated. The streamwas rich inHCO3

− andNa+. Application ofEFC to this solution, led to the formation of ice and NaHCO3 at −3.9 °C with 5.8 wt.% NaHCO3.Second, precipitation of CaCO3 by mixing the RO concentrate with the CIEX regenerant in a ratio of 2.8:1 andadjusting the pH to 11 was investigated. A 0.25 M NaCl solution remained after the treatment, which upon con-centration could be used to regenerate the resin.Third, application of EFC to a synthetic CIEX regenerant was also investigated. The multicomponent solutioncontained NaCl, MgCl2 and CaCl2. Upon EFC treatment, ice formed first and then NaCl · 2H2O at −29 °C. Thecomposition of eutectic point at−29.4 °C was 2.48 wt.% Na and 15.85 wt.% Cl.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Water scarcity is one of the main issues of this century. The applica-tion of membrane technology is playing an important role in theattempt to fight this problem. Nanofiltration (NF) and reverse osmosis(RO) membranes allow the use of alternative sources of water to pro-duce high quality water. However, its application entails the productionof a concentrate stream for which a destination has to be found. Inprinciple, for low volume, the waste stream could be transported tothe sea.However, for large capacity plants this is still a serious challenge.Alternatively, the remaining concentrate streammight be dewatered. In

nstitute, Groningenhaven 7, P.O.31 306069581.vador Cob).

literature evaporation is mentioned as an option to deal with the brineproblem. In arid areas evaporation ponds are used [1,2], but in humidclimates evaporation has to be realized by putting thermal energy intothe process. Given the fact that evaporative crystallization (EC) is ener-gy intensive and thus costly, it is worthwhile to explore the possibilitiesof alternative technologies. Eutectic freeze crystallization (EFC) is an al-ternative technology that is capable of separating aqueous solutionsinto pure water and pure solidified solutes. By cooling down the brineto the eutectic point (EP), ice and salt crystallize and these crystals canbe easily separated due to their density difference. The energy requiredto separate the water as ice is significantly less than that required toseparate it by evaporation, indicated by the fact that the heat of fusionof ice (6.01 kJ/mol) is less than the heat of evaporation of water(40.65 kJ/mol) [3]. In a previous study [4] it was shown that the energycost to treat an industrial KNO3–HNO3–H2O process stream with EFCwas 69% lower compared to EC. Furthermore, Fernández-Torres et al.

37S. Salvador Cob et al. / Desalination 344 (2014) 36–47

[5] performed a comparative life cycle assessment of EFC and EC. Thetwo techniques were compared to a 4 wt.% Na2SO4 solution, showingthat EFC was strongly preferred than EC, due to the 6 to 7 times lessenergy required.

EFC can be used as the final step after any high recovery processwhere a mixture of salt and water is produced. For example, Randallet al. [6] applied EFC on brine (0.6 vol.% of feed water volume), whichremained after extraction of gypsum from a HiPRO (High Recovery Pre-cipitating RO) concentrate (1 vol.% of the feed water). The brinecontained, among other ions, 7 g/L Na+, 1 g/L Ca2+ and 16 g/L SO4

2−.Their study shows that pure ice (N95% purity), pure calcium sulfateand pure sodium sulfate could be recovered from the brine. Ultimately,in the combined HiPRO, gypsum formation and EFC process, 99.6% offeed water was recovered as water, 0.3% as salts, and only 0.1% endedup aswaste. Thus, the total salt andwater recoverywas nearly 99.9% [6].

The nearly zero liquid discharge (ZLD) concept has recently been in-vestigated by various groups, and was the focus of several reviews[7–10]. Usually, membranes are used in the production of water innear-ZLD studies. Cation exchange (CIEX) resin may also be employed,to remove bivalent cations to prevent scaling. In a previous study feedwater was first treated by CIEX, then it was treated by NF, and theconcentrate was further treated by RO. The used pilot plant ran attotal system recoveries between 91 and 98% [11]. In the case of workingat 98% total recovery there is still a remaining 2% concentrate stream.Since EFC seemed a promising technique and requires less energythan for instance a three-stage evaporation system [4], it was chosento treat the RO concentrate. This RO concentrate stream was relativelyrich in HCO3

− (0.9 wt.%) and Na+ (0.3 wt.%), thus, the production ofNaHCO3 under eutectic conditions was expected.

Although the application of ionic exchange reduces scaling problemsand makes a reduction of the brine stream possible, regeneration intro-duces a newwaste material, the spent regenerant. In principle, the totalsalt waste stream increases because salts are needed to regenerate theionic exchange column. This solution is normally disposed directly tothe environment. Since the disposal of this solution to the environmentis undesirable, the challenge is, therefore, to reuse the CIEX spentregenerant. In this case study, the resin is in the sodium form andNaCl is used to regenerate it. The resulting regenerant can be treatedwith Na2CO3 to precipitate the calcium and to be able to re-use theregenerant. The possibility of treating the regenerant with the saltproduced from the RO concentrate is investigated.

The CIEX spent regenerant is amulti-component solution containinga mixture of NaCl, CaCl2 and MgCl2, among other ions. Another attrac-tive alternative is to treat this stream with EFC in order to producepure ice and recover pure salt.We know from literature that the solubil-ity lines and eutectic points of the binary systems are: NaCl–H2O:EP −21.2 °C, 23.3 wt.%, MgCl2–H2O: EP −33.6 °C, 32.3 wt.% andCaCl2–H2O: EP −55 °C, 29.8 wt.%. However, actual eutectic condi-tionsmight deviate from the ones for pure systems for themulticompo-nent solution due to the presence of impurities (dissolved ions). Thisoption has also been studied in this research.

Fig. 1. Scheme of the ZLD proce

2. Materials and methods

2.1. Scheme of the process

TheRO concentrate and CIEX spent regenerantwere obtained fromapilot plant described elsewhere [11]. When working at a 98% totalsystem recovery, the pilot plant treats 120 L of tap water and produces2.4 L/h RO concentrate and 100 L/week of regenerant, according toFig. 1.

Three different strategies to increase the total system recovery havebeen studied:

Strategy 1 Application of EFC to the RO concentrate. Pure ice and puresalt can be recovered. The salt produced can be used totreat the CIEX spent regenerant, or potentially be sold.

Strategy 2 Mixing of theROconcentratewith the CIEX spent regenerantto precipitate CaCO3 and re-use the regenerant for regenera-tion of the CIEX resin.

Strategy 3 Application of EFC to the CIEX spent regenerant. Pure ice andpure salt can be recovered.

2.2. Experimental methodology for Strategy 1

The starting solution for performing the experiments was the ROconcentrate from the pilot plant described in detail elsewhere [11].This RO concentrate was however, still rather diluted and far fromthe first expected eutectic point of NaHCO3 [12]. The RO concentratewas therefore, approximately 6 times concentrated by evaporationin order to get close to the expected eutectic point (it was pre-concentrated by evaporation just for practical reasons during the exper-iment, but it could had been pre-concentrated by freeze concentration).The concentration of the RO concentrate (pH 8.6) is shown in Table 1.

During the pre-concentration step, some salt formation already oc-curred. The subsequent EFC experiment was therefore conducted afterfiltering the salt from the solution with a 4.2 μm pore size filter(Whatman 1442 090). Fig. 2 gives a detailed scheme of the procedure.

The filtered pre-concentrated solution was placed in a 1 L plasticbeaker positioned inside a 2 L glass jacketed crystallizer (Fig. 3). AnIKA stirrer with variable speed control was used to provide adequatemixing of the solution inside the crystallizer. Cooling was achievedindirectly with a Lauda RE207 thermostatic unit. Two PT-100 sensorsconnected to an ASL F250 precision switchbox measured the tempera-ture of the solution and the coolant with a resolution of ±0.001 °C.The values were recorded in a computer via the Lab View program.The temperature of the coolant was set at−10 °C.

2.2.1. Sample analysesAfter ice formation, mother liquid samples were collected between

each 0.2 and 0.6 °C decrease for analysis. The samples were filteredwith a 0.22 μm Millipore filter and analyzed with Ion Couple PlasmaAtomic Emission Spectrometry (ICP-AES) with an accuracy of ±5%

ss studied in this research.

Fig. 3. Scheme of the crystallizer used during the EFC process.

Table 2Spent regenerant composition.

mg/L

Table 1RO concentrate average quality.

mg/L

Ca2+ 2.67Na+ 3300Mg2+ 0.25K+ 2.69HCO3

− 8900CO3

2− 170Cl− 205SO4

2− 21PO4

3− 4Cu2+ 0.47Fe3+ 0.052SiO2 128DOC 84

38 S. Salvador Cob et al. / Desalination 344 (2014) 36–47

and titration to determine the amount of HCO3− and CO3

2− using aTitrator TIM 865 (radiometer analytical) with an accuracy of ±2.5%.To calculate the amount of NaHCO3 and Na2CO3 it was assumed thatall the HCO3

− was associated to Na+ and the remaining Na+ linked tothe corresponding amount of the CO3

2−. Ice samples were filtered after2.5 h of operation and washed 6 times with pre-cooled pure waterand analyzed with ICP-AES. Salt samples were collected via filtrationand were characterized using a JEOL-6480LV scanning electron micro-scope (SEM) (JEOL Company) equipped with a Noran system SIXX-ray microanalysis (EDX) system (Thermo Electron Corporation) todetermine the structure and the composition. The samples were coatedwith a thin (10 nm) Au layer. An accelerating voltage of 6 kV was usedfor SEM observation and 10 kV for the EDX analysis. To try to determinethe type of salt, X-ray diffraction (XRD) analyses were performed withthe sample kept at room temperature (Bruker D8 Advance) andwith the sample collected during EFC experiment kept in a freezer(PANalytical X'Pert PRO X-ray diffractometer).

2.3. Experimental methodology for Strategy 2

The cation exchange resin is regenerated every week with a 100 L15wt.% NaCl solution. The spent regenerant was collected and the com-position was determined with ICP-AES (Table 2). 1 L of this regenerantwas poured into a 4 L glass beaker and the required amount of the ROconcentrate was added. For the composition of the RO concentrate,see Table 1. The pHwas adjusted to 11± 0.2 with a 5MNaOH solution.The beaker was stirred at 200 rpm in an incubator at 20 ºC. The Ca2+

concentration was measured every 5 min with a Ca2+ selectiveelectrode and when a stable concentration was reached, after 30 min,the experiment was stopped. Once the precipitate settled down, thesolution was filtered with a 0.45 μm filter and the precipitate was

Fig. 2. Experimental procedure scheme for treatment of RO concentrate with EFC.

dried in an oven and weighed. The precipitate was characterized withXRD and SEM–EDX as described in the previous section.

2.4. Experimental methodology for Strategy 3

A concentrated synthetic solution mimicking the typical CIEX spentregenerant was prepared (Table 3) (a synthetic solution was preparedfor practical reasons, avoiding the pre-concentration of the spentregenerant).

The average composition was 6.1 wt.% NaCl, 3.3 wt.% MgCl2 and14.2 wt.% CaCl2. The solution was prepared using 99.8 wt.% NaCl(Riedel-de Haën), 99 wt.% MgCl · 6H2O (Fluka), 97 wt.% CaCl2 (VWR)and demineralized water. The experiments were performed in a com-puter controlled EFC unit. The crystallizer used was a 12 L, batch type,Scraped Cooled Wall Crystallizer designed for EFC processes. Coolingwas achieved by circulating a Kyro 85 cooling liquid through a thermo-static unit, Lauda RUK 90 SW. The temperature of the cooling liquidwas

Ca2+ 15,300Na+ 10,800Mg2+ 300K+ 60Sr2+ 20Cl− 131,600SO4

2− 220SiO2 0.01

Table 3Synthetic spent regenerant composition.

mg/L

Ca2+ 51,200Na+ 24,000Mg2+ 7700K+ 47Sr2+ 22Cl− 149,000SO4

2− 31

Table 4Pre-concentrated solution composition.

mg/L

Ca2+ 0Na+ 19,900Mg2+ 1.46K+ 24HCO3

− 36,600CO3

2− 10,170Cl− 990SO4

2− 106PO4

3− 29Cu2+ 3Fe3+ 0.086SiO2 25

Fig. 4. Scheme of the EFC experimental set-up.

39S. Salvador Cob et al. / Desalination 344 (2014) 36–47

controlled with an accuracy of ±0.1–0.5 °C. Temperature of thesolution was measured using an ASL F250 precision thermometer con-nected to a PT-100 temperature sensor with an accuracy of ±0.01 °C.

Fig. 5. XRD pattern of the salt form

Throughout the crystallizer 5 PT-100 temperature sensors were locatedwhichmeasured the in- and outlet temperature of the coolant, the tem-perature at the bottomof the crystallizer (twice) and the temperature atthe top of the crystallizer. The data of the temperature sensors werecollected every 10 s and were recorded with Lab View program. Thesecomponents are connected as shown in Fig. 4.

11 L of solution at room temperaturewere poured into the crystalliz-er. The rotation speed of the scraper was set at 70 rpm. The temperatureof the coolant was lowered to a 10 °C/h cooling rate. After reaching theEP, the temperature difference between the cooling liquid and the slurrywas kept between 2 °C and 4 °C. After detection of the first temperaturejump in the aqueous solution, which was the indication of crystal(either ice or salt) nucleation, samples of the nucleated crystals andmother liquor were collected every 0.5 °C. Ice samples were washed 3times with pre-cooled pure water. Salt samples were washed 3 timeswith a saturated NaCl aqueous solution at −2 °C. Washing steps wereperformed at−0.2 °C in a cold room.

2.4.1. Sample analysesAll the samples were analyzedwith Inductive Couple Plasma Optical

Emission Spectrometry (ICP-OES)with an accuracy of 2.5%. The salt wasalso analyzedwith XRD. Pictures of ice and salt were taken under a LeicaWILD M10 stereomicroscope attached with a Nikon Coolpix 4500camera.

3. Results and discussion

3.1. RO concentrate treatment with EFC

RO concentrate from high recovery (98%) operations still con-tains mainly water. RO concentrate from the studied ZLD processwas pre-concentrated (Table 4) (pH 9.5) and salt deposition in theconcentrated solution was observed. Using XRD (Fig. 5), a salt sam-ple (Fig. 6) collected at room temperature via filtration was identifiedas a mixture of aragonite (CaCO3), trisodium hydrogendicarbonatedihydrate (Na3(CO3)(HCO3) · 2H2O) also known as trona, and a majorpart as amorphous or microcrystalline, which was probably silica.

The XRD pattern (Fig. 5) showed only a few sharp crystalline peaks,themajor parts are so-called “amorphous bumps”meaning the samples

ed during pre-concentration.

Table 6Mother liquid composition at the EP.

mg/L

Ca2+ 0.0Na+ 27,600Mg2+ 2.8K+ 515HCO3

− 42,300CO3

2− 17,800Cl− 1840SO4

2− 174PO4

3− 52Cu2+ 4Fe3+ 0.0SiO2 19

Fig. 6. SEM image of the salt produced during pre-concentration.

40 S. Salvador Cob et al. / Desalination 344 (2014) 36–47

are amorphous or microcrystalline. The crystalline peaks were identi-fied as aragonite and trona [13].

Fig. 6 shows a more or less amorphous layer with needle likespherulitic crystals present on top of it. The EDX analysis revealed thedistribution of elements in the salt indicated in Table 5.

The elements present in the precipitate are in agreement with theXRD analysis. The presence of silicon shows that the amorphous partof the precipitate is thus silica. It can also be seen that in the pre-concentrated solution the amount of silicon has decreased comparedto the RO concentrate, confirming that silicon (probably in the form ofsilica) is part of the precipitate.

The eutectic freeze crystallization treatment was applied to the fil-tered pre-concentrated solution (Table 3) in order to separate waterand salt from the solution. Fig. 7 shows the temperature profile of thesolution and the coolant in time during the experiment. After 1.4 h ofcooling, ice nucleation was indicated by the sudden temperature jumpin the solution from −8.3 to−3.1 °C (Fig. 7). At this point the concen-tration of NaHCO3was 5.5wt.% andNa2CO3was 1.5 wt.%. After 4 h of iceproduction, there was again a temperature jump from−6.7 to−3.9 °C,

Table 5EDX analysis of the salt produced during pre-concentration.

Element Atom (%)

C 11.32O 69.44Na 9.83Si 8.86Cl 0.39Ca 0.16

Fig. 7. Temperature profile of the experiment and followed pathway/sampling points.

indicating salt formation. Therefore,−3.9 °C was detected to be the eu-tectic point (Point E in Fig. 7) with 5.8 wt.% of NaHCO3, 2.7 wt.% Na2CO3

and 9 wt.% total dissolved ion content (Table 6).At this moment (Point E in Fig. 7) an optical microscope picture was

taken of the produced salt (Fig. 8). The salt crystals formedwere needle-like agglomerates, very small in size (it can be better distinguished inFig. 11). Under eutectic conditions upon further cooling, the EP wasdecreased till−4.1 °C due to the presence of impurities favoring the for-mation of NaHCO3 with 5.9 wt.% of NaHCO3 and 3.1 wt.% Na2CO3 and9.6 wt.% total impurity content in the mother liquid (Point F in Fig. 7).

After ice formation (Point C in Fig. 7), upon cooling and further iceproduction, the concentration of impurities in the mother liquid in-creased in time (Fig. 9). At −4.66 °C, the total concentration of Na+

and HCO3− was 8.9 wt.%. The next major impurity was CO3

2− with1.5 wt.%, followed by Cl− with 0.2 wt.%. Some other impurities werealso present in minor amount, SO4

2−, K+, PO43−, Si, Cu2+, Mg2+ and

Fe3+, with a total 0.03 wt.%.XRD analysis (XRD pattern is shown in Fig. 10) of the salt collected at

the end of the experiment (Point F in Fig. 7) and kept in the freezer toprevent recrystallization into Thermonatrite (Na2CO3 · H2O) revealedthat the salt formed under eutectic conditions was pure Nahcolite(NaHCO3) [13].

The NaHCO3 crystals were also characterized with SEM–EDX(Fig. 11). They were needle-like small agglomerate crystals with asize between 2 and 5 μm. This shape and size are in agreement withliterature [12].

The EDX analysis (Table 7) showed that the composition of the saltwas in the range of Na2CO3 with extra oxygen. However, according tothe phase diagram (Fig. 12), XRD analysis, and SEM pictures, the saltformed is Nahcolite (NaHCO3).

The ternary phase diagram of the pure Na2CO3–NaHCO3 and H2Osystem was previously determined [12]. We can use this definedphase diagram and sketch the path taken in this experiment (Fig. 12).

Fig. 8. Salt picture at the eutectic point (Point E) under the optical microscope.

Fig. 11. SEM image of the obtained NaHCO3 crystals (Point F).

Table 7EDX analysis of the salt formed under eutecticconditions.

Element Atom (%)

C 12.5O 64.8Na 22.3Fe 0.3

Fig. 9. a. Upon cooling impurity distribution in the mother liquid after ice formation andbefore reaching the eutectic point (samples taken from Points C to E in Fig. 7). b. Uponcooling impurity distribution in the mother liquid after ice formation and before reachingthe eutectic point (samples taken from Points C to E in Fig. 7).

41S. Salvador Cob et al. / Desalination 344 (2014) 36–47

Point A represents the RO concentrate. Point B is the solution after evap-oration (pre-concentration step). C, D, E and F are points from the EFCprocess, which were already shown in Fig. 7. Point C is the momentwhere ice formation took place; Point D is the last point before the EP;Point E is the Eutectic point; Point F is the final EP of the experiment.From Points C to D the solution was concentrated via ice formation,into a zone where NaHCO3 was formed. The solution remained in

Fig. 10. XRD pattern of the salt formed under eutectic cond

metastable state, until salt nucleation took place at Point D. The onsetof salt crystallization, which is shown by the jump from Points D to E,allowed us to define the metastable zone width, the maximum super-saturation for NaHCO3 salt for this impure system. Thus, as seen inFigs. 7 and 12, the eutectic condition was reached at Point E. Since weworked with an impure system (mainly CO3

2− and excess Na+), it wasexpected that the eutectic point would slightly shift to a lower temper-ature and a lower concentration of NaHCO3. The measured EP for thepure Na+–HCO3

−–H2O system is at −2.23 °C and 6.2 wt.% NaHCO3

[12]. In our case, the EP occurred at a lower T, −3.9 °C, and at a lowerconcentration 5.8 wt.% NaHCO3 due to the presence of other ions inthe system. The lower T is logical as the additional ions depress thefreezing point of the solution. The fact that a lower concentration of

itions collected at the end of the experiment (Point F).

Fig. 12. Ternary phase diagram of NaHCO3, Na2CO3 and H2O system, and followedpathway through the phase diagram for the experiment (A–B–C–D–E–F).

Fig. 13. Optical image of ice crystals.

Table 8Distribution coefficients of impurities in the ice.

Element K (I)

Cu2+ 0.0160K+ 0.0715Na+ 0.0078SO4

2− 0.0076PO4

3− 0.0018Si 0.1737

42 S. Salvador Cob et al. / Desalination 344 (2014) 36–47

NaHCO3 is found for the eutectic conditions in our experiments as com-pared to the pure Na+–HCO3

−–H2O system is not surprising, as there is alarge excess of Na+ in our system (the total concentration of Na+(aq)

and HCO3−(aq) at the EP is 7.0 wt.%).

Under eutectic conditions, the collected ice crystals have a roundshape with a 50 to 100 μm average diameter (Fig. 13). The ice crystalswere washed with pure water at 0 °C to remove the adhering mother

Fig. 14. Concentration of compounds in the ice as a function of washings. (0 refers to unwa

liquid and the purity of the ice was measured by ICP-AES and titration.In Fig. 14 one can see that after 3 washing steps most of the impuritieswere removed from the ice. The main impurity left was sodium and itremained constant by increasing the number of washing steps. Afterthe 6thwash, therewere also someminor amounts of silicon and potas-sium remaining. The desired purity of the ice determines the number ofwashing steps that are required. Respect to ions as impurities, with 3washing steps more than 99.9% of ice purity was obtained.

We can calculate the distribution coefficient of the impurities in theice to estimate the partition of each impurity between the ice and thesolution. It is calculated as the ratio of the concentrations of impurity(I) in the ice product crystals (C) and the solution (S):

K Ið Þ ¼ IC½ �IS½ � ð1Þ

The distribution coefficient of each impurity has been determinedafter the 6th washing step (Table 8). The distribution coefficients ofeach impurity were quite low under eutectic operation. The highest dis-tribution coefficient was Si with 0.1737. However, it is difficult to knowif the impurities were part of the ice or if theywere part of the adheringmother liquor, and maybe even part of the washing water.

As it has been mentioned before, the RO concentrate was extractedfrom a pilot plant [11]. This pilot plant consists, basically, of a CIEXresin column for pre-treatment, followed by NF and RO membranes.When the pilot is working at 98% total recovery, 2% concentrate streamis remaining.With the incorporation of EFC, the concentrate stream canbe minimized, increasing the total system recovery.

It can be assumed that 80% of the NaHCO3 present in the RO concen-trate can be crystallized via EFC. First, ice has to crystallize out, to reachthe eutectic point where NaHCO3 is formed. After that, we assume thatice and salt are produced in the same amount as they are present ateutectic conditions. In Fig. 15, the process diagram and flow streamsare shown.

EFC could potentially recover 95% of the RO concentrate as purewater and, theoretically, 80% of the NaHCO3 present in this RO concen-trate. The total system recoverywould be increased from 98.0% to about99.7%, thus nearly ZLD.

shed ice; washing water contained 0.152 mg/L K, 0.089 mg/L Na and 0.002 mg/L PO4).

Fig. 15. Process scheme including EFC on the RO concentrate.

Table 9Composition of the measured (via ICP-AES) mixture after filtration and the calculatedcomposition only mixing (no reaction).

Only mixing(mg/L)

Mixture(mg/L)

Ca2+ 4030 0.0Na+ 5270 5730Mg2+ 81 1.91K+ 18 20Sr2+ b0.01 0.03Cl− 34,800 25,200SO4

2− 74 52SiO2 94 72

43S. Salvador Cob et al. / Desalination 344 (2014) 36–47

3.2. Mixing RO concentrate and CIEX spent regenerant

The use of CIEX reduces the total recovery of the system by about0.5–2%, depending mostly on the concentration of multivalent cationsin the feed water. In order to reach the ZLD concept, the CIEX spentregenerant needs to be treated as well (besides the RO concentrate).One option to increase the total recovery in our system is mixing theRO concentrate with the CIEX regenerant. The RO concentrate containsmainly Na+ and HCO3

− in water (Table 1), and the regenerant containsmainly Na+, Ca2+ and Cl− (Table 2) in water. From the previous work[14] the most efficient dose of NaHCO3 and optimal conditions to pre-cipitate CaCO3 from a regenerant are known. We confirmed that inthis case 2.8 L of the concentrate were needed to precipitate the Ca2+

present in 1 L of the regenerant (1.5 wt.% Ca2+) and produce theoreti-cally 38 g of CaCO3. The required mixing ratio is close to the actualratio produced by the pilot plant, with 0.5% CIEX regenerant and 2%

Fig. 16. XRD pattern of t

RO concentrate. The saturation index of this mixture was calculatedfor CaCO3 at pH 11 using the software Phreeqc-2 [15] and it was 3.7for calcite and 3.5 for aragonite.

The two streams were mixed and the pH was adjusted to 11. After30 min the Ca2+ electrode measured a stable value of 0.0002 g/L. Thecomposition of the mixture after reaction and filtration was also mea-sured with ICP-AES (Table 9). Both measurements show that nearly allthe Ca2+(aq) disappeared from the solution, most likely formingCaCO3(s). XRD analysis (Fig. 16) identified the precipitate as pure calcite(CaCO3) as well [13]. Table 9 also gives the calculated composition ofthe mixture, based on the 1:2.8 mixing ratio and the composition ofthe RO concentrate (Table 1) and regenerant (Table 2). It can be con-cluded by comparing the measured and calculated composition (basedonly on mixing, with no reaction taking place), that Mg2+(aq) alsovanished from the solution. Mg2+(aq) could have either precipitated asMg(OH)2 and/or MgCO3 as both were supersaturated in the mixtureat pH 11 (as calculated from the respective solubility products givenin [16]), or co-precipitated with CaCO3. The difference in the predictedand measured Cl− concentration is probably due to analytical error.

From the SEM pictures (Fig. 17) we can observe the spherulitic par-ticles with a faceted surface of sizes between 3 and 8 μm. Similar pic-tures of calcite can be found in literature [17]. Calcite is the moststable polymorph of calcium carbonate. First, most probably amorphouscalcium carbonate (ACC) nanoparticles formed vaterite and thenvaterite transformed to calcite via a dissolution and recrystallizationmechanism [18]. This is the common mechanism in solutions with abasic pH [19], as it is here.

The EDXanalysis (Table 10) shows that the elemental composition isin accordance with the CaCO3 atoms, but with extra oxygen and thepresence of some impurities, as sodium andmagnesium. Quite possibly,

he precipitated salt.

Table 10EDX analysis of the precipitated salt.

Element Atom %

C 12.87O 69.61Ca 15.75Na 0.72Mg 0.65

Fig. 18. Temperature profile during EFC operation.

44 S. Salvador Cob et al. / Desalination 344 (2014) 36–47

water is caught inside the quickly crystallizing particles at the locallyvery high supersaturation after the addition of 5 M NaOH. The remain-ing mixture contained 0.25 M NaCl; it would be preferable to concen-trate this solution in order to achieve a higher concentration which ismore efficient for regeneration of the CIEX. Electrodialysis would be apotential technique to concentrate this solution [20].

Compared to the EFC treatment of RO concentrate, the treatment op-tion of mixing regenerant and RO concentrate does not recover purewater, and the precipitate is of relatively low quality. Thus, from thepoint of view of the formed products, treatment by EFC is preferable.However, based on energy consumption, this option is cheaper thanEFC.

Fig. 19. Mother liquor composition at different temperatures after reaching eutecticconditions for the first salt.

3.3. Application of EFC to the CIEX spent regenerant

Besides mixing of RO and CIEX regenerant, EFC could be applied totreat the CIEX spent regenerant to recover both water and salt. EFCtreatment was applied to a synthetic regenerant solution (Table 3).Fig. 18 shows the temperature profile in time during the experiment.

Three sensors measured the temperature values inside the crystal-lizer. As the crystallizer was perfectly mixed, the temperature readingswere lined with each other. In the solution temperature readings, twojumps were noticed. The first jump was the ice nucleation jump,which approximately occurred around−28.5 °C. Due to ice nucleation,a certain amount of energy (heat of crystallization) was released andgave a slight temperature increase indicated with the ice nucleationarrow. After the ice formation, upon cooling the solution further, theice crystals continued to grow. This increased the concentration of thesolution. At approximately −29.7 °C the second temperature jump oc-curred, indicating salt nucleation. After this jump the system reachedto EP of the first salt at −29 °C, which was expected to be NaCl. Com-pared to the literature value of the highest eutectic temperature,which is for theNaCl–H2O system at−21.1 °C [21], the eutectic temper-ature of the synthetic streamwas 8 °C lower. The reason of this behavioris again, like in the case of the EFC treatment of the concentrated ROconcentrate, due to the freezing point depression by the presence ofother ions in the solution. At the eutectic temperature of −29.4 °C,

Fig. 17. SEM picture of t

the composition was: Na 2.48 wt.%, Ca 5.49 wt.%, Mg 0.83 wt.%, and Cl15.85 wt.%.

When eutectic conditions for the first salt were achieved, thetemperature kept on decreasing due to the total composition changeof the mother liquor. As a result of ice and first salt crystallization, theslurry started to become saturated for the second and third saltnucleations. This composition change is presented in Fig. 19.

Ice samples were collected after the removal of themother liquid byfiltration (unwashed and after 3 timeswashed using pre-cooledwater).The uptake of impurities in the ice crystals is given for the syntheticsolution (Fig. 20).

After directly vacuum filtering the ice crystals at EP, it was seen thatthe ice crystals were highly contaminated by Cl, Ca, Na and Mg of themother liquid. Washing the ice crystals with pre-cooled water reducedthe Ca, Na and Mg contents to about 79, 57 and 11 mg/L, respectively;

he precipitated salt.

Fig. 20. Concentration of compounds in the ice as a function of washings (0 refers tounwashed ice; washing water contained 0.62 mg/L Ca).

Fig. 21. Concentration of compounds in the salt as a function of washings (0 refers tounwashed salt; washing liquid contained 16.8 mg/L K).

Table 12Distribution coefficients of impurities in the saltcrystals.

Element K (I)

Ca2+ 0K+ 0.95SO4

2− 0Sr2+ 0Mg2+ 0

45S. Salvador Cob et al. / Desalination 344 (2014) 36–47

whereas Cl was still the highest impurity with 574 mg/L. Further wash-ing steps might decrease Cl into lower values. Minor impurity of SO4

was decreased to 0.9 mg/L whereas K and Sr on the ice crystals werebelow the detection limit of the ICP-AES.

The distribution coefficient of each impurity has been determinedafter the 3rd washing step (Table 11).

The distribution coefficients of each impurity were quite low undereutectic operation. The highest distribution coefficient was SO4 with0.096. The distribution coefficients of the rest of the impurities were inminor level.

Salt crystalswere also collected after the removal ofmother liquid byfiltration and after three washing steps with NaCl saturated solution.The uptake of impurities in the salt crystals is given in Fig. 21.

Looking at the salt impurity content, it is seen that Ca was thehighest impurity (3734 mg/L), followed by Mg detected on the un-washed salt crystals. Due to three washing steps, Ca and Mg impuritieswere removed. The other minor impurity, K, increased slightly aftereach washing step due to its high content in the washing liquid.

The distribution coefficient of each impurity has been determinedafter the 3rd washing step (Table 12).

The distribution coefficients were zero for all the impurities exceptfor K with a high value of 0.95.

A photo of the salt crystals (Fig. 22) shows that most of the crystalshave the typical cubical NaCl · 2H2O shape, in agreementwith literature[22].

The salt crystals were analyzed with XRD (Fig. 23). The XRD patternindicated that the salt crystallized under the eutectic conditions andanalyzed at room temperature corresponded to halite (NaCl) [13].Since the XRD measurement was done at room temperature, the NaCl· 2H2O crystals that were formed at the EP in the EFC process wererecrystallized into NaCl.

The complexity of this system lies in the low temperatures that haveto be achieved. Therefore, in the presented work, the second and thirdeutectic points could not be reached using the starting solution. Thiswas because when the solution temperature reached lower than−35 °C, the crystallizer contained too much ice, the viscosity of theslurry became higher and the system had to face too much scalingthat the scrapers could no longer rotate. For future work, it would be

Table 11Distribution coefficients of impurities in the ice.

Element K (I)

Ca2+ 0.0026K+ 0.00Na+ 0.0055SO4

2− 0.096Sr2+ 0.00Mg+ 0.0024

highly recommended to use a crystallizer with a stronger scraping sys-tem to avoid ice scaling [23] and be able to reach the eutectic points ofthe second and third salts.

4. Conclusions

In this study three different strategies were studied to increasethe total recovery of a system composed of cation exchange resin,nanofiltration and reverse osmosis. Both the RO concentrate and CIEXspent regenerant are usually disposed of, but they contain valuablewater and salts.

• The RO concentrate stream was partially separated into salt andwater. EFC was applied to 6 times pre-concentrated and filtered ROconcentrate. The application of EFC led to the production of pure iceand pure nahcolite. The eutectic conditions were reached at −3.9 °Cand 5.8 wt.% NaHCO3 in the presence of 9 wt.% total dissolved ions.The eutectic line decreased by the production of ice and salt to−4.1 °C in the presence of 9.6 wt.% total dissolved ions content. Thetotal water recovery could potentially be increased by application ofEFC to the RO concentrate from 98.0 to 99.7%.

Fig. 22. Photo of the salt crystals under eutectic conditions.

Fig. 23. XRD pattern of the precipitated salt (10 samples measured).

46 S. Salvador Cob et al. / Desalination 344 (2014) 36–47

• Mixingof CIEX spent regenerant and RO concentrate, and addition of abase, causes precipitation of CaCO3 (N99% CaCO3 precipitated), butNaCl remains in the solution. Compared to the application of EFC,EFC leads to higher purity products (water/ice, and precipitate/salt).However, it is cheaper than EFC; theCaCO3 produced could potentiallybe sold and the remaining solution could be concentrated and reusedto regenerate the resin.

• Application of EFC to CIEX spent regenerant showed potentially that itis possible to treat NaCl–MgCl2–CaCl2 multicomponent syntheticstream using EFC. Upon cooling the stream, the first eutectic pointwas achieved at−29 °C for the ice–NaCl · 2H2O formation. At the eu-tectic temperature of −29.4 °C, the composition was: Na 2.48 wt.%,Ca 5.49 wt.%, Mg 0.83 wt.%, and Cl 15.85 wt.%. The ice and NaCl· 2H2O salt crystals were pure. However, the energy consumptionfor this system is high, due to the low temperatures that need to bereached. Furthermore, it is strongly recommended to operate thesystem with a crystallizer with a stronger scraping system to avoidice scaling.

These preliminary results show the application of EFC as a promisingtechnique to achieve near zero waste discharge in combinationwith ROmembranes by the production of reusable ice and salt from wastestreams. The application of Strategy 1 together with Strategy 3, wouldfurther increase the total system recovery to nearly zero liquiddischarge.

Acknowledgements

This work was performed in the cooperation framework of Wetsus,Centre of Excellence for Sustainable Water Technology (www.wetsus.nl). Wetsus is funded by the Dutch Ministry of Economic Affairs,Ministry of Infrastructure and the Environment, the European RegionalDevelopment Fund, the Province of Fryslân, and the NorthernNetherlands Provinces. The authors would like to thank the participantsof the research theme “CleanWater Technology” for the fruitful discus-sions and their financial support. We would like to thank Michel vanden Brink for the ICP analysis, Arie Zwijnenburg for the SEM–EDXanalysis and Ruud Hendrikx and Kees Goubitz for the XRD analysis.We would also like to thank Cyril Maffezzoni and Carlant Lopez for

their experimental contribution to this work. Finally, we also wouldlike to thank the reviewers for their comments, which helped toimprove the quality of this paper.

References

[1] M. Mickley, Review of concentrate management options, http://texaswater.tamu.edu/readings/desal/concentratedisposal.pdf 2004.

[2] M. Ahmed, et al., Use of evaporation ponds for brine disposal in desalination plants,Desalination 130 (2) (2000) 155–168.

[3] J. Nathoo, R. Jivanji, A.E. Lewis, Freezing your brines off: eutectic freeze crystalliza-tion for brine treatment, International mine water conference, 2009, (South Africa).

[4] F.v.d Ham, Eutectic freeze crystallization, Technical University of Delft, TheNetherlands, 1999. 122.

[5] M.J. Fernández-Torres, et al., A comparative life cycle assessment of eutectic freezecrystallisation and evaporative crystallisation for the treatment of saline wastewa-ter, Desalination 306 (2012) 17–23.

[6] D.G. Randall, J. Nathoo, A.E. Lewis, A case study for treating a reverse osmosis brineusing Eutectic Freeze Crystallization—approaching a zero waste process, Desalina-tion 266 (1–3) (2011) 256–262.

[7] A. Neilly, V. Jegatheesan, L. Shu, Evaluating the potential for zero discharge fromreverse osmosis desalination using integrated processes — a review, Desalin.Water Treat. 11 (1–3) (2009) 58–65.

[8] A. Pérez-González, et al., State of the art and review on the treatment technologiesof water reverse osmosis concentrates, Water Res. 46 (2) (2012) 267–283.

[9] J. Morillo, et al., Comparative study of brine management technologies for desalina-tion plants, Desalination 336 (2014) 32–49.

[10] A. Subramani, J.G. Jacangelo, Treatment technologies for reverse osmosisconcentrate volume minimization: a review, Sep. Purif. Technol. 122 (2014)472–489.

[11] S. Salvador Cob, et al., Silica and silicate precipitation as limiting factors in high-recovery reverse osmosis operations, J. Membr. Sci. 423–424 (15) (2012) 1–10.

[12] M. Rodriguez Pascual, et al., Determination of the eutectic solubility lines of theternary system NaHCO3–Na2CO3–H2O, Chem. Eng. Res. Des. 88 (10) (2010)1365–1371.

[13] R.T. Downs, The RRUFF Project: an integrated study of the chemistry, crystallogra-phy, Raman and infrared spectroscopy of minerals, 2006.

[14] W.G. Siegers, L.B., J. Post, M. Riemersma, Ionenwisseling als alternatief voorkorrelreactoren, regeneraat, Kiwa Water Research, DHV BTO 2004.066, TheNetherlands, 2004. 59.

[15] D.L. Parkhurst, C.A.J. Appelo, User's guide to PHREEQC (version 2) — a computerprogram for speciation, batch-reaction, one-dimensional transport, and inversegeochemical calculations, U.S. Geological Survey Water-Resources InvestigationsReport, 99-4259, 1999, (312 pp.).

[16] D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, Internet Version 2005, CRCPress, Boca Raton, FL, 2005.

[17] J.-P. Andreassen, Growth and aggregation phenomena in precipitation of calciumcarbonate, Norwegian University of Science and Technology, Norway, 2001. 121.

47S. Salvador Cob et al. / Desalination 344 (2014) 36–47

[18] J.D. Rodriguez-Blanco, S. Shaw, L.G. Benning, The kinetics and mechanisms of amor-phous calcium carbonate (ACC) crystallization to calcite, via vaterite, Nanoscale 3(1) (2011) 265–271.

[19] J.D. Rodriguez-Blanco, et al., The role of pH andMgon the stability and crystallizationof amorphous calcium carbonate, J. Alloys Compd. 536 (Suppl. 1) (2012) S477–S479.

[20] K. Ghyselbrecht, et al., Desalination of an industrial saline water with conventionaland bipolar membrane electrodialysis, Desalination 318 (2013) 9–18.

[21] D.A. Swenne, The eutectic crystallization of NaCl·2H2O and ice, 1983. (Eindhoven).[22] T.G. Zijlema, et al., The production of pure NaCl by the antisolvent crystallisation

of NaCl·2H2O and a consecutive recrystallisation step, J. Cryst. Growth 198–199(Part 1) (1999) 789–795 (0).

[23] M. Rodriguez Pascual, et al., A novel scraped cooled wall crystallizer. Recovery ofsodium carbonate and ice from an industrial aqueous solution by eutectic freezecrystallization, Chem. Eng. Res. Des. 88 (9) (2010) 1252–1258