Es

STR

h

•

•

•

a

ARRAA

KSCPPW

0h

Journal of Hazardous Materials 261 (2013) 29– 37

Contents lists available at ScienceDirect

Journal of Hazardous Materials

jou rn al hom epage: www.elsev ier .com/ locate / jhazmat

nhancement of struvite purity by re-dissolution of calcium ions inynthetic wastewaters

ang-Hun Lee, Byeoung-Hak Yoo, Sun-Kyoung Kim, Seung Joo Lim, Jun Young Kim,ak-Hyun Kim ∗

esearch Division for Industry & Environment, Korea Atomic Energy Research Institute, 1266 Sinjeong, Jeongeup, Jeollabuk-do, 580-185, Republic of Korea

i g h l i g h t s

In synthetic wastewaters, dynamic Caremovals were characterized alongstruvite formation.Ca2+ and PO4

3− ions were quicklyremoved from the wastewaters byprecipitation.The precipitated PO4

3− with Ca couldbe dissolved and used for struvite for-mation.

g r a p h i c a l a b s t r a c t

r t i c l e i n f o

rticle history:eceived 5 March 2013eceived in revised form 24 June 2013ccepted 28 June 2013vailable online 8 July 2013

eywords:truvitealciumhosphorus recoveryrecipitationastewater

a b s t r a c t

Although it is widely known that the presence of Ca ions inhibits the nucleation and growth of struvite,which consists of NH4

+, PO43−, and Mg2+, there is a lack of knowledge on actual Ca contents in struvite

co-precipitates at various N and P concentrations and the corresponding effects on the sizes of the pre-cipitates. Therefore, to address this challenge, this study designed synthetic wastewaters including thevariety of N and P concentrations, and conducted batch experimental reactions with each wastewaterto investigate Ca precipitation and size distributions of the precipitates. The molar ratio of Mg:P:N wasconfined to 1:1:7, while the initial Ca2+ concentrations were chosen to be 30–60 mg/L, which are typicalCa concentrations in real wastewaters. The result of the batch experiments confirmed that the presenceof Ca caused smaller solids than struvite as indicated in previous studies, and there was competitionbetween Ca-phosphate and Mg-N- PO4 (struvite) reactions, as expected. At the beginning of the experi-ment (∼1 min), fast Ca-phosphate precipitation was dominant because free Ca and P ions were quicklyremoved while Mg and N concentrations gradually reduced. However, as the nucleation and crystalgrowth processes elapsed, dissolved Mg and N concentrations continuously decreased, but dissolved Ca

concentrations could rise again at high N and P concentration conditions. The interesting phenomenonis that such increases of Ca concentrations probably results from the thermodynamic energy differencesbetween struvite and Ca-phosphate formations. A high thermodynamic driving force of struvite precip-itation could drive the re-dissolution of Ca-ions from the Ca-phosphate compounds with low saturationstates. This result is expected to be applied for increasing the struvite purity by the Ca re-dissolutionthrough the thermodynamic spontaneity without additional energy input.

∗ Corresponding author. Tel.: +82 63 570 3343; fax: +82 63 570 3348.E-mail addresses: tkhk@kaeri.re.kr, tkhkim@naver.com (T.-H. Kim).

304-3894/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jhazmat.2013.06.072

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Phosphorus (P) is one of the primary nutrients generatingeutrophication in aquatic systems [1–3]. To prevent eutrophication,municipal or agricultural wastewaters have been treated to reducethe P concentrations in the wastewater reaching surface water

3 ardou

spfpttafw

rmanP(wvAgtse3s[pccv

Cituppcrtvfvca

pgehs

TI

0 S.-H. Lee et al. / Journal of Haz

treams. While unregulated P is a pollutant in a water body, phos-horous is also a useful resource, such as in agricultural fertilizers,ood supply, and industrial raw materials [1–3]. Unfortunately,hosphorous resources have mostly been obtained from mineralshat should definitely be limited by the recent enormous utiliza-ion. Based on a previous study [4], phosphorous mineral resourcesre economically feasible for only 50 years. Therefore, P recoveryrom wastewater can be advantageous with respect to preventingater pollution and the devastation of mineral resources.

Struvite (MgNH4PO4·6H2O) precipitation, which was originallyegarded as an operational problem occurring in wastewater treat-ent plants [5,6], was recently considered as a useful P recovery

nd re-use as fertilizer. Generally, fertilizers should contain theutrition elements of N (nitrogen), P, and K (potassium). Although

can be precipitated with other ions instead of Mg, struviteMgNH4PO4) precipitation is preferred because it contains N asell as P. Also N is often rich in wastewater and therefore stru-

ite is desirable form for re-use as fertilizer and nutrient removal. successful P recovery should require an effective nucleation androwth of struvite crystals so that desirable amounts of precipi-ated struvite can be recovered typically through the gravitationalettling process. However, there are challenges to overcome; forxample, calcium ions (Ca), of which the typical concentrations are0–60 mg/L in municipal wastewater plants, are known as repre-entative ions that hamper struvite crystal nucleation and growth6]. Ca2+ ions actively react with phosphate to form calcium phos-hates. Previous studies on the influence of calcium on struviterystallization reported that Ca ions with struvite co-precipitationan retard the nucleation induction and inhibit the growth for stru-ite crystal formation [5,6].

The precipitation reactions of Ca ions with phosphate ions anda interference with the struvite formation have been investigated

n previous studies. For example, Le Corre et al. [6] investigatedhe impact of Ca solution concentrations on struvite crystallizationsing a range of analytical instruments including pH, absorbance,article size, and XRD (X-ray Diffractometer). It is known that theresence of Ca ions in a solution has a significant impact on struviterystallization in terms of the size, shape, and purity of the productecovered. In a previous study [6,7], it was found that increasinghe Ca concentration reduces the crystal size and inhibits the stru-ite growth, or affects the struvite crystallization and leads to theormation of an amorphous substance rather than crystalline stru-ite. At molar ratios of Mg:Ca > 1:1, it was reported that no morerystalline compound is formed, but a substance identified as anmorphous calcium phosphate was seen [6].

However, further studies are still required on calcium co-recipitation with phosphate and its inhibition to struvite crystal

rowth at various solution ionic concentration conditions. Forxample, some wastewaters such as livestock wastewaters, whichave often been studied for struvite recovery [8–10], have inten-ively concentrated on N, P, and Mg concentrations. Therefore, the

able 1nitial reaction conditions applied in this study.

Runs N + P

Low NP + low pH + no Ca Ntotal = 105 mg/L Ptotal = 32 mg/L

Low NP + low pH + High Ca

Low NP + high pH + no Ca

Low NP + high pH + High Ca

Mid NP + low pH + Low Ca Ntotal = 297 mg/L Ptotal = 102 mg/L

Mid NP + low pH + High Ca

Mid NP + high pH + Low Ca

Mid NP + high pH + High Ca

High NP + low pH + low Ca Ntotal = 1015 mg/L Ptotal = 309 mg/L

High NP + low pH + no Ca

Mid NP + high pH + low Ca

Mid NP + high pH + high Ca

s Materials 261 (2013) 29– 37

ultimate objective of this study is to explore desirable solution ionicconditions, especially incorporating Ca ions, to obtain the effectiveP-recovery as a struvite compound with high purity and propercrystal sizes. To achieve this, this study first needs to focus on inves-tigating the dynamic variation of Ca concentrations in the syntheticwastewater solutions in batch reactions, reflecting Ca contents inthe precipitates. In addition, the size distributions of the precipi-tates would be observed with the presence of Ca at various N andP concentrations.

2. Materials and methods

2.1. Design of reaction conditions

To undertake a series of batch experiments, a total of 12 syn-thetic wastewater solutions were prepared with various initialionic concentrations for each, including Mg2+, P, and N with aconfined molar ratio of 1:1:7, based on the N:P ratios of typicalwastewater sludge [11]. Phosphorus concentrations of 1, 3, and10 mM; nitrogen (N) concentrations of 7, 20, and 70 mM; and mag-nesium (Mg) concentrations of 1, 3, and 10 mM were chosen. Inaddition, Ca2+ ion concentrations of zero, 0.75, and 1.5 mM (0, 30, or60 mg/L), reflecting typical concentrations in water or wastewater,were added to the solution. The individual reaction conditions andthe corresponding initial ionic concentrations are shown in Table 1.

2.2. Preparation of wastewaters

First, stock solutions were prepared using the desired N andP concentrations by dissolving an analytical reagent grade ofNa2HPO4·6H2O (Sigma ACS Reagent, Fisher Scientific, USA) andNH4Cl (Sigma ACS Reagent, Fisher Scientific, USA) chemicals in dis-tilled water. The initial pH was then adjusted to 8.7 or 9.7 using 5 NNaOH or 1 N HCl solutions. MgCl2·6H2O (Sigma ACS Reagent, FisherScientific, USA) and CaCl2 (Sigma ACS Reagent, Fisher Scientific,USA) were added to the solutions in the batch reactors immediatelybefore the experiments were initiated. The agitation for the batchreactions were implemented using a jar tester (JHS-2/90, Republicof Korea) with a constant mixing speed of 120 rpm at 22 ◦C. The con-ditions were chosen based on previous studies [12,13]. The totalreaction duration for each batch experiment was 3 hrs, and thesampling time points were specified among 1, 5, 10, 20, 30, 60,and 180 min to observe the reaction kinetics. Approximately 8 mLsolutions were sampled and filtrated using 0.2 �m syringe filterscombined with 20 mL syringes (NORM-JECT, Germany) to removesolid precipitates. For a subsequent measurement of the ion con-centrations, 0.1 mL of 99% sulfuric acid was added to each sample to

prevent unwanted precipitation and then stored at 4 ◦C. Solid pre-cipitates were sampled by filtering the solution through 0.45 �mfilters at some sample time points, and dried at room temperature.To assess the struvite purity or to identify fates of the removed ions

pH Ca (mg/L) Mg (mg/L)

8.75 0 2229 21

9.72 0 2130 22

8.7 31 7161 72

9.71 30 6861 70

8.72 29 2520 253

9.72 30 25159 249

ardou

iHi

2

iHttOOtXtattosdpt

3

3

tCtirdtttNiip

giCmcadw

3

tttqsataN

S.-H. Lee et al. / Journal of Haz

n the precipitates, some solid samples were re-dissolved in a 1 NCl solution for subsequent elemental analysis of the re-dissolved

on concentrations.

.3. Analytical techniques

The concentrations of total P (Ptotal) and total N (Ntotal) dissolvedn the solutions were measured by a spectrophotometer (DR5000,ACH Inc., USA). The Nessler method was utilized to measure

he NH4+-N concentrations using the spectrophotometer. Free and

otal concentrations of Mg2+ and Ca2+ were measured using ICP-EP (Inductively Coupled Plasma-Optical Emission Spectroscopy:PTIMA 7300 DV, PerkinElmer, USA). The precipitates were charac-

erized using XRD (X-Ray Diffractometer, X’Pert PRO Multi Purpose-Ray Diffractometer, PANalytical, Holland). The particle size dis-

ributions of the precipitates were measured using a particle sizenalyzer (DelsaTM nano C, Beckman Coulter, Inc., USA), based onhe laser scattering principle, after each experimental reaction waserminated. The particles for the particle size measurement werebtained directly from the synthetic wastewaters to incorporatemall precipitates. The obtained solution samples were settledown for a single day to augment the concentrations of suspendedrecipitates and consequently increase the optimal detectability ofhe particle size analyzer.

. Results

.1. Ca concentrations

Fig. 1 shows the Ca ion concentrations as a function of reactionime for each reaction condition. The reaction conditions withouta addition are neglected in this plot. At high pH and low NP concen-ration conditions, the removed Ca concentration, compared withts initial concentration, was maintained at 70%–80% through theeactions. Conversely, at the low pH and low NP conditions, a lowissolved Ca removal with less than 40%, in comparison to the ini-ial concentration, was observed through the reactions. This implieshat the high precipitated Ca amount that temporally occurred athe high pH condition is relevant to the co-precipitation of Mg andH4

+ N bound to the phosphate; the Mg precipitation is compet-tive to the Ca precipitation at the beginning of the precipitationn this study. The low pH condition maintained a rather steady Carecipitation that slightly increased as time elapsed.

At the mid and high NP concentrations, the dissolved Careatly decreased at the beginning of the reactions and surpris-ngly increased afterward. In detail, the removed Ca ratios to initiala concentrations at 1 min and 3 h were 70–95% and 30–60% at theid NP concentrations, respectively. Similarly, those at the high NP

oncentrations were up to 52%-89% of the removal ratios at 1 minnd then declined to 20%. Based on a mass balance analysis by re-issolution of the precipitated samples, most of the removed Caas contained in the precipitated solids.

.2. Other ionic concentrations

Fig. 2 shows changes in the N and P concentrations as a func-ion of the reaction time duration in the batch reactors, in whichhe results of the low, middle, and high NP concentration condi-ions are presented in Fig. 2a–c, respectively. The P concentrationsuickly decreased at the beginning of each reaction, while ratherlow declines were observed for the N concentrations. Fig. 2a shows

strong pH dependence of the changes in the N and P concentra-ions at the low NP conditions, while such effects became weakert the other higher NP conditions. At the low NP conditions, the

and P concentrations greatly decreased with respect to time at

s Materials 261 (2013) 29– 37 31

the high pH, and the pH effect was also observed for N concentra-tion changes at the mid NP conditions. Overall, the presence of Caslightly inhibited N precipitation, but its effects on temporal N orP concentration changes were not strong compared with those ofthe initial NP or pH levels. At higher NP and lower initial pH levels,the pH declines were as much as 0.2–0.7 or greater. The presenceof Ca exhibits no strong effect on the pH change.

Fig. 3 shows the Mg ion concentrations according to the elapsedtime for each reactor condition. Low equilibrium removals of Mgfrom the solutions ranging in 0–75% of precipitation were observedat the low NP conditions, while higher equilibrium removals wereachieved at up to 74–87% and even greater than 90% at themid and high NP conditions, respectively, because of the highsuper-saturation state leading to active Mg precipitations. The Mgremoval rates were also higher at high super-saturation ratios. Only10 min were spent for high Mg removals of 90% at high NP condi-tions, regardless of other pH or Ca conditions. In contrast, morethan 10 min was spent to achieve the equilibriums at the mid NPand high Ca conditions and at the low NP conditions.

As for the effects of pH and Ca ions, the Mg precipitations werehighly influenced by the solution pH and the Ca concentrations atthe low NP concentrations. For example, the maximal removals of48% at the low Ca concentrations and 73% at zero Ca at the high pHconditions were clearly higher than less than 20% of Mg removals atthe low pH conditions. At the low NP conditions, fast Mg removalswere shown only at the high pH and low Ca concentrations, whilesteady removals were observed at the other conditions. Clearly dis-tinct Mg removals at the mid NP concentrations were observed asslower Mg removals for the high Ca conditions than otherwise. Atthe high NP conditions, fast and high Mg removals were achieveddue to the high super-saturation condition, regardless of the pH orCa concentrations.

3.3. Particle size distributions

Fig. 4 exhibits the size distributions of the precipitates (struviteand Ca phosphates) that were produced in the reactions. Each sizedistribution was presented by plotting the sizes corresponding to25% (d25), 50% (d50) and 75% (d75) in volume-based cumulativedistributions. In particular, this study focused on the effects of Caion concentrations on the particle size distributions. The horizontaland vertical axes represented the reaction conditions and precipi-tated particle sizes (d25, d50, or d75). The resultant data show thatsmaller d25 values were observed with higher Ca concentrationswhen the other conditions were identical; this phenomenon wasreported in the literature, showing that the number of small par-ticles increased by the presence of Ca ions [6]. The presence of Caresults in the presence of small crystals of sizes less than 2 �m. Athigher initial Ca concentrations of 60 mg/L, the data showed a risein the population of particles smaller than 1 �m than for the resultswithout Ca addition or 30 mg/L of initial Ca concentration.

However, overall size ranges of d50 are relatively invariant overthe reaction conditions except at low NP + high pH and at midNP + low pH levels. For the case of d75, there are contradictingresults in that larger sizes were observed with the lower Ca con-centration at mid NP + low pH conditions, and otherwise at lowNP + low pH conditions. Other d75 sizes are not much different witheach other with the presence or absence of Ca. The high pH levelsat low NP + low Ca concentrations produced smaller d50 and d75.Interestingly, mid NP + low Ca conditions show the largest sizesamong all conditions.

3.4. Crystal characterization

Fig. 5a–d exhibit XRD patterns of the precipitates at lowNP + high pH + low Ca condition, at low NP + high pH + no Ca

32 S.-H. Lee et al. / Journal of Hazardous Materials 261 (2013) 29– 37

Fig. 1. Variation of Ca ion concentrations in the synthetic wastewater solutions according to reaction time duration, for the reaction conditions of (a) low NP, (b) mid NP,and (c) high NP.

Fig. 2. Variation of N and P concentrations in the synthetic wastewater solutions as a function of the reaction time duration, for the reaction conditions of (a) low NP, (b) midNP, and (c) high NP.

S.-H. Lee et al. / Journal of Hazardous Materials 261 (2013) 29– 37 33

Fig. 3. Mg ion concentrations according to the reaction time elapsed, for the reaction conditions of (a) low NP, (b) mid NP, and (c) high NP.

y the

cNtNea

Fb

Fig. 4. Size distributions of the precipitates, represented b

ondition, at the beginning of the reaction (1 min) with midP + high pH condition, and at the end (3 hrs) of the reaction with

he same condition, respectively. This study chose the two mid

P + high pH conditions, because those produced the most notableffect resulting from the presence or absence of Ca ions and suit-ble amounts of precipitates could be obtained. The first condition

ig. 5. (a–d). XRD patterns of the precipitates at the reaction conditions of (a) low NP + higeginning time point (∼1 min), and (d) mid NP + high pH + high Ca with the end time poin

three sizes: d25, d50, and d75 for each reaction condition.

(Fig. 5a) had the highest Ca/Mg ratio (∼0.7) and resulted in no clearpeak location which probably reflects that XRD cannot detect anycrystal structures in the precipitates. This phenomenon agrees with

the results of the previous studies [6,7], in that the Ca compounds,co-precipitated with struvite, were identified mostly as the amor-phous structures. In contrast, the second one (Fig. 5b) showed the

h pH + low Ca, (b) low NP + high pH + no Ca, (c) mid NP + high pH + high Ca with thet (∼3 h).

3 ardous Materials 261 (2013) 29– 37

tatfetar

pdCiSartip(

4

tVsrsoi

Table 2Equilibrium solubility constants for the precipitates at 25 ◦C.

Name Chemical Formula pKsp

ACP1 (am1) Ca3(PO4)2 −25.5ACP2 (am2) Ca3(PO4)2 −28.25TCP (beta) Ca3(PO4)2 −28.92

Fb

4 S.-H. Lee et al. / Journal of Haz

ypical struvite pattern with its high XRD peaks [6,7,14]. In Fig. 5c–d,lthough the XRD patterns of the two cases were both identified ashat of struvite, the pattern at the beginning of the reaction showedurther reduced peak sizes than that at the end of the reaction. Inter-stingly, this result agrees with the batch experimental results ofhis study of lower (or higher) Ca concentrations (or precipitation)t the beginning of the reaction (Fig. 5c) than at the end of theeaction (Fig. 5d).

The SEM images in Fig. 6a exhibit the small precipitates by theresence of Ca2+ ions produced at low NP + high pH + low Ca con-ition. When compared with the image from the reaction withouta2+ ions (low NP + high pH + no Ca condition) in Fig. 6b, this result

s consistent with the particle size analysis data shown above. TheEM images in Fig. 6c–d show the precipitates at the beginningnd at the end of mid NP + high pH + high Ca reaction condition,espectively. The difference between the two images is not as dis-inct as Fig. 6a-6b, but, because of the higher Ca content, the SEMmage at the beginning (Fig. 6c) clearly exhibit small and irregularrecipitates in comparison with those at the end of the reactionFig. 6d).

. Discussion

To address the re-dissolution of the precipitated Ca ions usinghe thermodynamic equilibrium principles, this study utilizedisual MINTEQ software. Fig. 7 shows the saturation indices of theolid components that can be precipitated in the reactions. Three

eaction time points were chosen as 0, 1, and 180 min to repre-ent the saturation conditions before the reactions, at the beginningf the reactions, and at the equilibriums, respectively. The eas-ly supersaturated compounds detected by Visual MNTEQ include

ig. 6. (a–d). SEM images of the precipitates at the reaction conditions of (a) low NP + higeginning time point (∼1 min), and (d) mid NP + high pH + high Ca with the end time poin

OCP Ca4H(PO4)3:3H2O(s) −47.95Struvite (MAP) Mg(NH4)(PO4) −13.26

two kinds of ACP (Amorphous Calcium Phosphate: Ca3(HPO4)2:am1 and am2) and TCP (Tricalcium phosphate: beta) that containidentical chemical formula to those of ACP, OCP (Octacalcium phos-phate: Ca4H(PO4)3), and struvite (MAP). The equilibrium solubilityconstants of those compounds, used in type6 database in MINTEQ,are shown in Table 2; all the constants on the Ca compounds areloaded in MINTEQ, while that of struvite is directly inserted usingthe most recent previous experimental data [15]. These compoundslisted in Table 2 have actually been found in wastewater treatment[16]. The program also showed Hydroxyapatite (HAP) as a highlysupersaturated compound during the entire reaction duration, butthis study neglected the compound since the supersaturation wasalways maintained until the end of the reactions, and thus the HAPprecipitation might not be relevant with the Ca re-dissolution.

The saturation indices (SIs) of struvite was between 1.0–3.5before the reactions (0 min) for all reaction conditions, but reachequilibrium (SI range: −0.5 to 0.0) state at the end of the batch reac-tion time at 180 min for all conditions except low NP + low pH + low

Ca. This implies that the ions of PO4

3−, NH4+, and Mg2+, which

are incorporated in struvite formation, were consumed until thestruvite precipitation reaction reached its equilibrium. In contrast,although the SIs of the Ca phosphate compounds also decreased,

h pH + low Ca, (b) low NP + high pH + no Ca, (c) mid NP + high pH + high Ca with thet (∼3 h).

S.-H. Lee et al. / Journal of Hazardous Materials 261 (2013) 29– 37 35

Fig. 7. (a–i). Saturation indices of the solid components that are potentially precipitated at three reaction time points (0, 1, and 180 min) in the reactions of (a) low NP + lowp + lowN Ca.

ttla∼

irtwmipatcSmrninqimtr

s

H + low Ca, (b) low NP + high pH + low Ca, (c) mid NP + low pH + low Ca, (d) mid NPP + low pH + low Ca, (h) mid NP + high pH + low Ca, and (i) high NP + high pH + high

hese reactions were still super-saturated until the end of the reac-ions. For the condition of low NP + low pH + low Ca, the relativelyow ionic concentrations provide a slow reaction rate, and the SIt 180 min may still exhibit supersaturation states for struvite (SI0.24) and Ca phosphates.

Shortly after the reactions began (1 min), the quickly decreasedonic concentrations and the corresponding SIs reflect highlyeduced saturation states as compared with those before the reac-ion started. In particular, the reduction of SI was large for am1,hich often fell to an undersaturation. This implies that the ther-odynamic driving force of the struvite precipitation leading to

ts equilibrium is dominant over that of amorphous calcium phos-hate. When struvite SIs at the initial time are compared with thoset the final stage, large SI drops were observed for all the reac-ion conditions, except a low NP + low pH + low Ca condition. Inontrast, Ca-phosphate compounds had similar or even increasedIs over the time elapsed; this trend is noticeably apparent inid NP concentrations. Such increased SI is probably due to the

elease of Ca2+ by Mg2+ and NH4+ ions, with a strong thermody-

amic driving force of struvite precipitation, taking the phosphateons in the undersaturated Ca-phosphate compounds at the begin-ing. This phenomenon of the complicated SI of Ca-phosphates isuite noticeable at mid NP + high pH + low Ca conditions. Interest-

ngly, the condition shows undersaturated or equilibrated states forost of the Ca-phosphates at a 1 min reaction time. This supports

he possible influences of thermodynamic supersaturation on Cae-dissolution.

Fig. 8 shows schematic demonstrations to visualize the micro-copic processes of struvite and Ca-phosphate formations at mid or

pH + high Ca, (e) mid NP + high pH + low Ca, (f) mid NP + high pH + high Ca, (g) high

high NP conditions. The “Mg-N-P” in larger rectangles and “Ca-P”in smaller circles represent individual struvite and Ca-phosphateprecipitates, respectively. Also, the symbols of “Ca”, “P”, and “N”stand for dissolved ions of Ca2+, NH4

+, and PO43− that can partici-

pate in the precipitation. Fig. 8a describes the initial stage at whichCa-phosphate precipitation actively occurs through the rapid bind-ing between Ca2+ and PO4

3−. Consequently, the dissolved Ca2+ andPO4

3− ions were quickly removed from the solution, as comparedwith Mg2+ and NH4

+ ions. Fig. 8b–d show the subsequent stages ofthe initial reactions: first, Mg2+, NH4

+, and PO43− are bound to form

struvite (Fig. 8b), and thus SI of Ca-phosphates is further reducedbecause of the PO4

3− removed by the struvite formation. SomeCa-phosphates reaching undersaturated or equilibrium states canrelease Ca and PO4

3− ions into the solution (Fig. 8c) and the re-dissolved PO4

3− can be bound to Mg2+ and NH4+ ions to form

struvite (Fig. 8d).This study exhibits that the presence of Ca2+ led to a creation of

small particles that probably either contained small Ca-phosphatesor inhibited struvite growth. The variation in the size distributionsshould depend on the Ca concentrations and NP concentrations.As observed in Fig. 4, the optimal size distribution that reflects theexistence of larger particle sizes appears to be achieved at the midNP concentrations with low Ca concentrations. Interestingly, theparticle sizes under the two conditions are larger than those at highNP concentrations. A probable reason for this is that high NP ionic

concentrations produced small struvite crystals, while low concen-trations showed large crystal sizes [6]. This is perhaps because highsuper-saturation conditions tend to generate many nuclei seeds,rather than crystal growth. Alternatively, the presence of Ca2+ can

36 S.-H. Lee et al. / Journal of Hazardous Materials 261 (2013) 29– 37

Fig. 8. (a–d). Schematic demonstrations for the microscopic views of Ca-phosphate re-dissolution along struvite formations at mid or high NP conditions: the “Mg N P” withthe large rectangles and “Ca P” with the small circles represent struvite and Ca-phosphate precipitates, respectively. The symbols of “Ca”, “P”, and “N” represent dissolvedC d staga recip

iScssWtcttnsotCi

istwao(att

a2+, NH4+, and PO4

3− ions. (a) The initial stage with Ca precipitation, (b) the seconnd PO4 occur, and (d) the final stage that struvite continues to form by Mg and N p

mprove the agglomeration between struvite precipitate particles.truvite homogeneous surfaces are negatively charged [17,18], andonsequently, repulsive forces are exerted between homogeneousurfaces. Thus, a proper concentration of Ca2+ between the struviteurfaces can reduce the repulsion and enhanced agglomeration.

hen the Ca2+ concentrations were too low or too high, the nega-ive charges could not be reduced enough, or new positive repulsiveharges could be created, respectively. Probably, mid NP concen-rations with 30 mg/L of Ca2+ concentrations in this study may be athe proper concentrations. Unfortunately, the re-dissolution phe-omenon of the precipitated Ca does not significantly affect theizes of distributions, in that, based on the experimental resultsf this study, the presence of Ca ions leads to the production ofhe small precipitates at any NP concentrations. The role of thea re-dissolution in precipitate size distributions should be further

nvestigated in future studies.The Ca re-dissolution phenomenon has rarely been reported

n previous studies, but a recent article by Capdevielle et al. [19]howed Ca re-dissolution where Ca ions were mostly precipi-ated within 30 min along struvite precipitation in synthetic swineastewaters. The wastewaters contained many ionic components

s SO42−, NO3

−, K+, and carbonates. As a result, higher than 90%f P was recovered as large crystals of struvite at a low Mg:Ca

2.25:1) and a high N:P (3:1) ratios, stirring rates of 45 − 90 rpm)nd low temperature (below 20 ◦C). Although their experimen-al conditions were somewhat different, it would be interestinghat their experimental results as well as Ca re-dissolution well

e when struvite also begins to form, (c) the third stage when the dissociation of Caitated with the re-dissolved PO4.

agree to those of this study. For example, a high N:P ratio improvedstruvite precipitation over Ca PO4; ACP (amorphous calcium phos-phate) co-precipitated with struvite was the main component ofprecipitated Ca; small precipitated particles mostly consisted of Ca-PO4 (ACP). Such high correspondences between Capdevielle et al.’sresults and this study imply that occurrence of Ca re-dissolutionphenomenon is not limited to a special ionic conditions but canbe prevalent during struvite precipitation from the wastewaterscontaining high P concentrations with Ca presence.

5. Conclusion

This study including the Ca2+ re-dissolution at high N(>300 mg/L) and P (>100 mg/L) concentration conditions providesuseful information to maintain a high struvite purity in existenceof Ca ions. It is because of the thermodynamic spontaneity thatdoes not need additional energy consumption. Based on compar-ative batch experiments using three NP, two pH, and three Caconcentrations, it was observed that dissolved Ca ions led to a cre-ation of small precipitates. Also, based on the competition betweenCa-phosphate and struvite precipitation reactions, the phosphate

incorporated in early Ca-phosphate precipitation was later partic-ipated in struvite formation. With the particle size analysis, it wasalso observed that the maximized crystal growth for a desirablerecovery through the gravitation settling were achieved with mid

ardou

NC

A

tb

R

[

[

[

[

[

[

[

[

[

S.-H. Lee et al. / Journal of Haz

P concentrations (300 Ntotal mg/L and 100 Ptotal mg/L) and lowa2+ concentrations.

cknowledgement

This research was supported by the Nuclear R&D programhrough the National Research Foundation of Korea (NRF)fundedy the Ministry of Education, Science and Technology.

eferences

[1] M.I. Ali, P.A. Schneider, A fed-batch design approach of struvite system in con-trolled supersaturation, Chem. Eng. Sci. 61 (12) (2006) 3951–3961.

[2] M.I. Ali, P.A. Schneider, An approach of estimating struvite growth kineticincorporating thermodynamic and solution chemistry, kinetic and processdescription, Chem. Eng. Sci. 63 (2008) 3514–3525.

[3] J. Wang, J.G. Burken, X. Zhang, R. Surampalli, Engineered struvite precipitation:Impacts of component-ion molar ratios and pH, J. Environ. Eng. 131 (2005)1433–1440.

[4] J. Driver, D. Lijmbach, I. Steen, Why recover phosphorus for recycling and how?Environ. Technol. 20 (1999) 651–662.

[5] Y. Jaffer, T.A. Clark, P. Pearce, S.A. Parsons, Potential phosphorus recovery bystruvite formation, Water Res. 36 (2002) 1834–1842.

[6] K.S. Le Corre, E. Valsami-Jones, P. Hobbs, S.A. Parsons, Impact of calciumon struvite crystal size, shape and purity, J. Cryst. Growth 283 (2005)

514–522.

[7] Z-L. Ye, S-H. Chen, S-M. Wang, L-F. Lin, Y-J. Yan, Z-J. Zhang, J-S. Chen, Shortcommunication: phosphorus recovery from synthetic swine wastewater bychemical precipitation using response surface methodology, J. Hazard. Mater.176 (2010) 1083–1088.

[

s Materials 261 (2013) 29– 37 37

[8] N.O. Nelson, R.L. Mikkelsen, D.L. Hesterberg, Struvite precipitation in anaerobicswine lagoon liquid: effect of pH and Mg:P ratio and determination of rateconstant, Bioresour. Technol. 89 (2003) 229–236.

[9] Y-H. Song, G-L. Qiu, P. Yuan, X-Y. Cui, J-F. Peng, P. Zeng, L. Duan, L-C. Xiang,F. Qian, Nutrients removal and recovery from anaerobically digested swinewastewater by struvite crystallization without chemical additions, J. Hazard.Mater. 190 (2011) 140–149.

10] K. Suzuki, Y. Tanaka, K. Kuroda, D. Hanajima, Y. Fukumoto, T. Yasuda, M. Waki,Removal and recovery of phosphorous from swine wastewater by demon-stration crystallization reactor and struvite accumulation device, Bioresour.Technol. 98 (2007) 1573–1578.

11] Metcalf and Eddy, Inc., Wastewater Engineering: Treatment, Disposal andReuse, 3rd ed., McGraw Hill, USA, 1991.

12] M.I.H. Bhuiyan, D.S. Mavinic, R.D. Beckie, Nucleation and growth kinetics ofstruvite in a fluidized bed reactor, J. Cryst. Growth 310 (2008) 1187–1194.

13] Z. Liu, Q. Zhao, L. Wei, D. Wu, L. Ma, Effect of struvite seed crystal on MAPcrystallization, J Chem. Technol. Biotechnol. 86 (2011) 1394–1398.

14] N.C. Bouropoulos, P.G. Koutsoukos, Spontaneous precipitation of struvite fromaqueous solutions, J. Cryst. Growth 213 (2000) 381.

15] K.N. Ohlinger, T.M. Young, E.D. Schroeder, Predicting struvite formation indigestion, Water Res. 32 (1998) 3607–3614.

16] E. Valsami-Jones, Mineralogical controls on phosphorus recovery from wastew-aters, Mineral. Mag. 65 (2001) 611–620.

17] K.S. Le Corre, E. Valsami-Jones, P. Hobbs, B. Jefferson, S.A. Parsons, Agglomera-tion of struvite crystals, Water Res. 41 (2007) 419–425.

18] C. Zhang, Y. Chen, Simultaneous nitrogen and phosphorus recovery fromsludge-fermentation liquid mixture and application of the fermentation liq-uid to enhance municipal wastewater biological nutrient removal, Environ. Sci.

Technol. 43 (2009) 6164–6170.

19] A. Capdevielle, E. Sykorová, B. Biscans, F. Béline, M-L. Daumer, Opti-mization of struvite precipitation in synthetic biologically treated swinewastewater—determination of the optimal process parameters, J. HazardMater. 244–245 (2013) 357–369.