ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 9 6 5 – 2 9 7 1

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auCanada H1X 2B2. Te

E-mail address:

journal homepage: www.elsevier.com/locate/watres

Phosphorus removal from wastewater by mineral apatite

Nathalie Belliera, Florent Chazarenca,b,�, Yves Comeaua

aDepartment of Civil, Geological and Mining Engineering, Ecole Polytechnique, 2900 Edouard-Montpetit, Montreal, Que., Canada H3T 1J4bInstitut de recherche en biologie vegetale, Universite de Montreal, 4101 rue Sherbrooke Est, Montreal, Que., Canada H1X 2B2

a r t i c l e i n f o

Article history:

Received 4 April 2006

Received in revised form

12 May 2006

Accepted 18 May 2006

Available online 10 July 2006

Keywords:

Apatite

Hydroxyapatite crystallization

Phosphorus removal

Wastewater treatment

nt matter & 2006 Elsevie.2006.05.016

thor. Institut de recherchl.: +1 514 872 3942; fax: +

Florent.Chazarenc@umon

A B S T R A C T

Natural apatite has emerged as potentially effective for phosphorus (P) removal from

wastewater. The retention capacity of apatite is attributed to a lower activation energy

barrier required to form hydroxyapatite (HAP) by crystallization. The aim of our study was

to test the P removal potential of four apatites found in North America. Minerals were

collected from two geologically different formations: sedimentary apatites from Florida

and igneous apatites from Quebec. A granular size ranging from 2.5 to 10 mm to prevent

clogging in wastewater applications was used. Isotherms (24 and 96 h) were drawn after

batch tests using the Langmuir model which indicated that sedimentary apatites presented

a higher P-affinity (KL ¼ 0.009 L/g) than igneous apatites (KLE0.004 L/g). The higher density

of igneous material probably explained this difference. P-retention capacities were

determined to be around 0.3 mg P/g apatite (24 h). A 30 mg P/L synthetic effluent was fed

during 39 days to four lab-scale columns. A mixture of sedimentary material (apatite and

limestone 50–50%, w/w) showed a complete P-retention during 15 days which then

declined to 65% until the end of the 39 days lab scale test period. A limitation in calcium

may have limited nucleation processes. The same mixture used in a field scale test showed

60% P-retention from a secondary effluent (30 mg COD/L, 10 mg Pt/L) during 65 days without

clogging.

& 2006 Elsevier Ltd. All rights reserved.

1. Introduction

Adverse effects of eutrophication due to the presence of

anthropogenic phosphorus (P) in surface waters are well-

established. Conventional technologies for P-removal from

wastewater are physical processes (settling, filtration), che-

mical precipitation (with aluminium, iron and calcium salts),

and biological processes that rely on biomass growth

(bacteria, algae, plants) or intracellular bacterial polypho-

sphates accumulation (De-Bashan and Bashan, 2004). Low

cost, low maintenance extensive treatment processes based

on P-retention in filters containing reactive media have been

developed and showed promising results (De-Bashan and

Bashan, 2004). The aim of this project was to favour P-

r Ltd. All rights reserved.

e en biologie vegetale, U1 514 872 9406.treal.ca (F. Chazarenc).

crystallization in a filter located at the downstream end of the

sludge treatment line to reduce clogging by suspended solids.

Natural apatite, a mineral found in igneous, sedimentary and

metamorphic rocks (Nriagu and Moore, 1984), and in teeths

and bones, was shown to favour P-removal from wastewater

(Joko, 1984; Jang and Kang, 2002; Molle et al., 2005). The scope

of this paper was to test and evaluate the P-removal potential

of several apatites found in North America.

The driving mechanisms in the process of P-crystallization

consist essentially in nucleation (precipitation of hydroxya-

patite (HAP): Ca10(PO4)6OH2), followed by crystal growth.

Nucleation is believed to be initiated by the formation of a

metastable precursor, characterised by an amorphous phase

consisting in the combination of calcium phosphates at

niversite de Montreal, 4101 rue Sherbrooke Est, Montreal, Que.,

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 9 6 5 – 2 9 7 12966

a Ca/P molar ratio of between 1 and 1.5 (Zoltek, 1974). The

presence of a crystal acts like a catalyst which lowers the

activation energy barrier between the crystal and HAP. Seeded

precipitation of calcium phosphate is favoured by a high pH,

an increased contact time and a high Ca/P molar ratio. Several

media favouring the seeded precipitation of phosphate from

wastewater were tested, such as magnetite minerals

(Karapinar et al., 2004), steel slags (Shilton et al., 2005; Kostura

et al., 2005; Kim et al., 2006; Korkusuz et al., 2005; Naylor et al.,

2003), calcite limestone and concrete (Song et al., 2006;

Comeau et al., 2001; Molle et al., 2003). P-retention capacity

of mineral apatites for wastewater treatment varied between

2.7 and 4.8 mg P/g apatite in batch experiments which were

followed by successful column tests (Joko, 1984; Molle et al.,

2005).

For P-removal using natural apatites, adsorption mechan-

isms are also believed to play a role and the distinction

between adsorption and crystallization is not clear (Molle

et al., 2005). Considering that their concomitant action is

expected to occur, adsorption is believed to enhance nuclea-

tion/crystallization (Stumm and Morgan, 1996).

In this paper, the P-retention potential of three igneous

apatites from Quebec, one sedimentary apatite from Florida

and one sedimentary limestone was studied. Batch tests were

first performed which enabled to draw isotherms and rank

the samples according to their P-removal potential. Then,

flow-through lab and field scale column tests were conducted

which showed the better potential of sedimentary apatites

mixed with limestone over igneous apatites.

2. Material and methods

2.1. Media tested and sample preparation

Four apatite-containing materials and a limestone, identified

by their suppliers, were tested (Table 1). Samples were

prepared as follows: crushing, rinsing, air-drying and sieving.

During extraction, the associated gangue of the Cargill apatite

was separated by the supplier. This may have reduced the

amount of calcium and hydroxide available in the sample.

Table 1 – Apatite-containing media tested

Media supplier Location Geographiccoordinatesa

M

Cargill (USA) Florida 2714104900N Ph

8115004300W

Soquem (Quebec) Sept-Iles 5011502800N

6612904400W

Arianne (Quebec) Lac St-Jean 4915400500N

7014503000W

Niobec (Quebec) St-H de

Chicoutimi

4813200800N

7110902800W

Graymont

(Quebec)

Joliette 4610004100N

7312702600W

a Geographic Reference System: WGS 84.b Niobec is exploiting niobium oxide (Nb2O5).

In view of full-scale filters where clogging should be

avoided, a granular size of 2.5–10 mm for batch and column

experiments was used.

Particle-size distribution was determined (ASTM D 421-422)

to estimate the specific surface area of the 2.5–10 mm

material assuming a spherical shape. The density of each

material was determined according to ASTM standard test

methods (ASTM C 127-01 for 5–10 mm fraction, and ASTM D

854-02 for 2.5–5 mm fraction).

2.2. Batch tests

Batch tests were conducted with synthetic solutions (KH2PO4

in distiled water) with a gyratory shaker (160 rpm) at 2272 1C.

A mass of 35 g of material was placed in a 1 L glass flask filled

with 700 mL of solution. The pH (8.070.1) and conductivity

(10007100mS/cm) were adjusted using NaOH and NaCl to be

similar to a secondary treated effluent. Isotherm experiments

were conducted at P solutions ranging from 5 to 150 mg P/L

and lasting 24 and 96 h. All samples were analysed for pH,

conductivity, turbidity, suspended solids and orthopho-

sphates (o-PO4).

P adsorption capacity was modelled using Langmuir

isotherms, linearized as

Ce

qe

¼1KLþ

aL

KLCe, (1)

where Ce is the equilibrium concentration of P in solution

(mg P/L) and qe is the amount of P adsorbed (mg P/g material).

The Langmuir constants are graphically defined by KL and aL.

KL describes the affinity between P and the material, and is

represented by the initial slope. The constant KL/aL deter-

mines the P adsorption maximum.

2.3. Column tests

Lab scale column tests were carried out using four 100 mm

diameter columns filled with 2.5 L of material (Niobec, Cargill,

Graymont and Cargill/Graymont mixture 50–50%w/w). A

synthetic P solution (P–PO4 30 mg/L, tap water, pH 7.570.1)

was continuously pumped into the bottom of each column at

ine or prospectdeposit

Identification Geologicalformation

osphate rock mine Francolite-

hydroxyapatite

Sedimentary

Prospect apatite/

magnetite

Fluoroapatite Igneous

Prospect apatite/

ilmenite

Apatite Igneous

Niobium mineb Fluoroapatite Igneous

Limestone quarry Limestone Sedimentary

ARTICLE IN PRESS

Table 2 – Chemical and physical characteristics of the five tested media

Material Chemical composition (%w/w) Density (g/cm3) Specif. area (m2/kg) Particle sizeb

Si Al Ca Mg Fe P d10 (mm) d60 (mm)

Cargill 5.5 0.8 31.4 0.7 1.1 11.3 2.85 0.72 2.8 4.5

Soquem 10.3 4.0 8.6 4.0 22.0 2.4 3.51 0.48 3.0 6.0

Arianne 15.0 4.7 7.0 4.7 19.0 2.0 3.37 0.53 3.0 6.0

Niobec 2.7 0.7 25.7 6.3 3.6 5.2 3.00 0.58 3.0 6.0

Graymont n.a.a n.a. n.a. n.a. n.a. n.a. 2.72 0.57 3.5 7.0

a n.a: not available.b d10, d60: grain size of 10% and 60% material by weight.

Table 3 – Influent and effluent characteristics of lab scalecolumns test

Conductivity(mS/cm)

pH Ca2+

(mg/L)

Synthetic

influent

531726 7.970.1 31.870.7

Niobec effluent 575763 7.670.3 31.170.9

Cargill effluent 491739 7.170.4 2.872.5

Graymont

effluent

571749 7.970.2 32.372.6

Cargill/

Graymont

effluent

560751 7.670.3 29.171.7

WAT E R R E S E A R C H 40 (2006) 2965– 2971 2967

a flow rate of 0.7 L/d for 39 days (hydraulic retention time of

1.5 d considering void space [HRTv]). Samples were taken at

the inlet and outlet of the columns, every day during the first

15 days and every three days until day 39.

Field scale tests were carried out using a 6.5 L total volume

column filled with a mix of Cargill apatite and Graymont

limestone (50–50%w/w). The columns were installed at a fish

farm plant and were fed with a constructed wetland effluent

(30 mg COD/L, 10 mg Pt/L) such that the HRTv was initially set

at 0.33 d and operated during 65 days.

The concentration was fixed at of 30 mg P/L in lab scale

tests, based on annual average concentrations measured in a

wetland effluent from a fish farm plant. Since the field scale

tests were conducted during summer, P retention was more

important in the constructed wetland resulting in a lower P

concentration of the wetland effluent (10 mg P/L).

2.4. Analytical methods

Orthophosphates were measured using the Quickchem flow

injection analysis method # 10-115-01-1-Q derived from the

automated ascorbic acid reduction method (Standard Meth-

ods, 1998). Ionic measurements conducted on samples

collected from column tests, as well as chemical composition

of tested minerals, were analysed by the atomic absorption

spectroscopy method (Standard Methods, 1998).

3. Results and discussion

3.1. Chemical and physical media characteristics

The Ca and P content of the Cargill apatite (Table 2) was in the

same range as those tested by Molle et al. (2005) which

containing 37.3% Ca and 16.8% P, and materials tested by Joko

(1984) containing 37.6% Ca and 15.2% P. The Ca/P molar ratio

of 2.78 of the Cargill apatite, compared to the ratio of 1.67 of

pure hydroxyapatite, suggests that some P was substituted by

CO32� which is generally the case in carbonate-substituted

(low-grade) phosphate rock like francolite (Nriagu and Moore,

1984). The presence of calcite or dolomite in the gangue

associated to the apatite may also have contributed to

increase this ratio (Slansky, 1986). The occurrence of iso-

morphic substitutions in apatite from Morocco (Molle et al.,

2005), giving a lower density of 2.48 g/cm3 compared to that

from Cargill (Table 2), resulted in an increase of the internal

porosity (Zapata and Roy, 2004) and in a larger surface of

reaction (Table 3).

Igneous rocks are characterized by a higher density (Table

2), suggesting a lower internal porosity due to their mode of

crystallization in magma (Zapata and Roy, 2004). Apatites

from Soquem, Arianne and Niobec showed concentrations

varying between 2% and 5% P (Table 2) which is in the range of

natural igneous apatite (Nriagu and Moore, 1984). Aluminium,

magnesium and iron were also present in a relatively high

proportion, notably in the Soquem and Arianne apatites

(Nriagu and Moore, 1984; Al 0.05%).

3.2. Batch tests

A solubilization batch test (in distiled water) showed no

electrical conductivity and pH variation with apatite samples,

while dissolution of limestone from Graymont led to an

increase in pH (from 8.0 to 9.5) and in conductivity. Langmuir

isotherms determined after 24 h, showed a limited affinity

between apatites and P, as indicated by the smooth initial

slopes (Fig. 1). Graymont limestone 24 h isotherms showed a

constant affinity with P either because adsorption took place

regardless of the material, or due to the increase of adsorbate

which proportionally augmented adsorbing sites (Sposito,

1989).

ARTICLE IN PRESS

0.0

0.1

0.2

0.3

0 20 40 60 80 100 120 140

Ce (mgP/L)

qe

(mg

P/g

mat

eria

l)

Cargill Langmuir Cargill

Graymont Langmuir Graymont

Cargill Langmuir Cargill

Graymont Langmuir Graymont

0.0

0.1

0.2

0.3

0 20 40 60 80 100 120 140

Ce (mgP/L)

qe

(mg

P/g

mat

eria

l)

Soquem Langmuir SoquemArianne Langmuir ArianneNiobec Langmuir Niobec

Soquem Langmuir SoquemArianne Langmuir ArianneNiobec Langmuir Niobec

0.0

0.2

0.4

0.6

0.8

1.0

0 20 40 60 80 100 120 140

Ce (mg P/L)

qe

(mg

P/g

mat

eria

l)

0.0

0.2

0.4

0.6

0.8

1.0

0 20 40 60 80 100 120 140

Ce (mg P/L)

qe

(mg

P/g

mat

eria

l)

(A) (B)

(D)(C)

Fig. 1 – Equilibrium isotherms of phosphate (22 1C) (Cargill apatite and Graymont limestone: (A) 24 h, (C) 96 h; Soquem,

Arianne and Niobec apatites: (B) 24 h, (D) 96 h).

WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 9 6 5 – 2 9 7 12968

P-retention capacities calculated from the 24 h Langmuir

isotherms (in mg P/g material: Graymont: 1.09; Arianne: 0.41;

Soquem: 0.37; Cargill: 0.31; Niobec: 0.28) were low compared

to those found in the literature (2.7–4.8 mg P/g apatite; Molle

et al., 2005; Joko, 1984). According to 96 h isotherms (Fig. 1), a

greater affinity was found between P and sedimentary

materials. Igneous apatites were found to be less reactive

with P, possibly due to their lower internal porosity. Con-

sidering the importance of calcium in P-retention mechan-

isms, the limited dissolution observed for the Cargill apatite

and the absence of calcium in the synthetic P-solution

possibly led to a limitation in available Ca.

The low specific surface area of the material used in this

study, due to the relatively coarse material, influenced the

P-removal efficiency. Furthermore, apatite crystal sizes were

relatively small (0.1–0.5 mm in Niobec apatite), compared to

the granular size selected (2.5–10 mm), reducing the rates of P

adsorption. Using 24 h Langmuir isotherms, the tested

materials ordered by decreasing affinity based on KL coeffi-

cients were as follows: Cargill4Arianne4Niobec4Soquem4-

Graymont. Some indications in batch experiments suggest

that adsorption occurred almost instantaneously and was

rapidly followed by precipitation. Adsorption being a surface

phenomenon, it is considered that the high densities

associated with igneous materials was not favourable to

these types of apatites as adsorbents, in which poor internal

porosities were expected. Since the Cargill apatite presented a

lower density compared to igneous apatites, this could

explain the high affinity found for the sedimentary

apatite. No improvement for the igneous apatite was appar-

ent after 96 h.

Fluoroapatite presents a pH zero point of charge (pHZPC)

varying between 4 and 6, a lower value than for hydroxyapa-

tite (7.6–8.6), which may discredit compounds rich in fluor-

apatite (Arianne, and Soquem) as favourable substrates for

phosphate adsorption. Apatites are, however, largely asso-

ciated with other minerals in their natural state, and

adsorption may occur in non-apatitic mineral support

(especially in Arianne and Soquem media).

Another linearization of Langmuir isotherm (Kd ¼ qe/

Ce ¼ f(qe)) allowed the identification of two potential

P-retention mechanisms for Cargill apatite (results not

shown). This was illustrated in others studies (Søvik and

Kløve, 2005; Molle et al., 2005) and could correspond to an

adsorption mechanism at low P content followed by pre-

cipitation mechanism at great P content. But the maximum

initial P concentration in our batch experiment (150 mg P/L)

was too low to clearly identify these two potential P-retention

mechanisms and to go further in the discussion.

3.3. Lab scale column experiments

Considering batch test results, only one of the igneous apatite

(Niobec) was tested in lab scale column experiments. For

increased availability of calcium the Cargill apatite was mixed

with Graymont limestone.

P-retention by the Niobec apatite was poor and the column

seemed to become saturated after 15 days (Fig. 2), confirming

the relatively poor efficiency of this type of material. During

the first 15 days the retention by Cargill and Cargill/Graymont

mixture was comparable and close to 100%. Then a slight

decrease appears for Cargill’s. The removal of the associated

ARTICLE IN PRESS

0

100

200

300

400

500

600

0 100 200 300 400 500 600 700 800 900

Padded (mgP cumulated)

P r

emo

ved

(m

gP

cu

mu

late

d)

NiobecCargillGraymontCargill/Graymont

100% retention

50% retention

Fig. 2 – P-removal as a function of P added (lab scale experiment).

Table 4 – Influent and effluent characteristics of field scale column

Sampling point pH Cond (mS/cm) COD (mg/L) Total P (mg P/L)

Inlet 7.170.3 241768 32711 7.672.3

Outlet 7.870.4 286738 2877 1.971.1

WAT E R R E S E A R C H 40 (2006) 2965– 2971 2969

gangue of the Cargill apatite may have reduced the amount of

calcium and hydroxide availability of the sample thus

reducing its nucleation performance. In column tests using

untreated natural apatites, Molle et al. (2005) showed a

significant dissolution of calcium carbonate (5 mg/L Ca)

leading to a pH increase from 7.0 to 8.0.

Graymont limestone showed a relatively stable retention

over time that remained near 50%. Effluent analysis at the

outlet of each column showed for Cargill’s a significant

consumption of Ca2+ over time and a decrease in pH

(Table 4) which suggested HAP formation and crystallization.

Results showed that Graymont limestone favoured P

precipitation as it represented a source of Ca2+ and OH�.

The P retention curve for the Cargill/Graymont mixture

appeared to be the addition of the curves for the two distinct

materials, avoiding the initial poor removal observed with

Graymont’s alone. This suggested that Cargill’s probably

provided a better seed precipitant than Graymont’s. At the

outlet of the Cargill/Graymont mixture, the pH and Ca

concentration were higher than in Cargill’s and appeared to

favour P-retention (which equals about 200 g of P retained per

m3 of column after 39 days), but they were probably not high

enough to obtain more than 60% P-retention.

In case of P levels of about 15–20 mg/L, the minimum inlet

Ca level recommended for Ca–P precipitation, according to

Jang and Kang (2002), should be 40–60 mg/L. For the field scale

column test, it was chosen to use a single column of a

limestone/apatite mixture as it represented the most promis-

ing media tested. The Cargill/Graymont mixture seemed to be

the best compromise to increase the pH, dissolve some Ca

and favour HAP precipitation.

3.4. Field scale column experiments

The column was installed downstream of a constructed

wetland of a fresh water fish farm wastewater treatment

plant. Despite the variable composition of the influent

(Table 4), the column worked with a total P retention

efficiency of 60% without showing signs of either clogging or

salting out (Fig. 3).

During the field scale test, there was a conflicting effect of

pH. On one hand a higher pH at the column inlet favore Ca

dissolution, but on the other hand it led to a reduction of HAP

formation by reducing the available OH� in solution. Having a

too great pH variability at the inlet could become a problem to

ensure a long term P retention, the best pH being around

neutrality for HAP.

The field scale column efficiency was enough to reduce

P-level to reach 2 mg P/L but the long-term stability of the

process remains to be established. The next step would be to

test mesocosm columns under a prolonged period (e.g. 1 year)

under varying influent and operational conditions such as Ca

availability, retention time and pH.

4. Conclusion

Phosphorus affinity was found to be higher with Cargill

sedimentary apatite as substrate, compared to igneous

apatites in batch experiments. Sedimentary apatites are

believed to favour crystallization of HAP due to intrinsic

characteristics, notably their internal porosity. The presence

of secondary minerals associated to the gangue would be

ARTICLE IN PRESS

Inlet ≈ 15.5 mg P/d

Retention ≈ 9.5 mg P/d

0

200

400

600

800

1000

1200

0 10 20 30 40 50 60 70

Elapsed time (d)

Rem

ove

d P

(m

gP

cu

mu

late

d)

Inlet cumulated PRetained cumulated P

Fig. 3 – P-retention efficiency during the field scale experiment.

WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 9 6 5 – 2 9 7 12970

beneficial by contributing to a source of Ca, which is essential

for Ca–P precipitation. Relatively low P-retention capacities

were obtained, probably because the available Ca from

dissolution rapidly became depleted. Testing pure Cargill

material including the gangue would be of interest.

With igneous apatites, neither HAP crystallization nor Ca–P

precipitation were believed to have occurred considering the

higher affinity of P with other metals like Fe, Mg, Al as pointed

out in other studies. Crystallography tests could confirm this

interpretation.

Best pH conditions are conflicting as a low pH would favour

Ca dissolution and a high pH would favour HAP crystal-

lization. In small scale wastewater treatment plants, pH

fluctuations may be important thus reducing P-retention

availability. A source of calcium and hydroxide should be

provided to avoid such conflicting operation conditions.

Sedimentary and igneous North American apatites showed

low and poor retention capacity for P-removal, respectively.

Nevertheless, the potential of apatites for crystallization and

P-recycling merits attention. Further studies should be

conducted with regards to the critical conditions associated

to pH, calcium, and orthophosphates concentrations.

Acknowledgements

The authors thank Denis Bouchard for technical support and

Dwight Houweling for reviewing an earlier draft of the

manuscript. This research was financed by the Natural

Sciences and Engineering Research Council of Canada.

R E F E R E N C E S

Comeau, Y., Brisson, J., Reville, J.-P., Forget, C., Drizo, A., 2001.

Phosphorus removal from trout farm effluents by constructed

wetlands. Water Sci. Technol. 44 (11–12), 55–60.

De-Bashan, L.E., Bashan, Y., 2004. Recent advances in removingphosphorus from wastewater and its future use as fertilizer(1997–2003). Water Res. 38 (19), 4222–4246.

Jang, H., Kang, S.H., 2002. Phosphorus removal using cow bone inhydroxyapatite crystallization. Water Res. 36 (5), 1324–1330.

Joko, I., 1984. Phosphorus removal from wastewater by thecrystallization method. Water Sci. Technol. 17, 121–132.

Karapinar, N., Hoffmann, E., Hahn, H.H., 2004. Magnetite seededprecipitation of phosphate. Water Res. 38 (13), 3059–3066.

Kim, E.-H., Lee, D.-W., Hwang, H.-K., Yim, S., 2006. Recovery ofphosphates from wastewater using converter slag: Kineticsanalysis of a completely mixed phosphorus crystallizationprocess. Chemosphere 63 (2), 192–201.

Korkusuz, E.A., Beklioglu, M., Demirer, G.N., 2005. Comparison ofthe treatment performances of blast furnace slag-based andgravel-based vertical flow wetlands operated identically fordomestic wastewater treatment in Turkey. Ecol. Eng. 24 (3),185–198.

Kostura, B., Kulveitova, H., Lesko, J., 2005. Blast furnace slags assorbents of phosphate from water solutions. Water Res. 39 (9),1795–1802.

Molle, P., Lienard, A., Grasmick, A., Iwema, A., 2003. Phosphorusretention in subsurface constructed wetlands: Investigationsfocused on calcareous materials and their chemical reactions.Water Sci. Technol. 48 (5), 75–83.

Molle, P., Lienard, A., Grasmick, A., Iwema, A., Kabbabi, A., 2005.Apatite as an interesting seed to remove phosphorus fromwastewater in constructed wetlands. Water Sci. Technol. 51(9), 193–203.

Naylor, S., Brisson, J., Labelle, M.-A., Drizo, A., Comeau, Y., 2003.Treatment of freshwater fish farm effluent using constructedwetlands: The role of plants and substrate. Water Sci. Technol.48 (5), 215–222.

Nriagu, J.O., Moore, P.B., 1984. Phosphate Minerals. Springer-Verlag, Heidelberg.

Shilton, A., Pratt, S., Drizo, A., Mahmood, B., Banker, S., Billings, L.,Glenny, S., Luo, D., 2005. ‘Active’ filters for upgrading phos-phorus removal from pond systems. Water Sci. Technol. 51(12), 111–116.

Slansky, M., 1986. Geology of Sedimentary Phosphates. NorthOxford Academic Publishers Ltd.

Song, Y., Weidler, P.G., Berg, U., Nuesch, R., Donnert, D., 2006.Calcite-seeded crystallization of calcium phosphate for phos-phorus recovery. Chemosphere 63 (2), 236–243.

ARTICLE IN PRESS

WAT E R R E S E A R C H 40 (2006) 2965– 2971 2971

Sposito, G., 1989. The Chemistry of Soils. Oxford University Press,

New-York.

Standard Methods (Standard Methods for the Examination of

Water and Wastewater), 1998. 20th ed., American Public

Health Association/American Water Works Association/Water

Environment Federation, Washington, DC, USA.

Søvik, A.K., Kløve, B., 2005. Phosphorus retention processes in

shell sand filter systems treating municipal wastewater. Ecol.

Eng. 25 (2), 168–182.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry—ChemicalEquilibria and Rates in Natural Waters, third ed. Wiley-Interscience Publishers, Iowa.

Zapata, F., Roy, N.R., 2004. Use of phosphate rocks for sustainableagriculture, Report ISBN 92-5-105030-9, Food and AgricultureOrganization of the United Nations-FAO Land and WaterDevelopment Division and International Atomic EnergyAgency, Rome, Italy.

Zoltek Jr., J., 1974. Phosphorus removal by orthophosphatenucleation. J. Water Pollut. Control Fed. 46, 2498–2520.