International Journal of Environmental Science and Technology

https://doi.org/10.1007/s13762-018-02179-1

ORIGINAL PAPER

Phosphorus recovery as calcium phosphate by a pellet reactor

pre‑treating domestic wastewater before entering a constructed
wetland
A. M. Rugaika1,2 · R. Van Deun3 · K. N. Njau2 · B. Van der Bruggen1,4

Received: 1 August 2018 / Revised: 21 November 2018 / Accepted: 17 December 2018

© Islamic Azad University (IAU) 2019

Abstract
Horizontal subsurface flow constructed wetlands poorly remove phosphorus from wastewater, resulting in phosphorus levels
above the required limits in constructed wetland effluents. Since a pellet reactor can recover phosphorus through calcium
phosphate precipitation/crystallization, using it as a pre-treatment system prior to horizontal subsurface flow constructed
wetland could be a sustainable solution for phosphate scarcity. The operational conditions required for phosphate recovery
in a pellet reactor were evaluated and compared with the Visual MINTEQ version 3.0 with the aim of checking its suitability
to simulate the pellet reactor removal efficiencies. Such conditions include the initial phosphate concentration, pH, [Ca]/
[P] molar ratio and hydraulic loading rate. The results showed an increase in phosphate removal efficiency with increased
initial phosphate concentration, pH, [Ca]/[P] molar ratio and decreased hydraulic loading rate. However, the model calcula-
tion gave higher removal efficiencies than experimental results due to its inability to take into account the system kinetics
which is an important component in pellet reactor operation and its assumption that precipitation reactions take place at
constant pH. The effects of carbonate on phosphate precipitation were also investigated, and the removal efficiency of
61.9% without carbonate was improved to 63.2, 64.3 and 66.4% with a carbonate concentration of 0.25 mM, 0.5 mM and
2 mM, respectively, at pH 9, initial phosphate concentration of 1 mM, [Ca]/[P] molar ratio of 1.5 and hydraulic loading
rate of 57 m/h. Thus, the presence of carbonates in domestic wastewater is advantageous as it promotes calcium phosphate
precipitation in a pellet reactor.

Keywords  Horizontal subsurface flow constructed wetland · Phosphorus · Removal efficiency · Saturation index ·
Operational condition

Introduction

Phosphorus (P) is a limiting element for phytoplankton

growth and, therefore, one of the leading causes of eutrophi-
cation in natural waters. The release of inadequately treated
Editorial responsibility: M. Abbaspour.
wastewater in natural water bodies causes this destructive
* A. M. Rugaika and devastating process into the aquatic environment. This
anita.rugaika@nm‑aist.ac.tz resulted into a growing interest in research studies with the
1 aim of developing technologies that remove P to the levels
Department of Chemical Engineering, KU Leuven,
Celestijnenlaan 200F, 3001 Leuven, Belgium that meet the required limits. Since P is a non-renewable
2 resource and there were assumptions that the existing rock
The Nelson Mandela African Institution of Science
and Technology, P.O. Box 447, Arusha, Tanzania phosphate reserve will decline in the next 50–100 years
3 (Steen 1998), the attention shifted into P recovery with the
Thomas More University College, Kleinhoefstraat 4,
2440 Geel, Belgium aim of improving aquatic environment and saving the phos-
4 phate rock from depletion (Tran et al. 2014).
Faculty of Engineering and the Built Environment, Tshwane
University of Technology, Private Bag X680, Pretoria 0001, A number of approaches have been applied to recy-
South Africa cle P from wastewater, including chemical precipitation,

13
Vol.:(0123456789)

biological P removal, crystallization technology and sludge-
based processes (Morse et al. 1998). The authors went fur-
ther explaining that chemical precipitation technology can
be easily operated and installed and that it promotes high P
recycling. However, the technology produces a high amount
of sludge, requires chemicals, and recycled P tends to vary.
Biological P removal is a complex technology to install and
to operate, and it is more difficult in handling the gener-
ated sludge. On the other hand, it does not require chemicals
and apart from P it can also remove nitrogen. Sludge-based
processes are a complex technology that requires chemical
input, but it has increased the sludge value. Crystallization
of P in a fluidized bed reactor, also denoted as pellet reactor
in accordance with the terminology used in drinking water
treatment, is a widely used technique as it produces high-
purity phosphate, has a low water content in the produced
pellets, and pellets can be reused (Giesen 1999; Mahvi
et al. 2005). Apart from P recovery, a pellet reactor could
be used in heavy metal recovery (Zhou et al. 1999; Guillard
and Lewis 2001), fluoride recovery (Aldaco et al. 2007) and
softening of drinking water (Mahvi et al. 2005).
A pellet reactor is a fluidized system in which calcium
phosphate is principally precipitated/crystallized on seed
grains, typically sand present in the reactor. Precipitation of
calcium phosphate is induced by pH adjustment and calcium
ions addition in water containing P. The precipitation of a cal-
cium salt is a complex subject due to the formation of different
phases at different experimental conditions, such as brushite,
hydroxyapatite and amorphous calcium phosphate. The differ-
ent conditions leading to different phases include pH, super-
saturation level, temperature, [Ca]/[P] molar ratio, calcium and
phosphate ions concentrations, ionic strength and ion types
(Baronne and Nancollas 1977; Boskey and Posner 1973; Van
Kemenade and de Bruyn 1987; Montastruc et al. 2003).
A number of studies have been conducted with the aim
of improving P recovery from wastewater in a pellet reac-
tor. A study by Crutchik et al. (2017) used industrial grade
Mg(OH)2 as a source of magnesium and alkali for struvite
crystallization which resulted into improved P recovery.
Tran et al. (2014) used selectrodialysis as a technique to
pre-concentrate phosphate in the feed prior to the crystal-
lization process in a pellet reactor with the aim of obtaining
a high efficiency for phosphate recovery. Li et al. (2013)
tested struvite crystallization in a high-strength nitrogen
(N) and P stream and observed over 95% of N, P and Mg
removals being achieved at pH of 7.6. Song et al. (2007)
compared seed materials apatite, juraperle and phosphate-
modified juraperle and observed phosphate-modified jurap-
erle to be the most appropriate seed material for P recovery
by crystallization of calcium phosphate. Song et al. (2006)
13
International Journal of Environmental Science and Technology
showed calcite to be effective as seed material for calcium
phosphate crystallization from hard water. A study by Seck-
ler et al. (1996) explained that with proper pH adjustment
a pellet reactor could recover 80–95% of phosphate in the
presence of carbonate and magnesium ions through calcium
phosphate precipitation/crystallization and 50–65% in their
absence.
Horizontal subsurface flow constructed wetlands (HSSF-
CWs) are systems that treat wastewater in the same way
as natural wetlands; however, they are operated in a more
controlled environment. Despite the fact that these systems
perform well for the reduction in some pollutants, they have
been observed to poorly remove P from wastewater resulting
into P levels above the required limits in CWs effluent. A
study by Njau et al. (2011) showed the phosphate removal
efficiency of only 21, 29 and 59% for Shinyanga Prison CW,
Moshi Urban Water and Sewerage Authority CW (MUWSA-
CW) and Ruaha Secondary School CWs, respectively. The
nuisance of eutrophication in natural water bodies and the
urge to meet the effluent standard before wastewater dis-
charge has accelerated into a number of studies with the
aim of improving the removal of residual P in HSSF-CWs
before the discharge of the effluent to a receiving water body.
A study by Seo et al. (2005) showed that oyster shells, when
added to the CW media, enhance P adsorption and hence
increase the longevity of the system due to saturation of P.
Brix et al. (2001) suggested mixing of calcite and crushed
marble with sand or gravel media in HSSF-CWs as they
have a high P sorption capacity. A study by Drizo et al.
(1999) suggested the use of shale as a substrate media for
HSSF-CW as it enhances P removal from wastewater. Drizo
et al. (2006) reported the use of electric arc furnace steel
slag as another solution for P removal in CWs through both
adsorption and precipitation mechanisms. Zhu et al. (1997)
reported lightweight aggregates (LWA) material to remove
up to 95% of P from HSSF-CWs. Since these sorption and
precipitation mechanisms are finite, the substrate media will
eventually need to be removed/changed after reaching satu-
ration, which is expensive. Besides, the possibility of P to
be returned back to the CW system is high under a certain
chemical condition. Thus, the need of a sustainable means to
recover P from a HSSF-CW and the ability of a pellet reactor
to produce high-purity phosphate and reusable pellets with
low water content has prompted a need to use a pellet reactor
to pre-treat domestic wastewater before being further treated
in a HSSF-CW system. This study therefore uses a pellet
reactor to improve the P removal efficiency of HSSF-CW
system that treats domestic wastewater.
Since a pellet reactor requires a high initial P concentra-
tion for efficient P recovery (Tran et al. 2014), this study will
International Journal of Environmental Science and Technology
use it as a pre-treatment system with the aim of capturing
high P concentrations from wastewater. The use of a pellet
reactor system aims at recovering phosphate for application
on agriculture prior to the effluent discharge into the HSSF-
CW system. Despite the fact that treated wastewater could be
used directly in agriculture (Kihila et al. 2014), pellet reac-
tor will recover phosphate as a solid material, which can be
stored for future use and becomes a substitute for the threat-
ened phosphate rock, which is depleting. Specifically, this
study aims at determining appropriate conditions required
by the pellet reactor acting like a pre-treatment system to
recover P from domestic wastewater before it enters a HSSF-
CW system for further treatment. Moreover, the suitability
of the visual MINTEQ version 3.0 software to simulate the
pellet reactor removal efficiencies will be assessed. This
research was conducted in Leuven, Belgium, in 2016.
Materials and methods
Wastewater characteristics
Synthetic wastewater used in the experiments was prepared
by mixing K ­ H2PO4 with deionized water to obtain the
desired phosphate concentration (0.5 mM, 1 mM, 2.5 mM
and 4 mM) in a 60-L tank. In order to vary the pH of the pre-
pared wastewater, 6 M NaOH solution was used. Required
carbonate concentrations (0.25 mM, 0.5 mM and 2 mM)
were prepared by mixing appropriate amounts of N ­ a2CO3 in
­ a2+ ions were introduced by adding
synthetic wastewater. C
­CaCl2.2H2O in synthetic wastewater. Therefore, synthetic
wastewater without organic matter at 25 °C was comprised
of pH (7 to 11), PO3−
4 (0.5 to 4 mM), CO3 (0.25 to 2 mM),
2−
­K (0.5 to 4 mM), ­Na (0.5 to 4 mM), ­Cl− (3 to 12 mM),
+ +
­H+ (1 to 8 mM) and ­Ca2+ (0.75 to 6 mM) depending on the
experiment.
Fluidized laboratory‑scale pellet reactor operation
A laboratory-scale pellet reactor (glass reactor) was used to
conduct experiments with the aim of removing phosphate
from synthetic wastewater by precipitation in a pellet reac-
tor acting like a crystallator (DHV Consulting Engineers
1991). Operational conditions such as pH, [Ca]/[P] molar
ratio, initial phosphate concentrations and hydraulic loading
rate (HLR) that promote precipitation process were inves-
tigated. Sand (400 g) from Polytechnisch Bedrijf Compa-
ktuna Company in Ghent, Belgium, composed of 99.5%
­SiO2 and 0.03% ­Fe2O3 which was used as seed material was
filled in the reactor. In operation, the wastewater contain-
ing phosphate was pumped in an upward direction through
the fluidized bed with a HLR ranging from 44 to 70 m/h at
different pH. A pH meter (Orion model 420 A) was used
for pH measurement, while 6 M NaOH was used for adjust-
ing the pH. While wastewater was pumped upwards without
recirculation, a calcium chloride solution ­(CaCl2.2H2O) was
also added at a rate of 3 m/h, a few centimetres away from
the wastewater entry point in order to provide ­Ca2+ ions to
react with phosphate ions for calcium phosphate formation.
Precipitation process occurs when the solubility of calcium
phosphate is exceeded by introducing ­Ca2+ ions in waste-
water and adjusting its pH. The fluidized sand bed provided
a large specific surface area for the precipitation of calcium
phosphate to take place. When more and more calcium phos-
phate precipitate was coated on the sand surface, the sand
became heavy and moved to the bottom of the reactor. Dif-
ferent operational conditions were tested to determine the
appropriate conditions for phosphate precipitation. Thus,
a number of experiments were run in a pellet reactor at a
fixed temperature of 25 °C and an ionic strength calculated
from the reaction equilibrium. In the first experiment, an
initial phosphate concentration of 0.5 mM, 1 mM, 2.5 mM
and 4 mM was tested at pH 9, [Ca]/[P] molar ratio of 1.5
and HLR of 57 m/h. In the second experiment, the condi-
tions were set at an initial phosphate concentration of 1 mM,
[Ca]/[P] molar ratio of 1.5, HLR of 57 m/h except pH was
varied from 7 to 11. Third experiment was set at an initial
phosphate concentration of 1 mM, pH of 9, HLR of 57 m/h
except the [Ca]/[P] molar ratio was varied from 1.5 to 2.5.
Fourth experiment was set at an initial phosphate concentra-
tion of 1 mM, pH of 9, [Ca]/[P] molar ratio of 1.5 except
HLR was varied from 44 to 70 m/h. The last experiment was
set at an initial phosphate concentration of 1 mM, pH of 9,
[Ca]/[P] molar ratio of 1.5, HLR of 57 m/h, and carbonate
concentrations of 0.25 mM, 0.5 mM and 2 mM were added
with the aim of determining the effect of carbonate ions on
the precipitation process. Each experiment was run for 4 h.
pH 9 was mostly used in this study because it is practical,
especially when considering the reality of recovering phos-
phate from wastewater and discharge of treated wastewater
to the environment. The input data for visual MINTEQ ver-
sion 3.0 software are summarized in Table 1.
Probable reactions
Different calcium phosphate phases that can be formed at
different operational conditions are given in Table 2.
13
 International Journal of Environmental Science and Technology

Table 1  Summary of input data Experiment pH H+ (M) K+ (M) Ca2+ (M) Cl− (M) Na+ (M)
for visual MINTEQ version 3.0 4 (M)
PO3− 3 (M)
CO2−
software 1 9 0.001 0.0005 0.0005 0.00075 0.0015
9 0.002 0.001 0.001 0.0015 0.003
9 0.005 0.0025 0.0025 0.00375 0.0075
9 0.008 0.004 0.004 0.006 0.012
2 7 0.002 0.001 0.001 0.0015 0.003
8 0.002 0.001 0.001 0.0015 0.003
9 0.002 0.001 0.001 0.0015 0.003
10 0.002 0.001 0.001 0.0015 0.003
11 0.002 0.001 0.001 0.0015 0.003
3 9 0.002 0.001 0.001 0.0015 0.003
9 0.002 0.001 0.001 0.002 0.004
9 0.002 0.001 0.001 0.0025 0.005
4 9 0.002 0.001 0.001 0.0015 0.003
9 0.002 0.001 0.001 0.0015 0.003
9 0.002 0.001 0.001 0.0015 0.003
5 9 0.002 0.001 0.001 0.0015 0.003 0.00025 0.0005
9 0.002 0.001 0.001 0.0015 0.003 0.0005 0.001
9 0.002 0.001 0.001 0.0015 0.003 0.002 0.004

Table 2  Different calcium phosphate phases formed at different oper-

Visual MINTEQ version 3.0
ational conditions
The software was used for geochemical modelling. This
Mineral Reactions in a pellet reactor
is a chemical equilibrium model for water that calculates
Tricalcium phosphate 5Ca2+ + 2PO3−
4  → Ca3(PO4)2
metal speciation, solubility equilibria, sorption, etc. This
Octacalcium phosphate 4Ca2+ + H+ + 3PO3− thermodynamic simulation tool was used to predict the
4  + 3H2O → Ca4H(PO4)3
.3H2O precipitation potential of calcium with phosphate and car-
Monotite Ca2+ + HPO2−
4  → CaHPO4
bonate that were to be conducted in the laboratory. Con-
Brushite Ca2+ + HPO2− centrations of all ions present in the solution were fed in
4  + 2H2O → CaHPO4.2H2O
Hydroxyapatite 5Ca2+ + 3PO3− − the model, with the temperature fixed at 25 °C. The ionic
4  + OH  → Ca5(PO4)3(OH)
strength was calculated from all ion concentrations present
in the solution. Thus, the solid phase, species of the solid
Phosphate removal efficiency phase and the precipitation efficiency of the solid were
given depending on the input conditions. The suitability of
(
Ci − Co
) the model to simulate the pellet reactor removal efficien-
Removal efficiency(%) = ∗ 100% cies was checked by comparing the experimental results
Ci
and the results obtained from the model calculation. The
assumptions of this theoretical model include:
where Ci is the input phosphate concentration ­(mgPO4/L)
and Co is the output phosphate concentration ­(mgPO4/L).
1. The reactions between calcium and phosphate reach
equilibrium condition.
Sample analysis
2. Reactions take place at constant pH and temperature of
25 °C.
Two samples from the pellet reactor outlet were taken at
3. The solution is supersaturated.
30-min interval for the analysis of the remaining phosphate.
4. Only thermodynamic concept is taken into considera-
Ion chromatography (Dionex ICS-2000) was used to deter-
tion.
mine phosphate concentrations.

13
International Journal of Environmental Science and Technology

concentration increases the supersaturation index (SI), hence

Fig. 1  Phosphate removal efficiency at different initial phosphate con-
centrations at initial operational conditions of pH 9, [Ca]/[P] molar
ratio of 1.5 and HLR of 57 m/h
Statistical data analysis
Statistical data analysis was done using Origin Pro 9.0 soft-
ware (Origin Lab Corporation, Northampton, MA, USA).
Before the analysis, data were subjected to a normality test.
Normally distributed data were analysed using a parametric
ANOVA test and non-normally distributed data using a non-
parametric Kruskal–Wallis ANOVA. Significant differences
were represented by P < 0.05.
an increase in precipitation efficiency (Table 3). From the
model calculation, the phosphate removal efficiency of
hydroxyapatite was 89.9% for all initial phosphate concen-
trations tested. This was due to incomplete precipitation
reaction caused by [Ca]/[P] molar ratio of 1.5 used.
Despite the fact that the kinetics slowed down the precipi-
tation process in the pellet reactor, the limited reaction time
in the system also contributed to the experimental results
being significantly lower in removal efficiency (P = 0.016)
than the model calculation as the reaction equilibrium could
not be reached (Tran et al. 2014). Moreover, decreasing pH
during the precipitation reaction in a pellet reactor could
also be the reason for low removal efficiency of experi-
mental results compared to the model calculation with con-
stant pH. HPO2− 4 which has high affinity for calcium is the
predominant form of phosphate at pH 9. Thus, during the
precipitation reaction, the solution equilibrium shifts to the
right (reaction 2) with the release of hydrogen ions which
lower the pH, hence a decrease in the SI (Mekmene et al.
2009; Tran et al. 2014). However, the difference between the
experimental results and the model calculation was reduced
at phosphate concentration of 4 mM due to increased SI
which as a result made the reaction more efficient as the
induction time was reduced.
H3 PO4 ↔ H+ + H2 PO−4 (1)
H2 PO−4 ↔ H+ + HPO2− (2)
4

HPO2−
4
↔ H+ + PO3−
4 (3)

Results and discussion

Initial phosphate concentration

The experimental results showed an increase in phosphate

removal efficiency with an increase in initial phosphate con-
centration at pH 9, HLR of 57 m/h and [Ca]/[P] ratio of 1.5.
From Fig. 1, the removal efficiency was 54.1% at the initial
concentration of 0.5 mM and 77.1% at the initial concentra-
tion of 4 mM. This is in agreement with the study by Song
et al. (2002), which explained that the increase in phosphate

Table 3  SI of different calcium phosphate phases at different initial

phosphate concentration with pH 9 and [Ca]/[P] of 1.5

Mineral 0.5 mM 1 mM 2.5 mM 4 mM

Tricalcium phosphate 4.9 5.9 7.2 7.8

Octacalcium phosphate 4.6 6 7.8 8.6
Monotite − 0.7 0.3 0.8 1.1
Brushite − 0.3 0.1 0.5 0.8 Fig. 2  Phosphate removal efficiency at different initial pH values at
Hydroxyapatite 15.6 17.3 19.3 20.3 operational conditions of initial phosphate concentration of 1  mM,
[Ca]/[P] molar ratio of 1.5 and HLR of 57 m/h

13
 International Journal of Environmental Science and Technology

Table 4  SI of different calcium phosphate phases at different pH with to be 4.9%, while at pH 11 the removal efficiency was
initial phosphate concentration of 1 mM and [Ca]/[P] of 1.5 71.6%. The results obtained from the model calculation
Mineral pH 7 pH 8 pH 9 pH 10 pH 11 showed only a slight increase in removal efficiency with
an increase in the initial pH. At pH 7, the removal effi-
Tricalcium phosphate 3.6 5.8 7.2 7.6 7.6
ciency was 87.2%, while at pH 11 the removal efficiency
Octacalcium phosphate 4.3 6.6 7.8 7.4 6.5
was 89.9%. The difference in removal efficiency between
Monotite 0.9 1.1 0.8 0.1 − 0.8
model calculations and experimental results was signifi-
Brushite 0.7 0.8 0.5 − 0.2 − 1.1
cant (P = 0.009). High removal efficiency from the model
Hydroxyapatite 11.9 16.3 19.3 20.8 21.8
calculation was due to the assumption that precipitation
reaction takes place at constant pH (Mekmene et al. 2009),
whereas during the experiment the pH was decreasing,
causing low efficiency. Song et al. (2002) explained that
at constant pH, more base becomes available in the solu-
tion at higher pH than at low pH, leading to the improved
SI. This means at high pH more hydroxyl and phosphate
ions become available in the solution which favours the
precipitation process, hence an increase in SI (Table 4;
Fig. 3). The increase in pH has also attributed to the high
removal efficiency due to the change of phosphate com-
pounds in solution (Fig. 4) which tends to improve the SI.
That is, at high pH the ion concentrations of PO3−
4  ,
CaPO−4
+
and ­CaOH increase in solution, hence promoting the base
uptake reaction. The difference between the model calcula-
tion and the experimental results was observed to decrease
Fig. 3  SI of hydroxyapatite from the model as a function of pH at
operational conditions of initial phosphate concentration of 1  mM, with an increase in pH due to the increase in supersatura-
[Ca]/[P] molar ratio of 1.5 and HLR of 57 m/h tion level. However, the maximum removal efficiency from
the model was only 89.9% at pH 11 because the [Ca]/[P]
molar ratio (1.5) used was below the required ratio (1.67)
Initial pH for complete precipitation.

The experimental results showed an increase in phos-
phate removal efficiency with an increase in initial pH at
operational conditions of [Ca]/[P] molar ratio of 1.5, HLR
of 57 m/h and initial phosphate concentration of 1 mM
(Fig. 2). At pH 7, the removal efficiency was observed
Initial [Ca]/[P] molar ratio
The phosphate removal efficiency increased with an increase
in [Ca]/[P] molar ratio (Fig. 5). At a [Ca]/[P] molar ratio of
1.5, the removal efficiency was 61.9%, while at a [Ca]/[P]

Fig. 4  Phosphate speciation by
visual MINTEQ version 3.0 at
different pH during precipita-
tion process

13
International Journal of Environmental Science and Technology

Fig. 5  Phosphate removal efficiency at different initial [Ca]/[P] molar Fig. 7  Phosphate removal efficiency at different carbonate concentra-
ratio at initial operational conditions of pH 9, initial phosphate con- tions at operational conditions of pH 9, [Ca]/[P] molar ratio of 1.5,
centration of 1 mM and HLR of 57 m/h HLR of 57 m/h and initial phosphate concentration of 1 mM

Table 5  SI of calcium phosphate at different [Ca]/[P] molar ratio with molar ratio of 2.5 the removal efficiency was 79.2%. The
initial phosphate concentration of 1 mM and pH 9 reason for the increase is due to sufficient amount of calcium
Mineral 1.5 2 2.5 ions at higher molar ratio, allowing complete precipitation
reaction since from [Ca]/[P] molar ratio of 1.67 complete
Tricalcium phosphate 5.9 6.2 6.3 reaction occurs when assuming the precipitate forms as
Octacalcium phosphate 6 6.3 6.5 hydroxyapatite (Tran et al. 2014). As a result, SI of calcium
Monotite 0.3 0.4 0.4
phosphate is improved, hence an increase in the efficiency
Brushite 0.06 0.1 0.2
of the precipitation reaction (Table 5). This could also be
Hydroxyapatite 17.3 17.7 18.03
observed from the model calculation, which also showed an
increase in removal efficiency for a [Ca]/[P] molar ratio of
1.5 (89.9%) and 100% for [Ca]/[P] molar ratio of 2 and 2.5
(Fig. 5). From the model calculation, the efficiency reached
100% as the model does not consider the kinetics in the sys-
tem which tends to slow down the reaction process. Moreo-
ver, the higher removal efficiency given by the model was
attributed to the constant pH assumed by the model, whereas
the pH decrease in the experiments caused a low removal
efficiency, as it resulted in a low saturation level (Mekmene
et al. 2009).

Hydraulic loading rate

The experimental data showed a decrease in removal effi-

Fig. 6  Phosphate removal efficiency at different HLRs at initial
operational conditions of pH 9, [Ca]/[P] molar ratio of 1.5 and initial
phosphate concentration of 1 mM
ciency with increased HLR (Fig. 6) at the operational con-
ditions of a [Ca]/[P] molar ratio of 1.5, pH 9 and an initial
phosphate concentration of 1 mM. At the HLR of 44 m/h,
the removal efficiency was 65.9%, 61.9% at the HLR of
57 m/h and 54.2% at the HLR of 70 m/h. This could be
explained by the limited contact time between calcium and

13

phosphate for complete reaction at high HLRs. Since the
model calculation assumed complete reaction between cal-
cium and phosphate in solution, this could be a reason for
higher removal efficiency given by the model compared to
the experimental results (Tran et al. 2014).
Effect of carbonate on phosphate recovery
From the results obtained, the presence of carbonate has
shown to improve the removal efficiency of phosphate (63.2,
64.3 and 66.4%) for carbonate addition of 0.25 mM, 0.5 mM
and 2 mM, respectively, at pH of 9, a phosphate concentra-
tion of 1 mM, a [Ca]/[P] molar ratio of 1.5 and a HLR of
57 m/h (Fig. 7). However, in the absence of carbonate only
61.9% was observed. This is in agreement with the study
by Xuechu et al. (2009) which observed the enhancement
of phosphate removal due to the presence of bicarbonate
ions. Carbonate tends to appear as bicarbonate in a pellet
reactor due to lowering of pH during the precipitation pro-
cess, hence inhibiting the formation of calcium (Tran et al.
2014). The authors went further explaining that the increase
in bicarbonate levels promotes the precipitation of calcium
phosphate due to increased pH. This was not the case at
pH 11 given the same operational conditions and 2 mM
of carbonate concentrations added giving the efficiency of
68.1%, while before carbonate addition the efficiency was
71.6%. This high pH promoted calcium carbonate formation
leaving phosphate ions with few calcium ions to react with;
this decreased the SI of calcium phosphate (Seckler et al.
1990). The presence of high carbonate concentrations in the
solution with pH 9 to 11 could result in co-precipitation of
carbonate with phosphate, leading to the decrease in rela-
tive phosphorus content in the precipitate (Song et al. 2002;
Montastruc et al. 2003; De-Bashan and Bashan 2004). To
overcome this situation, Song et al. (2002) suggested the
increase in [Ca]/[P] molar ratio to less than 5 mM and pH
of less than 10.
Conclusion
Recovering phosphate as calcium phosphate salt in a pellet
reactor prior to treatment in a HSSF-CW system requires
proper operational conditions. It has been observed that
an increase in initial phosphate concentrations, [Ca]/[P]
molar ratio, pH and decrease in HLR improve phosphate
removal efficiency and, thus, its recovery. The presence
of compounds such as carbonates in domestic wastewater
has shown to promote the removal efficiency at pH 9. It
13
International Journal of Environmental Science and Technology
suppressed the removal efficiency at pH 11 due to the forma-
tion of calcium carbonate. Moreover, the visual MINTEQ
version 3.0 should only be used to give an indication of
whether calcium phosphate precipitation from the opera-
tional conditions fed in the model would take place or not
as it only considers the thermodynamics of the system. The
inability of the model to take into account the kinetics con-
cept and the assumption that precipitation reactions take
place at constant pH make it inappropriate for calculating
the precipitation efficiency of phosphate in a pellet reactor
as it gives higher efficiencies compared to the experimental
results. The demonstrated ability of a pellet reactor used
as a pre-treatment system to recover phosphate makes it a
sustainable solution for phosphate scarcity; however, further
research is required to check the suitability of pellet reactor
effluent containing low P for further treatment in a HSSF-
CW system. Since phosphate found in domestic wastewa-
ter is a renewable resource, it could become an additional
source to the existing non-renewable phosphate rock which
is depleting. Other environmental benefits which could result
from phosphate recovery include the improvement in the
aquatic environment due to the reduction in eutrophication.
Acknowledgement  This work was supported by the Flemish Inter-uni-
versity Council for University Development Cooperation (VLIR-UOS)
through an Institutional University Cooperation (IUC) programme with
the Nelson Mandela African Institution of Science and Technology
(NM-AIST) with grant number ZIUS2016AP029.
Compliance with ethical standards 
Conflict of interest  The authors declare that they have no conflict of
interest.
References
Aldaco R, Garea A, Irabien A (2007) Calcium fluoride recovery
from fluoride wastewater in a fluidized bed reactor. Water Res
41:810–818
Baronne JP, Nancollas GH (1977) The seeded growth of calcium phos-
phates: the effect of solid/solution ratio in controlling the nature of
growth phase. J Colloid Interface Sci 62(3):421–431
Boskey AL, Posner AS (1973) Conversion of amorphous calcium
phosphate to microcrystalline hydroxyapatite. A pH-depend-
ent, solution-mediated, solid–solid conversion. J Phys Chem
77(7):2313–2317
Brix H, Arias C, Del Bubba M (2001) Media selection for sustainable
phosphorus removal in subsurface flow constructed wetlands.
Water Sci Technol 44(11–12):47–54
Crutchik D, Morales N, Vazquez-Padin JR, Garrido JM (2017)
Enhancement of struvite pellets crystallization in a full-scale plant
using an industrial grade magnesium product. Water Sci Technol
75(3):609–618. https​://doi.org/10.2166/wst.2016.527
International Journal of Environmental Science and Technology
De-Bashan LE, Bashan Y (2004) Recent advances in removing phos-
phorus from wastewater and its future use as fertilizer (1997–
2003). Water Res 38:4222–4246. https​://doi.org/10.1016/j.watre​
s.2004.07.014
DHV Consulting Engineers (1991) Removing phosphate from munici-
pal wastewater at Westerbork, The Netherlands. DHV Consulting
Engineers, Amersfoort
Drizo A, Frost CA, Grace J, Smith KA (1999) Physico-chemical
screening of phosphate-removing substrates for use in constructed
wetland systems. Water Res 33(17):3595–3602
Drizo A, Forget C, Chapuis R, Comeau Y (2006) Phosphorus removal
by electric arc furnace steel slag and serpentinite. Water Res
40:1547–1554. https​://doi.org/10.1016/j.watre​s.2006.02.001
Giesen A (1999) Crystallisation process enables environmental friendly
phosphate removal at low costs. Environ Technol 20:769–775
Guillard D, Lewis AE (2001) Nickel carbonate precipitation in a fluid-
ized-bed reactor. Ind Eng Chem Res 40(23):5564–5569
Kihila J, Mtei KM, Njau KN (2014) Wastewater treatment for reuse
in urban agriculture; the case of Moshi. Phys Chem Earth
72–75:104–110. https​://doi.org/10.1016/j.pce.2014.10.004
Li YM, Liu MY, Yuan ZW, Zou JT (2013) Struvite pellet crystalliza-
tion in a high-strength nitrogen and phosphorus stream. Water Sci
Technol 68(6):1300–1305
Mahvi AH, Shafiee F, Naddfi K (2005) Feasibility study of crystalliza-
tion process for water softening in a pellet reactor. Int J Environ
Sci Technol 1(4):301–304
Mekmene O, Quillard S, Rouillon T, Bouler JM, Piot M, Gaucheron
F (2009) Effects of pH and Ca/P molar ratio on the quantity and
crystalline structure of calcium phosphates obtained from aqueous
solutions. Dairy Sci Technol 89:301–316. https:​ //doi.org/10.1051/
dst/20090​19
Montastruc L, Azzaro-Pantel C, Biscans B, Cabassud M, Domenech S,
Dibouleau L (2003) A general framework for pellet reactor model-
ling: application to P-recovery. Trans IChemE 81(9):1271–1278
Morse GK, Brett SW, Guy JA, Lester JN (1998) Review: phospho-
rus removal and recovery technologies. Sci Total Environ
212(1):69–81
Njau KN, Mwegoha WJS, Kimwaga RJ, Katima JHY (2011) Use of
engineered wetlands for onsite treatment of wastewater by the
local communities: experiences from Tanzania. Water Pract Tech-
nol. https​://doi.org/10.2166/wpt.2011.047
Seckler MM, Bransma OSC, Rosenden GM, van Ogh, JC, Delgage,
J and Eggers E (1990) Phosphate removal by means of a full
scale pellet reactor. In: 11th International Symposium on Indus-
trial Crystallization, 18–20 September, Garmisch-Partenkirchen,
Germany
Seckler MM, Bruinsma OSL, Van Rosmalen GM (1996) Calcium
phosphate precipitation in a fluidized bed in relation to process
conditions: a black box approach. Water Res 30(7):1677–1685
Seo D, Cho J, Lee H, Heo J (2005) Phosphorus retention capacity of
filter media for estimating the longevity of constructed wetland.
Water Res 39(11):2445–2457. https​://doi.org/10.1016/j.watre​
s.2005.04.032
Song Y, Hahn HH, Hoffmann E (2002) The effect of carbonate on the
precipitation of calcium phosphate. Environ Technol 23:207–215
Song Y, Weidler PG, Berg U, Nuesch R, Donnert D (2006) Calcite-
seeded crystallization of calcium phosphate for phosphorus recov-
ery. Chemosphere 63:236–243
Song Y, Donnert D, Berg U, Weidler PG, Nueesch R (2007) Seed
selections for crystallization of calcium phosphate for phosphorus
recovery. J Environ Sci 19:591–595
Steen I (1998) Phosphorus availability in the 21st century: management
of a non-renewable resource. Phosphorus Potassium 217:25–31
Tran ATK, Zhang Y, De Corte D, Hannes J, Ye W, Mondal P, Jullock
N, Meesschaert B, Pinoy L, Van der Bruggen B (2014) P-recovery
as calcium phosphate from wastewater using an integrated selec-
trodialysis/crystallization process. J Clean Prod 77:140–151. https​
://doi.org/10.1016/j.jclep​ro.2014.01.069
13

Van Kemenade MJJM, de Bruyn PL (1987) A kinetic study of precipi-
tation from supersaturated calcium phosphate solutions. J Colloid
Interface Sci 118(2):564–585
Xuechu C, Hainan K, Deyi W, Xinze W, Yongyong L (2009) Phosphate
removal and recovery through crystallization of hydroxyapatite
using xonotlite as seed crystal. J Environ Sci 21:575–580. https​://
doi.org/10.1016/S1001​-0742(08)62310​-4
13
International Journal of Environmental Science and Technology
Zhou P, Huang J, Li AWF, Wei S (1999) Heavy metal removal from
wastewater in fluidized bed reactor. Water Res 33(8):1918–1924
Zhu T, Jenssen PD, Maehlum T, Krogstad D (1997) Phosphorus
sorption and chemical characteristics of lightweight aggregates
(LWA)—potential filter media in treatment wetlands. Water Sci
Technol 35(5):103–108