Chemosphere 231 (2019) 528e537

Contents lists available at ScienceDirect

Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere

Fluoride removal from water by composite Al/Fe/Si/Mg pre-

polymerized coagulants: Characterization and application
Athanasia K. Tolkou a, Manassis Mitrakas b, Ioannis A. Katsoyiannis a, Mathias Ernst c,
Anastasios I. Zouboulis a, *
a
Laboratory of Chemical and Environmental Technology, Department of Chemistry, Aristotle University of Thessaloniki, 54124, Greece
b
Laboratory of Analytical Chemistry, Department of Chemical Engineering, Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece
c
Department of Water Supply, Technical University of Hamburg, 21073, Hamburg, Germany

h i g h l i g h t s

A Mg-containing composite (PSiFAC-Mg) coagulant was produced and characterized.

PSiFAC-Mg was more efficient than conventional Al-base coagulants for F-removal.

PSiFAC-Mg was efficient for F-removal in a wider pH range (6-8).

Mg content enhanced fluoride removal at pH values higher than 7.5

Residual aluminum concentration was under the respective legislation limit.

a r t i c l e i n f o a b s t r a c t

Article history: Fluoride, an anionic pollutant, is possibly to be found in excessive concentrations especially in
Received 9 April 2019 groundwaters and can show detrimental effects on human health, in concentrations higher than the
Received in revised form commonly applied legislation limit of 1.5 mg/L The most commonly applied method for water de-
19 May 2019
fluoridation is performed by Al-based coagulants, which however presents some important limita-
Accepted 21 May 2019
Available online 22 May 2019
tions, such as the applied relatively high dosage, producing rather excessive amounts of chemical sludge.
In this study, the use of novel pre-polymerized Al-based coagulants was examined, regarding their ef-
Handling Editor: Y Yeomin Yoon ficiency towards fluoride removal, as compared with the conventionally applied AlCl3. The novel co-
agulants were characterized by measuring the main physico-chemical properties, the aluminum species
Keywords: distribution, the zeta potential, the particles' size distribution and the produced flocs’ sizes. The results
Water treatment showed that the Mg-containing coagulant (PSiFAC-Mg30-10-15) was the most efficient, when applied in pH
Composite coagulants values relevant to fluoride-containing groundwaters; it was also the only coagulant, which increases its
Poly-aluminum-ferric-silicate-magnesium efficiency at pH values > 7. The uptake capacity of coagulants, regarding fluoride, to reach the residual/
coagulant (PSiFAC-Mg)
equilibrium concentration limit of 1.5 mg F/L (Q1.5-value) at the pH value 7.0 ± 0.1 were found 170, 134
Fluoride removal
and 94 mg F/g Al for the cases of PSiFAC-Mg30-10-15, AlCl3$6H2O and PSiFAC-Na1.5-10-15, respectively.
Accordingly, at the pH value 7.8 ± 0.2 the Q1.5-values were found 189, 118 and 41 mg F/g Al for the same
coagulants; whereas considering the residual aluminum concentration this was ranged at 15 ± 5, 25 ± 5
and 30 ± 5 mg Al/L, respectively. In addition, (beneficial) increase of residual magnesium concentration,
when applying the coagulant PSiFAC-Mg30-10-15 was 15 ± 5 mg/L.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction

Fluoride pollution of groundwaters is affecting more than 200

* Corresponding author. Aristotle University of Thessaloniki, Department of
Chemistry, Laboratory of Chemical and Environmental Technology, 54124, Thessa-
loniki, Greece.
E-mail addresses: tolkatha@chem.auth.gr (A.K. Tolkou), manasis@eng.auth.gr
(M. Mitrakas), katsogia@chem.auth.gr (I.A. Katsoyiannis), mathias.ernst@tuhh.de
(M. Ernst), zoubouli@chem.auth.gr (A.I. Zouboulis).
million people globally (WHO, 2011). Excessive fluoride concen-
trations found in groundwaters can be attribute to both natural and
anthropogenic activities. The major anthropogenic sources are
considered the excessive use of phosphate fertilizers, as well as the

https://doi.org/10.1016/j.chemosphere.2019.05.183
0045-6535/© 2019 Elsevier Ltd. All rights reserved.
 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
aluminum smelting and the irrigation by fluoride-contaminated
water (Pettenati et al., 2013). It is estimated that up to 0.34 mg/L
of fluoride can reach the groundwater sources by the use of
phosphate-containing fertilizers (Rao et al., 1997). The natural ac-
tivities which can enhance the fluoride presence in groundwater
aquifers are favored mainly by the high alkalinity and bicarbonate
content (Ali et al., 2016). Fluoride concentrations higher than 2 mg/
L can cause several health problems to humans, such as dental and
skeleton fluorosis, masculine infertility and even cancer (Ozsvath,
2009), while at concentrations up to 1.5 mg/L it is considered as
beneficial for human health. Therefore, the removal of fluoride from
waters to achieve final concentration between 0.5 and 1.5 mg/L is
essential (Fawell et al., 2006). Pollution by fluoride is a serious
problem in south-east Africa, as well as in the United States, Middle
East, South America, and Asian countries, with China, India,
Pakistan, Ethiopia and Kenya being the most affected countries (Ali
et al., 2016; Rasool et al., 2018). This problem is affecting mainly the
people in developing countries, in which drinking water is obtained
from groundwater wells, usually untreated; the effective removal of
fluoride in these countries is considered as a challenging issue,
mainly because of lack of necessary infrastructure and expertise.
Therefore, the removal of fluoride from drinking water sources
(de-fluoridation) has received particular attention, especially dur-
ing the past few years. Different treatment technologies have been
developed, such as ion-exchange (Apambire et al., 1997; Meenakshi
and Viswanathan, 2007), precipitation with iron salts (Biswas et al.,
2010), activated alumina (Gong et al., 2012a, b), alum sludge
(Sujana et al., 1998; Dassanayake et al., 2015), use of calcium agents
(Waghmare and Arfin, 2015). Especially for the developing coun-
tries the Nalgonda process is the most widely applied technology,
due to its simplicity and use of relatively low cost materials. This
technology is based on the combined use of alum and lime in a two-
step process, involving the direct addition of lime (or calcium ox-
ide) and alum (as hydrated aluminum salts), used as coagulation
agents (Nawlakhe and Paramasivam, 1993).
However, most of the aforementioned techniques exhibit spe-
cific disadvantages; for example, the adsorption process, when
using activated alumina, is highly pH-sensitive and most efficient at
pH values between 5 and 6.5, therefore not corresponding to the
common pH values of most groundwaters, especially those con-
taining fluoride, being usually higher than 7.5. As a result it presents
relatively lower adsorption capacity at pH > 7 and requires the
appropriate pre-treatment for lowering the pH value of input wa-
ter; additionally, the adsorption capacity is reduced after each
regeneration stage (Tomar and Kumar, 2013; Habuda-Stanic et al.,
2014). In addition, the use of bone char as an adsorbent may not
be acceptable is some countries, due to religious or cultural aspects
and the water in some cases, might also have unpleasant taste and
odor problems. Finally, this technology exhibits relatively low
adsorption capacity, which imply the need for frequent replace-
ment of adsorbent agent, resulting in higher operational costs (Zhu
et al., 2010; Habuda-Stani c M. et al., 2014).
In the case of ion-exchange, fluoride can be removed from water
with the use of a strongly basic anion exchange resin, containing
quaternary ammonium functional groups, but certain limitations,
such as the possible deterioration of water quality, the cost of resin,
its regeneration and secondary waste disposal can prevent this
process from being economical. In recent years, membrane pro-
cesses have emerged as a preferred alternative for drinking water
treatment; however, certain disadvantages, such as the high cost of
529
membranes, the disposal of concentrate and the eventual post-
treatment of water tend to make the process rather uneconomi-
cal (Battula and Cheukuri, 2014).
In the coagulationeprecipitation method, lime and alum are the
most commonly used coagulants; the addition of lime leads to the
precipitation of fluoride as insoluble solid and raises the pH value of
water to 11e12. The respective limitations are related to the rather
higher maintenance cost, as well as to larger space (treatment
tanks) requirements and higher residual aluminum concentrations
in the treated water (Ayoob et al., 2008; Tomar and Kumar, 2013).
Among the aforementioned technologies, conventional coagu-
lation is widely applied, due to its simplicity and high efficiency,
using mostly aluminum-based coagulants, such as aluminum
chloride (AlCl3) or poly-aluminum chloride (PACl) (Malhbtra et al.,
1997; Pinon-Miramontes et al., 2003; He et al., 2015). However, in
most relevant studies, the efficient removal of fluoride requires
quite high coagulant doses, which in some cases may reach up to
300 mg/L and consequently, higher residual aluminum concentra-
tion can remain in the treated (drinkable) water (Reardon and
Wang, 2000; Aoudj et al., 2012). Therefore, the development of
more efficient coagulants, which could remove fluoride with the
application of lower dosages, being at the same time more efficient
at a broader pH range, is considered as an important challenge.
In previous studies, the use of pre-polymerized coagulants,
based on iron or aluminum, have been applied and found quite
efficient for water and wastewater treatment, regarding for
example, the removal of turbidity (Tolkou et al., 2014a,b; 2015a,b),
arsenic (Katsoyiannis et al., 2017), and the reduction of membrane
fouling in MBR systems (Gkotsis et al., 2017), leaving very low re-
sidual metal concentrations, such as in the case of PSiFAC use, being
an aluminum-based pre-polymerized coagulant. Therefore, in the
present study, a tailored (for fluoride removal) pre-polymerized
coagulant was developed and compared with the conventionally
applied aluminum coagulant, as well as with the already tested in
previous studies pre-polymerized coagulant, such as PSiFAC-Na.
The specific coagulant developed in the present study (PSiFAC-
Mg) was a magnesium-modified PSiFAC, aiming at enhancing the
removal of fluoride by the creation of insoluble MgF2 complexes
(Martin, 1990). Mg is classified as a hard base, while F is consid-
ered as a hard acid; thus, it was considered that this material might
exhibit increased affinity against fluoride, due to the interaction
between the electron pair donors and acceptors; noting that the
most favorable interactions may occur, when the acid and base
present similar electronic character (Larson and McMahon, 1985;
Krahlab and Kemnitz, 2017).
Relevant studies investigating the efficiency of adsorbents,
containing different forms of MgO, have been previously reported
(Sundaram et al., 2009; Maliyekkal et al., 2010) and the high de-
fluoridation efficiency of MgO was noticed, which was ascribed to
its high value of Iso-Electric Point (IEP> 10), thus presenting posi-
tive surface charges at the desired water pH range (7-8) and fa-
voring the uptake of fluoride anions (F) (Oladoja et al., 2015).
Therefore, the objectives of this study were: (a) the character-
ization of novel aluminum-based coagulants (PSiFAC-Na and
PSiFAC-Mg), (b) the application of these reagents for the removal of
fluoride, investigating their efficiency in varying experimental
conditions, i.e. applying different fluoride or coagulant concentra-
tions, different (initial) pH values of water and examining the water
matrix to be treated, and (c) the determination of necessary co-
agulants’ removal capacity, able to achieve the residual legislative
530 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
concentration of 1.5 mg F/L (denoted as Q1.5-value). To the best of
our knowledge, the use of composite pre-polymerized coagulants
of aluminum was not systematically studied for the case of fluoride
removal, whereas the application of custom made PSiFAC-Mg re-
agent is reported for the first time.
2. Materials and methods
2.1. Materials
2.1.1. Chemicals
The initial solutions used for the preparation of pre-polymerized
coagulants were 0.5 M AlCl3$6H2O (Merck), 0.5 М FeCl3$6H2O
(Merck), and 0.5 M NaOH (Merck), or slurry of 0.5 M MgO (PMS2
pure magnesia) - as the added alkaline agents. The 0.5 M polysilicic
acid solution (pSi) was prepared, according to the literature
(Tzoupanos and Zouboulis, 2009; Zouboulis and Tzoupanos, 2009).
Sodium fluoride (NaF) solution (Merck) was used for the prepara-
tion of simulated fluoride-polluted water. А 1000 mg/L fluoride
stock solution was prepared by dissolving 2.210 g NaF in 1000 mL of
deionized water. Fluoride-containing working solutions were pre-
pared by the proper dissolution of stock solution to the tap water of
Thessaloniki (Greece). The pH was properly adjusted by using HCl
or NaOH solutions of appropriate concentrations in the range of
1e0.01 M.
2.1.2. Tested water
The main physicochemical characteristics of Thessaloniki's tap
water, partly (around 50%) originating from groundwater sources
and partly from surface (river) water, are shown in Table S1 (sup-
porting information). All simulated fluoride-polluted water sam-
ples were prepared at least 12 h before the experiments and the
main parameters of them are given in Table 1.
2.1.3. Preparation of coagulants
Two composite inorganic pre-polymerized coagulants, i.e. PSi-
FAC-Na1.5-10-15 (polyaluminum ferric silicate chloride) previously
synthesized, studied and applied by Tolkou et al. (2014a, b; 2015a,
b) and PSiFAC-Mg30-10-15 (polyaluminum ferric silicate magne-
sium), firstly prepared and applied in this study, were produced at
room temperature (22  C), applying specific experimental condi-
tions, as presented in Table 2. To compare better the results on de-
fluorization treatment, commercially available AlCl3$6H2O coagu-
lant was also examined.
2.2. Methods
2.2.1. Coagulants’ characterization
2.2.1.1. Physico-chemical properties. All prepared composite (inor-
ganic) coagulants were characterized mainly by measuring the pH
value of final solution (using a Metrohm Herisau pH-Meter), the
turbidity (using a Hach ratio/XR Turbidity-meter) and the con-
ductivity (using a Crison CM 35 conductivity-meter); their stability
(ageing) was also frequently checked.
Table 1
Main physicochemical parameters of simulated fluoride-polluted waters. 5.
Water samples Water type
Tap water 1 L tap water*
1: 1 water 0.5 L tap water and 0.5 L distilled water
1: 5 water 0.2 L tap water and 0.8 L distilled water
1: 10 water 0.1 L tap water and 0.9 L distilled water
2.2.1.2. Al species distribution - Ferron method. The distribution of
aluminum species was determined with the application of Al-
Ferron timed spectrophotometric method, which is based on the
different reaction times of aluminum species with the Ferron re-
agent (8-hydroxy-7- iodoquinoline-5-sulphonic acid) to form water
soluble complexes. These complexes absorb visible light with the
maximum being at 370 nm, hence the absorbance measurements at
this wavelength would allow the calculation of different species of
aluminum (i.e. Ala - monomers, Alb - polymers and Alc - Al(OH)3 -
solids). A Hitachi UV/vis spectrophotometer was used for this
purpose. The exact procedure was a modification of the relevant
method developed by Zhou et al. (2006).
2.2.2. Zeta-potential and size measurement
The zeta-potential and particle size distribution measurements
were both conducted with the ZETASIZER Nano-ZS (Malvern). For
each (adjusted) pH value, three measurements of zeta-potential
and the respective particle sizes were calculated. The tempera-
ture was set at 20  C. The particle sizes’ distribution, produced by
each coagulant, was calculated by assuming that their refractive
index is equal to the refractive index of aluminum silicate. The pH
adjustments were conducted by the MPT-2 MULTI PURPOSE
TITRATOR (Malvern).
2.2.3. Particle size distribution by the volume of flocs
The EyeTech Particle Size and Shape Analyser (Ankersmid) were
used for the flocs’ size measurements. The results show the particle
volume distribution and the cumulative volume of each sample.
The volume distribution shows the particles in a given size range,
expressed as percentage of total sample volume, while the cumu-
lative volume curve tracks the total volume of all size ranges, as
they approach 100%. The flocs were created by vigorous stirring of
solution at 160 rpm for 2 min and then by stirring at 40 rpm for
10 min (Tolkou et al., 2015a). The concentration of coagulant added
to 1 L of water was 10 mg/L and the pH value was accordingly
adjusted, when investigating the influence of (initial) pH.
2.2.4. Fluoride determination
The concentration of fluoride was determined by ion chroma-
tography (Metrohm IC Anion Column Metrosep Anion Dual 2),
using as eluents 1.3 mmol/L of sodium carbonate (Na2CO3) and
2.0 mmol/L of sodium bicarbonate (NaHCO3) and applying a flow of
0.8 mL/min. The retention time/min (unique for each ion) was used
for the respective measurements, noting that for the case of F is
3.53e4.02 min. The method detection limit (MDL) was 0.1 mg/L.
2.2.5. Residual aluminum concentration
The residual aluminum concentration was determined by the
Eriochrome cyanine R standard method (Clesceri et al., 1989).
Aluminum ions react with Eriochrome cyanine R dye at pH 6,
resulting to a colored compound with maximum absorbance at
535 nm. The methods detection limit (MDL) was 6 mg/L.
2.2.6. Residual magnesium concentration
The residual magnesium concentration was determined by
Initial pH pH after addition Conductivity Total hardness DOC (mg/L)
of 20 mg F/L (mS/cm)
7.5 7.31 692 280 mg CaCO3/L 0.2
7.2 7.28 362
7.2 7.34 174
7.03 7.26 84.3
 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537 531

Table 2
Preparation procedures of inorganic pre-polymerized coagulants used in this study.

Coagulant Molar Description Procedure

ratios

PSiFAC- [OH]/[Al]: [AlCl3 þ FeCl3] þ pSi þ NaOH Appropriate amount of FeCl3 solution was added in the AlCl3 solution, under vigorous stirring. Then, the pSi
Na1.5-10- 1.5 solution was added to the resulted Fe/Al solution at predetermined ratio and the alkaline solution (NaOH) was
15 [Al]/[Fe]: added slowly in the mixture reaching the desired pH value (3e3.5).
10
[Al þ Fe]/
[Si]: 15
PSiFAC- [Al]/[OH]: [AlCl3 þ FeCl3] þ pSi þ MgO Appropriate amount of FeCl3 solution was added in the AlCl3 solution, under vigorous stirring. Then, the pSi
Mg30-10- 30 solution was added to the resulted Fe/Al solution at predetermined ratio and the MgO slurry was added slowly in
15 [Al]/[Fe]: the mixture until reaching the desired pH value (2e2.5)
10
[Al þ Fe]/
[Si]: 15

flame Atomic Absorption Spectrophotometry, using a PerkinElmer silicate.

AAnalyst 800 instrument.
2.3. Coagulation performance (Jar-tests)
Jar tests were used for the examination of coagulants’ efficiency.
A jar-test apparatus (Aqualytic), equipped with six paddles and
employing 1 L glass beakers, was used for all coagulation experi-
ments. A total volume of 1000 mL of simulated fluoride-polluted
water was transferred to a 1 L beaker and pH was adjusted to the
desired value (between 6 and 8) and kept constant during the
coagulation tests. The solutions were stirred rapidly at 160 rpm for
2 min, followed by slow stirring at 40 rpm for 10 min and then, the
suspension was rested for 30 min without any stirring to allow the
sedimentation of produced flocs (Tolkou et al., 2015a). After sedi-
mentation, water samples were collected from the supernatant of
each beaker, approximately at 5 cm below the water surface,
filtered through 0.45 mm pore size nylon filter and the concentra-
tions of fluoride and other relevant parameters were determined in
the filtrate. The examined coagulants doses, expressed as mg Al/L,
were in the range of 5e50 mg Al/L.
3. Results and discussion
3.1. Characterization of coagulants
3.1.1. Physico-chemical properties
Table 3 presents the main properties of examined coagulants in
this study. It is shown that the applied preparation/synthesis
method is affecting the percentage of medium-sized forms of
polymerized aluminum (denoted as Alb). More specifically in the
case of PSiFAC-Na1.5-10-15, where NaOH solution is used as the
alkaline agent for the polymerization of coagulant, a relatively
higher content was found (i.e. Alb 51%), in contrast to the case of
PSiFAC-Mg30-10-15 (where Alb was 12%), using MgO as alternative
alkaline agent. Turbidity was increased, when NaOH is used, due to
the formation of several precipitated solids of aluminum, iron and
Table 3
The main physico-chemical properties of examined coagulant agents and the rele-
vant Al-species distribution.
Type pH Turbidity (NTU) Conductivity (mS/cm) Al-species
distribution
Ala
PSiFAC-Na1.5-10-15 3.5 211 23.7 44 51
PSiFAC-Mg30-10-15 2.3 98 24.5 81 12
AlCl3$6H2O 2 <1 21.0 100 e
3.1.2. Zeta-potential and size measurements
In Table 4, the values of the initial pH of samples, of zeta-
potential, of particle sizes and of Iso-Electric Points (IEPs) are dis-
played, for all the examined coagulants, by applying the same
coagulant dosage (10 mg Al/L) in deionized water. Fig. 1 shows the
relevant diagrams of zeta-potential and particle size measurements
of these coagulants, as well as a comparison diagram between
them. It is depicted that the rise of pH value has resulted in the
increase of particle sizes and in the reduction of zeta-potential
measurements of each sample. As the pH increases, the positive
charge of the particles decreases and tends to be neutralized at the
respective IEP and then, an overall negative charge is prevailing. As
a result, the (colloidal) system is being destabilized and bigger
particles are formed.
PSiFAC-Na1.5-10-15 is the aluminum-based coagulant that pre-
sented the highest initial positive value of zeta-potential (50.5 mV).
The initial pH value (after coagulant addition) and the respective
IEP for this coagulant were 4.9 and 9.10, respectively. The initial
particle size was approximately 189 nm, classifying it as a nanoscale
coagulant. PSiFAC-Mg30-10-15 presents also a high initial zeta-
potential value (38.2 mV). Its initial particle size was 449 nm and
increases substantially (up to 4000 nm) as the pH rises. The com-
mon commercially available aluminum-based coagulant, i.e.
AlCl3.6H2O, presents an initial zeta-potential value of 39.7 mV and a
particle size value of 234 nm. The results depict that all these
chemical agents can be classified as nanoscale coagulants.
Fig. 1 shows that for all these coagulants, the IEP values were
found to be greater than 8, which means that at lower pH values an
overall positive charge prevails, favoring the interaction of
aluminum species with the negatively charged fluoride species.
Also, the size of the particles in all these cases was increased
drastically at pH values around 7.0, which is most likely attributed
to the formation of Al(OH)3 precipitates. These results, regarding
the characterization of novel coagulants as compared with the
Table 4
Initial pH (after coagulant addition), zeta-potential, sizes and Iso-Electric Points (IEP)
values for the examined coagulants, tested at 10 mg Al/L concentration in distilled
water.
Coagulant Initial Initial zeta-potential Initial size Iso-Electric Point
pH (mV) (nm) (IEP)
PSiFAC-Na15- 4.87 50.5 189 9.10
Alb Alc
10-15
5 PSiFAC-Mg30- 4.61 38.3 449 8.3
7 10-15
e AlCl3$6H2O 5.07 39.7 234 9.02
532 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
Fig. 1. Zeta-potential (mV) and particle size measurements (nm) as a function of respective pH values; (a) PSiFAC-Na1.5-10-15, (b) PSiFAC-Mg30-10-15, (c) AlCl3$6H2O, and (d) com-
parison of zeta-potential values for the examined coagulants.
conventional one, can confirm their effectiveness as effective co-
agulants in the pH range, which is relevant to drinking water
treatment (i.e. 7-8).
3.1.3. Particle size distribution of flocs
In Table 5 the mean diameter values of the flocs produced by the
examined coagulants are presented, falling in the range 1e6 mm,
with the AlCl3 presenting the larger flocs under the studied con-
ditions. Additionally, Table 6 and Fig. 2 shows the effect of fluoride
presence on the size of the flocs for the case of PSiFAC-Mg30-10-15
addition; it can be noticed that the presence of fluoride contributed
to the formation of slightly larger flocs. These changes in the floc
size and formation were also found in a relevant study, when the
effect of fluoride on floc formation by coagulation using aluminum
chloride was examined (Gong et al., 2012b).
In particular, by using FTIR and XPS spectrometry, it was also
shown in this paper that the flocs after complexation/interaction
with fluoride anions were different than in the case of Al(OH)3 flocs
Table 5
Mean flocs’ size range (mm) of applied coagulants in tap water for the same coag-
ulant dosage (10 mg Al/L).
Coagulant Flocs' size range (mm)
PSiFAC-Na15-10-15 1.5e3
PSiFAC-Mg30-10-15 2e4
AlCl3$6H2O 4e6
Table 6
Mean flocs size value (mm) of PSiFAC-Mg30-10-15 coagulant as affected by the
fluoride presence in water; coagulant dosage 10 mg Al/L, pH 7.0 ± 0.1,
T ¼ 22±1  C.
Presence of F (mg/L) Mean flocs' size value (mm)
0 2.3
5 4.9
10 5
and that the size of flocs differed in the absence and presence of
fluoride. However, in their study, the floc size was decreased as the
ratio of fluoride to aluminum increased, which is in contrast to the
findings of this study, examining the use of pre-polymerized co-
agulants by using magnesium oxide as the alkali regulator. This
change might be attributed to the presence of magnesium, in
contrast to the pure aluminum chloride, which creates different
kind of flocs and involves also other mechanisms for the removal
for fluoride; i.e. in the case of magnesium-modified coagulant the
bridge flocculation mechanism seems to be more probable, while in
the presence of aluminum chloride the respective mechanism
seems to be is rather the sweep flocculation (Dubey et al., 2018).
Therefore, this might be an advantage of the selected coagulant, as
the larger the flocs are, the better the settling ability and thus, the
more efficient overall removal of fluoride from water.
 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
Fig. 2. Effect of fluoride presence in the distribution of particle sizes, regarding the addition of PSiFAC-Mg30-10-15 in tap water; (a) on the diameters of flocs sizes (mm), and (b) on the
cumulative size distribution as percentage of the particles sizes.
3.2. Effect of pH on fluoride removal
The coagulation is a physico-chemical process, which is highly
dependent on the solution pH (Aoudj et al., 2012; He et al., 2015),
especially in the case of aluminum-based coagulants, because the
range of pH in which aluminum exhibits its lower solubility is quite
narrow. The effect of pH on fluoride removal by the presence of
aluminum coagulants was studied over the pH range of 6e8 with
constant coagulant dose (30 mg Al/L, as found during preliminary
experiments) and initial fluoride concentration of 5 mg/L (Fig. 3). It
is worth noting that the pH was controlled and kept constant
during the coagulation tests. As it can be noticed, the maximum
fluoride removal was observed at pH around 7, when PSiFAC-Na1.5-
10-15 and conventional AlCl3$6H2O were used, while at the acidic
and especially at the alkaline pH values, the fluoride removal was
decreased, in contrast with the case of Mg-containing coagulant.
Fig. 3. Effect of pH values on fluoride removal, for initial fluoride concentration 5 mg/L
and coagulant dosage 30 mg Al/L (T ¼ 22±1  C) for the three examined coagulant
agents.
533
This is also consistent with the solubility of aluminum, which is
lowest at narrow pH value 6.8 ± 0.2. The residual fluoride concen-
trations at pH 6, 7 and 8 were 2.2, 1.3 and 1.6 mg/L for the case of
AlCl3$6H2O and 2.4, 1.7 and 2.6 mg/L for the case of PSiFAC-Na1.5-10-
15, respectively. These results indicate that for the cases of afore-
mentioned coagulants the insoluble Al(OH)3 is possibly the main
specie responsible for fluoride uptake. The Al(OH)3 flocs is believed
to adsorb F strongly and the formation of this precipitate is
optimal in the pH range 6.8 ± 0.2 (Aoudj et al., 2012). The dissolved
aluminum species Al3þ, Al(OH)2þ and Al(OH)2þ at pH < 6.8 and
Al(OH)-4 at pH > 6.8 can lead to the diminishing of Al(OH)3 pre-
cipitates, resulting in turn in lower fluoride uptake. These results
were also confirmed by the relevant literature (Hu et al., 2005;
Aoudj et al., 2012).
In contrast, PSiFAC-Mg30-10-15 exhibits a better fluoride removal
by increasing the pH in the range 6e8 (Fig. 3) and the residual
concentrations at pH 6, 7 and 8 were 1.7, 1.1 and 0.7 mg/L,
respectively.
This is most likely attributable to the ability of magnesium to
form the respective magnesium fluoride complexes (see also the
supporting information, Table S2); thus, acting synergistically with
Al species and improving the removal of fluoride. In summarizing,
the responsible mechanism of PSiFACMg30-10-15 for fluoride
removal is similar with the other coagulants (i.e. PSiFAC-Na1.5-10-15
and AlCl3$6H2O) for the pH values up to 7. However, at pH values
higher than 7, the increase of fluoride uptake capacity can be
attributed to the formation of insoluble Magnesium fluoride com-
plexes, which can replace effectively the expected loss of efficiency,
provided by aluminum at higher pH values. Conclusively, the
incorporation of magnesium to form a composite pre-polymerized
inorganic coagulant was found to increase significantly the capacity
for fluoride uptake, as compared to the other aluminum-based
coagulants examined in this research, especially for pH values
higher than 7.2. This finding has significant implications for the
treatment of fluoride-containing groundwaters, since most of them
usually exhibit pH values above 7.5 (Thole, 2013).
3.3. Determination of coagulants uptake capacity for fluoride
The fluoride uptake capacity of coagulants was subsequently
evaluated, considering the residual/equilibrium concentration
equal to drinking water regulation limit of 1.5 mg F/L (denoted as
Q1.5-value) at pH values 7.0 ± 0.1 and 7.8 ± 0.1. Since fluoride uptake
534 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537

Fig. 4. (a) Effect of coagulant dose to fluoride removal, and (b) the respective results fitting to Freundlich (solid line) and Langmuir (dashed line) models; initial fluoride con-
centration 5 mg F/L, pH 7.0 ± 0.1, T ¼ 22±1  C.

is expected to proceed through complexation by aluminum and/or
magnesium, the uptake capacity was determined by applying the
following experimental conditions:
✓ The same initial fluoride concentration (5 mg F/L) and different
coagulants doses/concentrations in the range 2e50 mg Al/L, and
✓ The same coagulants' dose (30 mg Al/L) and initial fluoride
concentrations in the range 2e20 mg F/L.
The results of coagulants’ type and concentration, when using a
fixed initial fluoride concentration (5 mg/L) at pH 7.0 ± 0.1 are
shown in Fig. 4a. All three coagulants showed rather similar
behavior, with the necessary coagulant dose being at least 20 mg/L
for achieving residual concentration of fluoride lower than the
drinking water regulation limit of 1.5 mg/L. These experimental
results were best fitted to empirical Freundlich model (Fig. 4b),
based on sorption onto a heterogeneous surface (Parab et al., 2005;
Katsoyiannis, 2007), as well as to the Langmuir model, which is
based on the assumption that all adsorption sites are of equal size
and shape, resulting in monolayer adsorption (Table 7). According
to the Freundlich equation, the determined Q1.5 values were 167,
135 and 96 mg F/g Al for the cases of PSiFAC-Mg30-10-15, AlCl3$6H2O
and PSiFAC-Na1.5-10-15, respectively. The evaluation of Q1.5 values
clarify that the PSiFAC-Mg30-10-15 presents the best uptake capacity
among them, which is almost 25% and 70% higher, than the
corresponding for the other two coagulants (AlCl3$6H2O and PSi-
FAC-Na1.5-10-15), respectively. Furthermore, it must be clarified that
the Qmax values according to the Langmuir equation are indicative,
since they were calculated by the extrapolation of experimental
data, focusing to low residual concentrations of fluoride in order to
clarify better the technologically important Q1.5-values (applicable
for the design of larger-scale treatment systems), while there are
not available data, regarding the residual concentrations at the
saturation conditions of examined coagulants/adsorbents.
The adsorption isotherms for the initial fluoride concentrations
in the range 2e20 mg F/L, using a fixed coagulant concentration
(30 mg Al/L) are presented in Fig. 5 and the relative constants by
fitting these data, according to the Freundlich equation, are also
incorporated in Table 7. The respective experimental results did not
fit well, according to the Langmuir model (as R2 < 0.9), and thus
they are not presented in Table 7, whereas they were best fitted
according to the Freundlich model. In the latter case the Q1.5 values
were found to be similar (the respective variation was less than 5%)
with the corresponding Q1.5 values, which were previously deter-
mined (i.e. with fixed fluoride concentration and different coagu-
lant doses).
The lower KF-values along with the higher n-values (according
to the Freundlich model) indicate the lower affinity between fluo-
ride and the created surface of precipitates and verify the hetero-
geneous adsorption mechanism. When considering a constant

Table 7
Freundlich and Langmuir constants derived from the application of respective isotherms to fluoride removal; experimental conditions: initial fluoride concentration 5 mg F/L,
pH 7.0 ± 0.1, T ¼ 22±1  C).

Coagulants Freundlich model Langmuir model

qe ¼ KFCne Ce 1 1
¼ þ Ce
Qe KL Qmax Qmax

KF mg/g (L/mg)n n R2 Q1.5 mg F/g Al Qmax mg F/g Al KL R2

L/mg

Coagulants' concentration range 2e50 mg Al/L, fixed initial fluoride concentration (5 mg F/L)
AlCl3$6H2O 107 0.573 0.988 135 721 0.203 0.998
PSiFAC-Na1.5-10-15 70 0.783 0.989 96 413 0.334 0.982
PSiFAC-Mg30-10-15 126 0.686 0.995 167 750 0.096 0.985
Fixed coagulants' dose (30 mg Al/L), fluoride concentration range 2e20 mg F/L
AlCl3$6H2O 87 1.051 0.996 133
PSiFAC-Na1.5-10-15 57 1.158 0.994 92
PSiFAC-Mg30-10-15 123 0.853 0.998 174
 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537 535

Table 8
BET fitting parameters for fluoride adsorption isotherms at 7.8 ± 0.2, according to the
equation: Q ¼ aC/[1 þ bC þ dC2]; experimental conditions: Co ¼ 5 mg F/L, pH
7.8 ± 0.2, T ¼ 22±1  C).

Coagulant a b C R2 Q1.5, (mg F/g Al)

AlCl3$6H2O 103 0.49 0.19 0.9968 118

PSiFAC-Na1.5-10-15 11 0.49 0.06 0.9848 41
PSiFAC-Mg30-10-15 1070 8.9 2.6 0.9891 189

✓ PSiFAC-Na1.5-10-15 (mean Q1.5 ¼ 94 mg F/g Al)

Co ¼ 5 mg F/L: Al dose ¼ [(10e1.5) mg F/L]/[94 mg F/g

Fig. 5. Adsorption isotherms based on various fluoride concentrations (2-20 mg F/L)
and fixed coagulants' dose (30 mg Al/L) at pH 7.0 ± 0.1 (T ¼ 22±1  C).
number of adsorption sites (i.e. addition of a constant coagulant
dose), then by increasing the adsorbate (fluoride) concentration,
this is gradually cover the adsorption sites, which present lower
affinity and possibly led in multilayer adsorption; therefore, results
in lower KF-values, higher n-values and did not fit to Langmuir
model.
Conclusively, PSiFAC-Mg30-10-15 proved as more efficient coag-
ulant for fluoride removal in comparison with the other examined
coagulants, while the respective doses for decreasing the initial
fluoride concentrations of 5 mg F/L at pH 7.0 ± 0.1 down to the
legislation limit (1.5 mg F/L) are:
✓ PSiFAC-Mg30-10-15 (mean Q1.5 ¼ 170 mg F/g Al):
Co ¼ 5 mg F/L: Al dose ¼ [(5e1.5) mg F/L]/[170 mg F/g
Al] ¼ 0.0205 g Al/L ¼ 20.6 mg Al/L.
Al] ¼ 0.0372 g Al/L ¼ 37.2 mg Al/L.
However, the uptake capacity of PSiFAC-Mg30-10-15 at pH
7.8 ± 0.2, commonly encountered at fluoride-rich groundwaters
was improved, while the corresponding uptake capacities of
AlCl3$6H2O and PSiFAC-Na1.5-10-15 were decreased, verifying the
results of Fig. 3. The adsorption isotherms for the same initial
fluoride concentration (5 mg/L) at pH 7.8 ± 0.2 and different co-
agulants’ doses 2-50 mg Al/L are shown in Fig. 6. The respective
adsorption data were found to best fit the BET model, indicating the
lower affinity of fluoride with aluminum at pH 7.8 ± 0.2, due to the
increase of negatively charged surface of aluminum species
[Al(OH)-] and the uptake of fluoride by physisorption. The respec-
tive model parameters, as well as the uptake capacity of coagulants
are presented in Table 8, which indicate that the uptake capacity
(Q1.5) of PSiFAC-Mg30-10-15 increased more than 10%, while the
corresponding of AlCl3$6H2O and PSiFAC-Na1.5-10-15 were decreased
more than 10% and 50%, respectively.
Therefore, the concentrations for decreasing an initial fluoride
concentration of 5 mg F/L to drinking water regulation limits were
modified accordingly in this case as follows:
➢ PSiFAC-Mg30-10-15 (Q1.5¼189 mg F/g Al):

✓ AlCl3$6H2O (mean Q1.5 ¼ 134 mg F/g Al). Co ¼ 5 mg F/L: Al dose ¼ [(5e1.5) mg F/L]/[189 mg F/g
Al] ¼ 0.0185 g Al/L ¼ 18.5 mg Al/L.
Co ¼ 5 mg F/L: Al dose ¼ [(10e1.5) mg F/L]/[134 mg F/g
Al] ¼ 0.0261 g Al/L ¼ 26.1 mg Al/L. ➢ AlCl3$6H2O (Q1.5 ¼ 118 mg F/g Al)

Co ¼ 5 mg F/L: Al dose ¼ [(10e1.5) mg F/L]/[118 mg F/g

Al] ¼ 0.0297 g Al/L ¼ 29.7 mg Al/L.

➢ PSiFAC-Na1.5-10-15 (Q1.5 ¼ 41 mg F/g Al)

Co ¼ 5 mg F/L: Al dose ¼ [(10e1.5) mg F/L]/[41 mg F/g

Al] ¼ 0.0854 g Al/L ¼ 85.4 mg Al/L.
Conclusively, the doses of AlCl3$6H2O and PSiFAC-Na1.5-10-15 for
diminishing an initial fluoride concentration of 5 mg/L to drinking
water regulation limits are 1.6 and 4.6 times higher, than the cor-
responding of PSiFAC-Mg30-10-15. Considering also the residual
aluminum, its concentration was ranged at 15 ± 5 mg Al/L, 25 ± 5 mg
Al/L and 30 ± 5 mg Al/L for the cases of AlCl3$6H2O, PSiFAC-Mg30-10-
15 and PSiFAC-Na1.5-10-15, respectively. Additionally, the contribu-
tion of selected coagulant PSiFAC-Mg30-10-15 for dissolved Mg in the
treated water and in the range of 2e50 mg Al/L as coagulant dos-
ages, was found to be in the range 15 ± 5 mg Mg/L, i.e. slightly
increased, which did not modified significantly the resulting water
quality and it can be considered as an advantage from the human
health point of view.

Fig. 6. Adsorption isotherms of fluoride by the application of three coagulant agents:

3.4. Effect of water matrix on fluoride removal
AlCl3$6H2O, PSiFAC-Na1.5-10-15 and PSiFAC-Mg30-10-15; experimental conditions:
Co ¼ 5 mg F/L, pH 7.8 ± 0.1, T ¼ 22±1  C). Fig. 7 shows the fluoride removal, using several dilutions of tap
536 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
Fig. 7. Effect of water matrix at fixed initial fluoride concentration of 5 mg/L; coagulant
dosage 30 mg Al/L, pH 7.2 ± 0.1, T ¼ 22±1  C.
water by deionized water, namely 1: 1, 1: 5, and 1: 10 (Table 1) in
order to check the effect of water matrix. Fluoride removal by the
coagulants PSiFAC-Na1.5-10-15 and AlCl3$6H2O was not affected by
the dilution with tap water, which means that is not substantially
affected by different water hardness and alkalinity. On the other
hand, the efficiency of PSiFAC-Mg30-10-15 on fluoride uptake was
found to increase by the dilution of tap water with deionized water,
with the improvement being better as the dilution gets from 1: 5 to
1: 10 ratio. This effect is most likely attributed to the magnesium
content of the coagulant. The presence of magnesium in tap water
matrix along with the MgF2 can form various complexes, such as
MgHCOþ þ Mg, already studied for the decrease of fouling in MBR processes
3 , MgCl (see Table S2 in the supplementary information),
which presents strong competition for the same adsorbing sites.
Therefore, by diluting the tap water with distilled water, the con-
centrations of MgHCOþ þ polluted groundwaters. The results showed that the newly pro-
3 , MgCl are decreased, thus increasing the
available sites for fluoride sorption. This effect sheds also light into
the mechanism of fluoride removal by the addition of magnesium,
providing information for the synthesis of more efficient co-
agulants in future research.
3.5. Residual aluminum concentration
The residual aluminum concentration is considered
Fig. 8. Residual aluminum concentration as function of water pH for the same co-
agulants' dose (30 mg Al/L).
particularly important parameter from the perspective of human
health and should be carefully considered, when an aluminum
coagulant is applied for drinking water treatment. According to
Fig. 8, it can be seen that the residual Al concentration remains well
under the respective legislation limit of 200 mg Al/L, according to
the EU Directive 98/83/EC, for all the examined coagulants at the
pH range (6-8) commonly encountered in most drinking water
sources. These results are in agreement with other studies, having
examined aluminum-based coagulants for the removal of fluoride,
in samples measured after a microfiltration process, as it was
conducted in the present study (Dubey et al., 2018). The use of all
coagulants in this study did not enrich the finished water with
unacceptable levels of aluminum and from this point of view can be
considered as safe for applications in drinking water treatment. The
lowest residual Al concentration was achieved at water pH 7, due to
the lowest solubility of aluminum at this pH value.
Summarizing, in the case of composite poly-aluminum-ferric-
silicate-magnesium coagulant (PSiFAC-Mg), no further flocculants
aid or polyelectrolyte are required; as a result, additional cost
benefits may arise by the utilization of this material, including the
avoidance of specific equipment for handling the polyelectrolyte
solution (e.g. dissolution and pumping systems). In this way, the
overall treatment procedure is simplified and the overall cost-
effectiveness is also improved. In addition, the results showed
that the Mg-containing coagulant (PSiFAC-Mg30-10-15) was the
most efficient, when applied in pH values relevant to fluoride-
containing groundwaters, which means that no extra cost and re-
agents are needed to adjust the pH accordingly.
4. Conclusions
In the present study, the novel aluminum-based coagulant
PSiFAC-Mg30-10-15 was compared with PSiFAC-Na1.5-10-15 (pre-
polymerized, also Al-based coagulant, but without the presence of
and for arsenic removal) and with the conventional and commer-
cially available AlCl3 for the removal of fluoride from simulated
duced magnesium-containing coagulant was more efficient, than
the other two (not containing magnesium) and the fluoride uptake
capacity of examined coagulants, followed the order: PSiFAC-Mg30-
10-15 > AlCl3$6H2O > PSiFAC-Na1.5-10-15. The optimum uptake ca-
pacity of AlCl3$6H2O and PSiFAC-Na1.5-10-15 was found at pH 7,
where aluminum presents the lower solubility, while the corre-
sponding of PSiFAC-Mg30-10-15 uptake capacity was more efficient
at pH 7, than the other two. Furthermore, the selected coagulant
as
continued to increase its efficiency at pH values higher than 7,
where the other two coagulants started to become less efficient for
the removal of fluoride at slightly alkaline pH values, commonly
encountered in most groundwaters.
The results have been further elaborated by using the Freundlich
and Langmuir models and the Q1.5-values were determined by the
application of Freundlich model. It was shown that the selected
coagulant PSiFAC-Mg30-10-15 can achieve higher uptake Q1.5 values
and lower residual F concentrations, than the respective regulation
limit at the pH 7.0 ± 0.1, i.e. Q1.5 ¼ 170 mg F/g Al for PSiFAC-Mg30-10-
15, Q1.5 ¼ 134 mg F/g Al for AlCl3$6H2O and Q1.5 ¼ 94 for PSiFAC1.5-10-
15. The uptake capacity of PSiFAC-Mg30-10-15 at pH 7.8 ± 0.2,
commonly encountered at fluoride-rich groundwaters, was further
improved (Q1.5 ¼ 189 mg F/g), while the corresponding uptake ca-
pacities of the other two examined coagulants, i.e. AlCl3$6H2O
(Q1.5 ¼ 118 mg F/g) and PSiFAC-Na1.5-10-15 (Q1.5 ¼ 41 mg F/g) were
decreased at this alkaline pH value. Additionally, for this pH value
(7.8), the adsorption data were best fitted to the BET model, indi-
cating the lower affinity of fluoride with aluminum, due to the
 A.K. Tolkou et al. / Chemosphere 231 (2019) 528e537
increase of respective negative surface charges [Al(OH)-] and the
fluoride uptake was performed mainly by physisorption. The re-
sidual aluminum concentration in the treated waters was found for
all coagulants and for all tested pH values and coagulant concen-
trations below the maximum concentration limit of Al in drinking
water (200 mg/L). Finally, a slight increase of Mg concentration in
the treated water can be considered rather as an additional
advantage of the new coagulant.
Conflicts of interest
The authors declare that they have no conflict of interest.
Acknowledgments
A part of the experimental part was executed under the
framework of the IKYDA 2016 project ACR-Tech, which involved
cooperation between the Aristotle University of Thessaloniki and
the Technical University of Hamburg-Germany
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.chemosphere.2019.05.183.
References
Ali, S., Thakur, S.K., Sarkar, A., Shekhar, S., 2016. Worldwide contamination of water
by fluoride. Environ. Chem. Lett. 14, 291e315.
Aoudj, S., Drouiche, N., Hecini, M., Ouslimane, T., Palaouane, B., 2012. Coagulation as
a post-treatment method for the defluoridation of photovoltaic cell
manufacturing wastewater. Proc. Eng. 33, 111e120.
Apambire, W.B., Boyle, D.R., Michel, F.A., 1997. Geochemistry, genesis, and health
implications of fluoriferous groundwaters in the upper regions of Ghana. En-
viron. Geol. 33 (1), 13e24.
Ayoob, S., Gupta, A.K., Bhat, V.T., 2008. A conceptual overview on sustainable
technologies for defluoridation of drinking water and removal mechanisms.
Crit. Rev. Environ. Sci. Technol. 38, 401e470.
Battula, S.K., Cheukuri, J., 2014. Effective removal of fluoride from ground water
using electro coagulation. Int. J. Eng. Res. Afr. 4 (2), 439e445.
Biswas, K., Gupta, K., Goswami, A., Chand Ghosh, U., 2010. Fluoride removal effi-
ciency from aqueous solution by synthetic iron(III)ealuminum(III)echromiu-
m(III) ternary mixed oxide. Desalination 255 (1e3), 44e51.
Clesceri, L., Greenberg, A., Trussell, R., 1989. Standard Methods for the Examination
of Water and Wastewater, seventeenth ed. APHAeAWWAeWEF, Washington,
DC.
Dassanayake, K.B., Jayasinghe, G.Y., Surapaneni, A., Hetherington, C., 2015. A review
on alum sludge reuse with special reference to agricultural applications and
future challenges. Waste Manag. 38, 321e335.
Dubey, S., Agarwal, M., Gupta, A.B., 2018. Experimental investigation of Al-F species
formation and transformation during coagulation for fluoride removal using
alum and PACl. J. Mol. Liq. 266, 349e360.
Fawell, J., Bailey, K., Chilton, J., Dahi, E., Fewtrell, L., Magara, Y., 2006. Fluoride in
Drinking-Water. IWA Publishing, London.
Gkotsis, P.K., Mitrakas, M.M., Tolkou, A.K., Zouboulis, A.I., 2017. Batch and contin-
uous dosing of conventional and composite coagulation agents for fouling
control in a pilot-scale MBR. Chem. Eng. J. 311, 255e264.
Gong, W.X., Qu, J.H., Liu, R.P., Lan, H.C., 2012a. Adsorption of fluoride onto different
types of aluminas. Chem. Eng. J. 189e190, 126e133.
Gong, W.X., Qu, J.H., Liu, R.P., Lan, H.C., 2012b. Effect of aluminum fluoride
complexation of fluoride removal by coagulation. Colloid. Surf. Physicochem.
Eng. Asp. 39, 88e93.
Habuda-Stani c, M., Ergovic Ravan ci
c, M., Flanagan, A., 2014. A Review on adsorption
of fluoride from aqueous solution. Materials 7, 6317e6366.
He, Z., Liu, R., Hu, Z., Liu, H., Qu, J., 2015. Defluoridation by Al-based coagulation and
adsorption: species transformation of aluminum and fluoride. Separ. Purif.
Technol. 148, 68e75.
Hu, C.Y., Lo, S.L., Kuan, W.H., 2005. Effects of the molar ration of hydroxide and
fluoride to Al(III) on fluoride removal by coagulation and electrocoagulation.
J. Colloid Interface Sci. 283, 472e476.
Katsoyiannis, I., Tzollas, N., Tolkou, A., Mitrakas, M., Ernst, M., Zouboulis, A., 2017.
Use of novel composite coagulants for arsenic removal from waters experi-
mental insight for the application of polyferric sulfate (PFS). Sustainability 9 (4),
590. https://doi.org/10.3390/su9040590.
Katsoyiannis, I.A., 2007. Carbonate effects and pH-dependence of uranium sorption
onto bacteriogenic iron oxides: kinetic and equilibrium studies. J. Hazard Mater.
537
139 (1), 31e37.
Krahlab, Τ., Kemnitz, Е., 2017. Aluminium fluoride e the strongest solid Lewis acid:
structure and reactivity. Catal. Sci. Technol. 7, 773e796.
Larson, J.W., McMahon, T.B., 1985. Fluoride and chloride affinities of main group
oxides, fluorides, oxofluorides, and alkyls. Quantitative scales of Lewis acidities
from ion cyclotron resonance halide-exchange equilibrium. J. Am. Chem. Soc.
107 (4), 766e773.
Malhbtra, S., Kulkarni, D.N., Pande, S.P., 1997. Effectiveness of poly aluminum
chloride (PAC) vis-a-vis alum in the removal of fluorides and heavy metals.
J. Environ. Sci. Health Part A: Environ. Sci. Eng. Toxicol. 32 (9e10), 2563e2574.
Maliyekkal, S.M., Anshup, K.R., Antony, T.P., 2010. High yield combustion synthesis
of nanomagnesia and its application for fluoride removal. Sci. Total Environ.
408, 2273e2282.
Martin, B., 1990. Bioinorganic Chemistry of Magnesium Chapter in Metal Ions in
Biological Systems (Astrid Sigel and Helmut Sigel).
Meenakshi, S., Viswanathan, N., 2007. Identification of selective ion-exchange resin
for fluoride sorption. J. Colloid Interface Sci. 308, 438e450.
Nawlakhe, W.G., Paramasivam, R., 1993. Defluoridation of potable water by Nal-
gonda technique. Curr. Sci. 65 (10), 743e748.
Oladoja, N.A., Chen, S., Drewes, J.E., Helmreich, B., 2015. Characterization of granular
matrix supported nano magnesium oxide as an adsorbent for defluoridation of
groundwater. Chem. Eng. J. 281, 632e643.
Ozsvath, D.L., 2009. Fluoride and environmental health: a review. Rev. Environ. Sci.
Biotechnol. 8, 59e79.
Parab, H., Joschi, S., Shenoy, N., Verma, R., Lali, A., Sudersanan, M., 2005. Uranium
removal from aqueous solution by coir pith: equilibrium and kinetic studies.
Bioresour. Technol. 96, 1241e1248.
Pettenati, M., Perrin, J., Pauwels, H., Ahmed, S., 2013. Simulating fluoride evolution
in groundwater using a reactive multicomponent transient transport model:
application to a crystalline aquifer of Southern India. Appl. Geochem. 29,
102e116.
Pinon-Miramontes, M., Bautista-Margulis, R.G., Perez-Hernandez, A., 2003. Removal
of arsenic and fluoride from drinking water with cake alum and a polymeric
anionic flocculent. Fluoride 36, 122e128.
Rasool, A., Farooqi, A., Xiao, T., Ali, W., Noor, S., Abiola, O., Ali, S., Nasim, W., 2018.
A review of global outlook on fluoride contamination in groundwater with
prominence on the Pakistan current situation. Environ. Geochem. Health 40,
1265e1281.
Reardon, E.J., Wang, Y., 2000. Limestone reactor for fluoride removal from waste
water. Environ. Sci. Technol. 34, 3247e3253.
Sujana, M.G., Thakur, R.S., Rao, S.B., 1998. Removal of fluoride from aqueous solution
by using alum sludge. J. Colloid Interface Sci. 206 (1), 94e101.
Sundaram, C., Viswanathan, N., Meenakshi, S., 2009. Defluoridation of water using
magnesia/chitosan composite. J. Hazard Mater. 163 (2e3), 618e624.
Thole, B., 2013. Ground Water Contamination with Fluoride and Potential Fluoride
Removal Technologies for East and Southern Africa, pp. 66e90 (Chapter 4) in
Perspectives in Water Pollution.
Tolkou, A., Zouboulis, A., Samaras, P., 2014a. Development of a sustainable treat-
ment process (MBR) for high-strength industrial wastewaters by the applica-
tion of composite coagulants. In: Proceedings of the International Congress in
Green Infrastructure and Sustainable Societies/Cities (GreInSus 2014), 8 e 10
May, 2014, Izmir, Turkey.
Tolkou, A., Zouboulis, A., Peleka, E., 2014b. Novel pre-polymerized coagulant agents
used for the treatment of industrial wastewaters. In: Proceedings of the 1st IWA
Specialized International Conference “EcoSTP- Eco Technologies for Wastewater
Treatment”, 293-298, 23 e 25 June, 2014b, Verona, Italy.
Tolkou, A.K., Zouboulis, A.I., 2015a. Synthesis and coagulation performance of
composite poly-aluminum-ferric-silicate-chloride coagulants in water and
wastewater. Desalination Water Treat. 53, 3309e3318.
Tolkou, A., Zouboulis, A., 2015b. Synthesis and characterization of a novel composite
pre-polymerized coagulant for water and wastewater treatment. Int. J. Environ.
Eng. 2 (1), 50e54.
Tomar, V., Kumar, D., 2013. A critical study on efficiency of different materials for
fluoride removal from aqueous media. Chem. Cent. J. 7 (1), 51. https://doi.org/
10.1186/1752-153X-7-51.
Tzoupanos, N.D., Zouboulis, A.I., Tsoleridis, C.A., 2009. A systematic study for the
characterization of a novel coagulant (polyaluminium silicate chloride). Colloid.
Surf. Physicochem. Eng. Asp. 342, 30e39.
Waghmare, S.S., Arfin, T., 2015. Fluoride removal from water by calcium materials: a
state-of the-art review. Int. J. Innov. Res. Sci. Eng. Technol. 4 (9), 2347e6710.
World Health Organization, 2011. Fluoride in drinking water. WHO Guidelines, for
Drinking Water Quality, 4th. Ed. World Health Organization, Geneva,
Switzerland.
Zhou, W., Gao, B., Yue, Q., Liu, L., Wang, Y., 2006. Al-Ferron kinetics and quantitative
calculation of Al(III) species in polyaluminum chloride coagulants. Colloid. Surf.
Physicochem. Eng. Asp. 278 (1e3), 235e240.
Zhu, H., Wang, H., Wang, G., Zhang, K., 2010. Removal of fluorine from water by the
aluminum-modified bone char. In: Proceedings of the 2010 International Con-
ference on Biology, Environment and Chemistry (ICBEC 2010), Hong Kong,
China, 28e30 December 2010, pp. 455e457.
Zouboulis, A.I., Tzoupanos, N.D., 2009. Polyaluminium silicate chloride e a sys-
tematic study for the preparation and application of an efficient coagulant for
water or wastewater treatment. J. Hazard Mater. 162, 1379e1389.