Journal of Environmental Chemical Engineering 7 (2019) 103084

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering

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

Electrolysis removal of fluoride by magnesium ion-assisted sacrificial iron T

electrode and the effect of coexisting ions
A.A.G.D. Amarasooriya , Tomonori Kawakami
⁎

Department of Environmental Engineering, Faculty of Engineering, Toyama Prefectural University, 5180 Kurokawa, Imizu-City, Toyama 939-0398, Japan

ARTICLE INFO ABSTRACT

Keywords: A novel system for removing fluoride (F−) by electrolysis (ELC) was investigated as magnesium (Mg2+) ion
Electrolysis assistive system with a sacrificial iron (Fe) electrode. To minimize and control the Fe leachate from the Fe
Mg2+ electrode, ELC was performed in two stages. First stage: Fe dissociation (Fe/stainless steel (SS) electrodes);
Fe second stage: ELC electrocoagulation (SS/platinum (Pt) electrodes). The effect of initial ion concentrations on F-
Co-Precipitation
removal was investigated with different levels of Fe (0–486 mg/L), Mg2+(0–48 mg/L) and coexisting ions cal-
cium (Ca2+), and carbonates (CO32-+HCO3−). Experimental results revealed that the proposed system’s Fe
(486 mg/L) alone may remove 17% of F−. However, incorporating Fe (52 mg/L) and Mg2+ (46 mg/L) sig-
nificantly increased the removal of F− to 77% for 5 mg/L initial F− level and the molar ratio of (0.24–0.94):1 of
the Fe: Mg mixture showed maximum F− removal properties. The F- was removed by both co-precipitation and
by Coulomb forces. It was found that F− was removed by co-precipitating with a mixture of Mg(OH)2 and Fe
(OH)3 as F− was metathesis with the OH- ion due to the similar radius of both ions. To achieve 1.5 mg/L of F−,
the desirable level recommended by the World Health Organization, minimum Mg2+ and Fe concentration ratios
of 5:20, 10:20, and 50:50 mg/L were required, respectively, for the initial 2, 3, and 5 mg/L F- concentrations.
The presence of Ca2+ ions and the presence of CO32-+HCO3− inhibited the system’s functionality significantly.
The optimum operating cost was calculated as 0.56–1.50 US$/m3.

1. Introduction Primarily, electrocoagulation use sacrificial electrode coagulant and

The contamination of groundwater, from naturally occurring rocks,
minerals, soils, or by human activity, limits its availability for human
consumption. Among the contaminants, fluoride (F−) plays a vital role,
since it is an essential micronutrient for the human body, and excessive
F− exposure can cause health issues like dental, skeletal fluorosis and
damage the ecosystem through gastrointestinal or respiratory track
accumulation [1]. The main requisite source of drinking water in most
countries, such as Africa, China, the Middle East, Sri Lanka, and India,
is groundwater rich in F−, where endemic fluorosis are commonly
found [2–4]. This situation has become a major global issue and the
World Health Organization (WHO) recommended a maximum F− level
of 1.5 mg/L for drinking water [5].
Fluoride is one of the most frequently studied contaminates in the
world [6]. For its removal, many technologies have evolved and been
used in the past decades, such as adsorption, chemical coagulation,
membrane separation, reverse osmosis, nanofiltration, ultrafiltration
and electrochemical methods [7–11]. Electrochemical methods can be
classified mainly into two areas, electrocoagulation and electrodialysis.
electrodialysis use less restrictive porous membrane for separating the
dissolved F− [9].
Among those technologies, an electrochemical method, electro-
coagulation (EC), is identified as an emerging technique for the de-
fluoridation of drinking water and industrial wastewater because of its
environmental compatibility, and significant process efficiency [12].
Additionally, the environmental sector has shown a growing interest in
the treatment of different types of water and wastewater by electro-
coagulation [13]. Conversely, electrocoagulation is still struggling to
emerge as a conventional water treatment technology due to the lack of
a systematic approach to EC reactor design/operation and the issue of
electrode reliability, particularly the passivation of the electrodes over
time [12].
Sacrificial aluminum (Al) is commonly used as an electrode material
for de-fluoridation due to its effective EC in removing F−. Adversely,
high Al intake causes neurotoxicity [14,15]. Therefore, the use of Al-
based sludges for the process of removing F- from water could nega-
tively impact human health and the ecosystem. Apart from the Al
electrodes in EC for F− removal, an iron (Fe) electrode was used. The Fe

⁎
Corresponding author.
E-mail address: gayanamarasooriya@gmail.com (A.A.G.D. Amarasooriya).

https://doi.org/10.1016/j.jece.2019.103084
Received 29 January 2019; Received in revised form 8 April 2019; Accepted 10 April 2019
Available online 12 April 2019
2213-3437/ © 2019 Elsevier Ltd. All rights reserved.
A.A.G.D. Amarasooriya and T. Kawakami
electrode in EC has the additional advantages of inexpensiveness,
abundance, and lower toxicity as compared to the Al electrode. How-
ever, few available studies describe its usability and the high ratio of Fe
sludge volume to F− removed could be the main reason that Fe elec-
trodes are not commonly studied [16–18].
Combining Fe with other elements could increase the F− removal in
the EC process. Among the studies of F- removal using sacrificial Fe
electrodes, few described the effect of coexisting ions on F− removal.
Drouiche et al. (2012) proposed removing F− from photovoltaic was-
tewater with Fe-EC bipolar electrodes by pretreating water with Ca
(OH)2 [17]. Later, Govindan et al. (2015) investigated the Fe-EC in-
fluence of coexisting ions Ca2+, Mg2+, and Al3+ on F− removal and
showed that Mg2+ has an effect on F- removal, although this was not
studied in detail [18]. Furthermore, their findings revealed that the
weight fraction of Fe volume to F− removal appears to be significantly
higher than that of the Al electrode EC. Concisely, neither study gave
much attention to minimizing Fe sludge volume, the effects of the
supplementary increasing and decreasing of Fe2+ and Mg2+, the effect
of Ca2+, alkalinity levels, and the assessment of OH− and H+ levels to
optimize the removal of F−.
Usually, anode water oxidation introduced H+, naturalization took
place near the cathode regime, buffering the pH to a near neutral level
by minimizing the EC’s coagulation efficiency and leading to the partial
formation of floc [19]. Furthermore, the non-faradaic contribution to
anodic dissolution leads to a higher electrode material loss [20].
Therefore, overcoming the total effects of the prime residual electrode
materials and the partial removal of F− in the final water become
problematic when using the common single cell EC system. Minimizing
these effects are inevitable when electrocoagulation is performed in a
common single cell EC system.
Electrolysis (ELC) is a widely used technology for the generation of
H2 and the production of Cl2. The ELC cell’s permeable diaphragm splits
the anode and cathode solution and allows the high pH and low pH
solutions to be conserved in each compartment. Furthermore, the
Coulomb forces in the ELC allows transferring ions between the ELC cell
diaphragm to concentrate or dilute. These specific functions can be
utilized to minimize the electrode dissociation and to optimize the
process performances of EC. Furthermore, to date, no ELC study on F−
removal with Fe electrode has been reported. Most importantly, F−
removal by precipitation is absolutely required in the wastewater
treatment, while total removal of both precipitation and coulomb re-
moval are important in the drinking water treatment.
Therefore, the main goal of this work is to study the F− removal by
precipitation (Sacrificial Fe electrode) and utilize coulomb forces for
the F- removal by using ELC. Accordingly, F− removal was studied with
and without Mg2+ ion assistive sacrificial Fe electrode system, and the
performance optimization of F− removal for different levels of Fe and
Mg2+ was studied to minimize Fe and Mg2+ concentrations (total
sludge volume). Furthermore, the effects of coexisting ions Ca2+, F−,
CO32-, and HCO3− on F- removal were also investigated.
2. Materials and methods
2.1. Chemicals, instruments and quality control
De fluoridation experiments were performed with artificially pre-
pared groundwater containing different concentrations of Fe, F−,
Mg2+,Ca2+, HCO3−, and CO32- ions. Analytical-grade chemicals, NaF
(F−), MgCl2.6H2O (Mg2+), CaCl2.2H2O (Ca2+), NaHCO3, and NaCl (as
electrolytes), purchased from Wako, Japan, were dissolved in distilled
water to prepare the artificial groundwater. Cation and anion con-
centrations were measured with ion chromatography (Anions: Thermo
ICS-2000; Separation column: Ion Pac AS18; Eluent: KOH 23–40 mmol/
L gradient/Cations: Thermo ICS-1500; Separation column: Ion Pac
CS12; Eluent: MSA 30 mmol/L isocratic; Thermo Scientific, USA). Prior
to the analysis, samples were filtered with 0.45 μm membrane filter.
2
Journal of Environmental Chemical Engineering 7 (2019) 103084
The standard solution was analyzed after every 20 samples to check the
detector sensitivity. Overall concentration variability of standard sam-
ples were obtained below 5% or otherwise, re-analysis was performed.
Alongside ion chromatographic measurement of F− ions, an ion-
selective electrode method was also performed by coupling the F−
electrode (ORION 9609BNWP, Thermo Scientific, USA) with an Orion
Star A324 meter (Thermo Scientific, USA). Calibration for each new
batch of the samples was performed with TISAB III (Thermo, USA)
buffer solution and intermediate standard samples were analyzed for
each 10 samples for quality control. The pH was measured using a
coupling pH glass electrode (ORION 8107UWMMD, Thermo Scientific,
USA) with an Orion Star A324 meter (Thermo Scientific, USA). Fe
concentrations were measured with atomic absorption spectro-
photometric (AAS) (Varian SpectrAA, USA). To measure total Fe with
AAS, 20 mL of solution collected from the first stage was dissolved in
50% HNO3 acid and filtered with a 0.45 μm membrane filter prior to
injecting AAS Blank analysis and the standard solutions were injected
after each 10 samples for the quality control.
2.2. Experimental setup and process description
Batch ELC experiments were performed with a square-shaped cell
having two equal compartments made of PVC
(9.5(H)×8.5(W)×10.5(L) cm3). Two equal compartments, an anode,
and a cathode were equally separated by using a mixed cellulose ester
membrane filter (membrane material: Mixed cellulos ester,
A020A142C) as a diaphragm with a pore size of 0.2 μm (Advantec,
Toyo Roshi Kaisha, Ltd., Japan). The membrane filter was not com-
mercial type diaphragm but it allows both aniton and cation transferred
through anode and cathode without special ion selectivity. Two paral-
lely connected iron (Fe) electrodes (effective length = 5.5 cm,
φ = 3 mm) and single U-shaped stainless steel (SS) (effective
length = 15 cm, φ = 1 mm) and platinum (Pt) (effective length = 15
cm, φ = 0.40 mm) wire electrodes were used for ELC throughout all
experiments at an inter-electrode distance of 5 cm by utilizing a con-
stant current power source (TSS TAKASAGO, Japan). Pt electrodes were
purchased from the Nilaco Corporation, Japan, and SS and Fe elec-
trodes were obtained from the local market in Japan.
ELC experiments were performed in two stages by utilizing 640 mL
of water in the first stage and 600 mL in the second stage. Twenty mL of
water was collected from each anode and cathode cell after the first
stage for total Fe measurements. First-stage ELC (Electrode configura-
tion: Fe anode/SS cathode) was performed for dissociation of Fe into
the anode solution, and second-stage ELC (Electrode configuration: Pt
anode/SS cathode) was for forming floc of metal hydroxides which co-
precipitates F− in the cathode. In the second stage, first-stage anodes
and cathodes were converse as cathodes and anodes, respectively, by
replacing the electrodes as shown in Fig. 1. After the second stage, the
cathode solution was the treated water with less F− concentration,
while the anode solution was to be discarded.
2.3. Experimental conditions and calculations
Applied charge loading (C/L) is very important in EC to control
electrode dissociation as well as the flocculation process. According to
Faraday’s law (Eq. (1)), the amount of Fe dissociated can be controlled
by changing the applied current or time. Practically 100% of electrons
did not contribute to the leaching of metal ions from the sacrificial
electrode. Some electrons could transfer to the ELC water molecule by
forming H+ ions in the vicinity of the anode. Therefore, calculating the
Fe leached could be problematic. To determine the Fe leached, the
weight differences of the Fe electrode, before and after the first-stage
current application, were used to calculate the total Fe dissolved. In
each experiment, prior to the first stage of ELC, Fe electrode surfaces
were polished with sandpaper, washed, and dried to remove the oxi-
dized layer. Furthermore, the concentration of Fe in the anode solution
A.A.G.D. Amarasooriya and T. Kawakami Journal of Environmental Chemical Engineering 7 (2019) 103084

Fig. 1. Electrolysis electrocoagulation cell configuration in the first and second stages.

was verified with AAS measurements.
Current (A). Time (s )
[Metal]M + dissolved (mol) = /M
Faraday constant (96485.33)(C / mol)
Where M is the valence of the ion
During the second-stage ELC, the pH in the anode effluent de-
creased, and the cathode effluent increased significantly due to the
electrolytic effect on the water molecule. The anode leads to water
oxidation releasing a hydrogen ion (causing lower pH) and oxygen to
the aqueous system. In the cathode, water molecules dissociate into
Hydrogen ions and OH− (causing high pH) by the interaction of water
molecules and electrons. Metallic Fe3+ or/and Fe2+ and Mg2+ ions
combined with OH− and generated metallic hydroxides. The pre-
cipitation of Fe and Mg hydroxides could eliminate F-. Moreover,
Coulomb forces contribute to transporting the ions through the dia-
phragm membrane. Considering both effects, total F- removal by
Coulomb forces and by precipitation were calculated by Eqs. (2) and
(3).
Fluoride transferred to the anode by Coulomb forces (mg) = (Ca
Percentage of F transferred to the anode by Coulomb forces
(C Ci)
= a × 100
Ci
Fluoride removed by precipitation can be described as (Total de-
crease in F− in the cathode - F− transferred to the anode by coulomb
forces)
Accordingly,
Fluoride removal by precipitation (mg) = (Ci Cf ) × V
(Ci Cf ) (Ca Ci )
Fluoride removal % by precipitation=
Ci
(2Ci Ca Cf )
= × 100,
Ci
where
Ci : Initial F− concentration in the anode and cathode (before first
stage) (mg/L)
Ca : Final F− concentration in the anode (after second stage) (mg/L)
Cf : Final F− concentration in the cathode(after second stage) (mg/
L)
V: Anode volume (L)=Cathode volume (L)
To date, no data were found in ELC for F− removal using the Fe
electrode; thus, it is worth studying the effect of dissolved Fe con-
centration on F− removal with and without Mg2+. The dissolved Fe
concentrations were controlled at the first stage by changing the ELC
time and current. After the first-stage Fe dissociation, ELC was con-
tinued for 60 minutes by applying a constant current of 100 mA to gain
a total of 1200 C/L for ELC operation. Total of 1200 C/L was utilized to
obtain the adequate amount of OH− moles in the cathode which could
precipitate 99% of dissolved Mg2+ and Fe in a form of its hydroxides.
To investigate the minimum initial Mg2+ concentration required to
(1) meet WHO’s 1.5 mg/L F− limit, initial F- concentrations ranging from 1
to 5 mg/L and Mg2+ concentrations ranging from 5 to 20 mg/L were
used at a constant concentration of initial Fe (20 mg/L) in ELC.
Groundwater or raw water contains several other ions that may
interfere or compete with F− during Mg and Fe assisted ELC. The most
common ion found in water is Ca2+. Therefore, the effect of the pre-
sence of the Ca2+ ion was studied. Initial levels of Fe 20 mg/L, Mg2+
20 mg/L, and F− 5 mg/L were investigated with varying Ca2+. Second-
stage C/L was kept at 1200 C/L throughout the experiments. Other than
the Ca2+, the most abundant naturally occurring ions in the water that
may affect process efficiency are carbonates (HCO3− and CO32-), which
cause alkalinity. Due to the diprotic nature of carbonic acid (H2CO3) in
the solution, both CO32- and HCO3− ions (hereafter (HCO3−+ CO32-))
are available, and their concentrations would change with the solution
pH. Therefore, (HCO3- + CO32-) as mmol/L accounted for the alkali-
nity. The charges of the water samples (HCO3- + CO32-) were calcu-
Ci ) × V
lated by accounting system ions charge balances. The tracer ion con-
centrations were neglected, and major ions were involved with the
calculation (Eq. (4)). For calculating individual HCO3- and CO32- con-
centrations as mmol/L, as described in Eqs. (7) and (6), we employed
(2) the carbonates’ charge balance equation (Eq. (5)) and carbonates’
equilibria (Eq. (4)). The K2 = 4.69 × 10 11 (mol/L) value was obtained
from Plummer and Busenberg's 1982 study findings [21].
(TC)= Total charge of CO32− and HCO3- = (sum of cation charge – the
sum of anion charge except for CO32− and HCO3-)
(Ca Ci) × V
TC = {2 × [CO 32 ] + [HCO 3 ]}(eq/ L) = {[H+] + [Na2 +]
× 100 + [NH4+] + [K+] + 2 × [Mg 2 + ] + 2 × [Ca2 +]}
{[OH ] + [F ] + [Cl ] + 2 × [SO 24 ] + [NO 3 ]}(eq/ L). (4)
(3)
The charge balance for CO32− and HCO3- can be written as (as-
sumed no dissociation of atmospheric CO2(g)) :
TC = 2[CO32 (aq)]equilibrium + [HCO3 (aq)]equilibrium (5)
HCO3− ↔ CO3- equilibria:
K2 = 4.69 x 10 11(mol/L)
HCO3 (aq)3 =========== CO32 (aq) + H+ (aq)
[CO3 2 (aq)]eqilibrium × [H+ (aq)]equilibrium
K2 =
[HCO3 (aq)]equilibrium (6)
Combining Eqs. (5) and (6)

3
A.A.G.D. Amarasooriya and T. Kawakami Journal of Environmental Chemical Engineering 7 (2019) 103084

Fig. 2. Variation of F− removal as a function of varying initial Fe concentrations/first-stage charge loadings in the cathode (initial F− = 5.0 mg/L, NaCl = 2 mg/L;
C/L applied at second-stage = 1200).

[HCO3 (aq)]equilibrium =
TC
(eq/ L) or (mol/L) As the initial Fe concentration increases, the total surface area of Fe
1 + (2K2/[H+ (aq)]equilibrium ) (OH)3 increases. Thus, the removal of F− should eventually increase
(7) agreeing with Fig. 2 results. When the initial Fe concentration was in-
creased, Fe(OH)3 formation increased, causing increased F− removal as
a form of precipitate in the cathode, creating an increment of F− ion
3. Results and discussion concentration gradient between cathode and anode. An increasing F-
ion concentration gradient leads to more F- ion transferring from the
3.1. Effect of the initial Fe concentration on F− removal in the absence of anode to the cathode bath by overcoming the applied Coulomb forces.
Mg2+ Clear evidence of this phenomenon can be seen in the 90 to 480 mg/L
initial Fe level in Fig. 2, as the removal of F− as precipitates increases
By varying the first-stage ELC time and current (for dissolving the over the initial Fe concentration, while F- removal by Coulomb forces
desired concentration of Fe) but by keeping the second-stage C/L con- plunged significantly.
stant (100 mA for 60 min = 1200 C/L), the rate of F− removal in the At the anode (first stage)
synthesized groundwater was investigated. Variations of F− removal is Fe(s) Fe3 +(aq) + 3e (8)
shown in Fig. 2 as a function of initial Fe concentration. According to
the results shown in Fig. 2, when the initial Fe concentrations (first- At the cathode (second stage)
stage charge loading) increased, F- removal by precipitation increased 2H2 O(l) + 2e 2OH (aq) + H2(g ) (9)
marginally. The maximum F− removal by precipitation (14%) was
achieved with an initial Fe concentration of 486 mg/L. For initial Fe KspFe (OH )3 = 2.79 × 10 39 (mol / L)4
concentrations of 50–100 mg/L, around 4% of F− was removed by Fe3 +(aq) + 3OH (aq) Fe (OH )3(s ) (10)
precipitation; however, higher F- removal 72–71% by Coulomb forces Fluoride co-precipitation
was experienced. This level was a 34% increase over the initial level of
zero Fe in the solution. Fe 3 + (aq) + (3 ) OH ( q) + F ( q) Fe (OH )3 F (s ) (11)
The predominant species of Fe in the pH range of 8–12 is Fe(OH)3
Fluoride Adsorption
[22]. It can be further certified with the Ksp values, 4.87 × 10−17
(mol/L)3 and 2.79 × 10-39 (mol/L)4 of Fe(OH)2 and Fe(OH)3 formation, Fe (OH )3 (s ) + F (aq) Fe (OH )(3 ) F( / ) (s ) + (OH ) (aq)
respectively [23]. Since, second-stage pH levels of 11.75–8.76 were
observed for initial Fe concentrations of 0–486 mg/L, precipitation of (12)
−
Fe(OH)3 take place in the cathode of the second stage. The amorphous With the limited data available from studies of F removal by Fe
metallic hydroxide (Al and Fe) formed has a large surface area with a electrode, and the data from the current study are summarized in
high adsorption capacity, which forms a floc and makes a bond with Table 1. According to the data summarized, it appears that the pro-
pollutants [19,24]. Furthermore, as described in previous studies, F- posed Fe-ELC system is capable of removing F- up to some degree.
removal may occur due to sweep coagulation or the enmeshment of F- However, it seems that removal was slightly less efficient than in pre-
ions by insoluble Fe hydroxide precipitate (co-precipitation) in the so- vious studies. As described by Drouiche (2012), optimum F− removal in
lution [16,25]. Accordingly, overall reactions in the first and second EC was achieved at pH 6, and, as the pH further increased, F- removal
stages can be summarized as Eqs. (8–12). affinity tended to decrease [17]. Similarly, the operational pH value of

Table 1
F− removal by sacrificial Fe electrodes in EC and ELC.
Initial Removal as precipitate % First stage C/L (Fe) Total C/L Fe2+ leachate (mg/L) Removal of F− by Fe (mg/g) Reference

F−(mg/L) Mg2+(mg/L)

25 0 40.0 0 3600 1042 9.6 EC- [16]

4.5 0 8.1 0 922 267 1.4 EC- [26]
10 0 18.0 0 1620 469 3.8 EC- [18]
5 0 4.1 176 1376 50 4.0 ELC- (this study)
5 0 4.2 332 1532 90 2.4
5 0 6.5 938 2138 282 1.2
5 0 14.4 1650 2850 486 1.1

4
A.A.G.D. Amarasooriya and T. Kawakami
Fig. 3. Variation of F− removal with initial Fe concentration, C/L applied at
first stage = 18.75–187.5, C/L applied at second stage = 1200, initial Mg
concentration = 20 mg/L, initial F− concentration = 5 mg/L.
this system was higher than pH 6 which contribute to the decreased F-
removal than neutral pH operated EC studies.
3.2. Effect of the initial Fe concentration on F− removal in the presence of
Mg2+
The role of the initial Fe concentration on ions and F− removal in
the presence of Mg2+ is shown in Fig. 3. Throughout the ELC experi-
ments, the initial Mg2+ level was kept constant at 20 mg/L, and initial
Fe concentrations were varied from 0 to 53 mg/L. According to Fig. 3a,
it is clear that in the absence of Fe, Mg2+ alone may remove F− as a
precipitate (37%). It is obvious that Mg(OH)2 contributed to removing
the F− in the form of a precipitate due to the higher alkaline nature at
the cathode. In addition F− could also precipitate as MgF2, however,
when comparing the solubility products (Ksp) of MgF2 (5.16 × 10-11
(mol/L)3) with that of Mg(OH)2 (5.61 × 10-12 (mol/L)3), most aban-
doned precipitate from the alkaline medium should be the Mg(OH)2.
The previous studies of de-fluoridation of water using nano-magnesium
oxide suggested that F− metathesis between Mg(OH)2 and F− could
occur due to the isoelectric nature and similar size of F− and OH- ions
[27,28]. Nevertheless, F- ion adsorption by Mg(OH)2 was not reported
in previous studies. Accordingly, it can be proposed that the F- ion
could be exchanged with the OH− ion and co-precipitated with Mg
(OH)2 (Eq. (13)).
5
Journal of Environmental Chemical Engineering 7 (2019) 103084
aMg (OH ) 2 (s ) + bF (aq) aMg (OH ) (2 b F(b / a) (s ) + bOH (aq)
a)
(13)
At an initial Fe levels of 11–43 mg/L, the removals of F− as pre-
cipitates were maximized and approximately remained at a constant
level. But, after the initial Fe level > 43 mg/L, F− removal as pre-
cipitate was decreased as shown in Fig. 2a. Therefore, the co-existence
of Mg hydroxide along with the Fe hydroxide at an Mg: Fe molar ratio
of 1: (0.24–0.94), was found to be the most favorable condition in re-
moving F−. As described previously, both Mg(OH)2 and Fe(OH)3 ex-
hibit F− co-precipitation properties. Consequently, increasing Fe con-
centrations in the range of 10–43 mg/L could increase F− removal if the
proper geometric orientation of floc is formed to enmesh and me-
tathesize the F− ion with OH−. As a result, the mechanism of F− re-
moval could be predominantly co-precipitation as expressed in Eq. (14).
xMg (OH )3 (s ) + yMg (OH )2 (s ) + zF (aq) { Fe (OH )3 }x {Mg (OH ) 2 }y…
(F ) z (s ) (14)
The data from Figs. 2 and 3 make it clear that the removal of F− by
Fe(OH)3 was weaker than that by Mg(OH)2. A further increase of the
initial Fe level > 43 mg/L dominates the formation of Fe(OH)3, and the
floc of the Mg(OH)2 surrounded by more Fe(OH)3. This condition could
start preventing the F− ion from metathesis with OH− ion in mostly F−
co-precipitating floc Mg(OH)2. Hence, the increase of Fe concentration
beyond 43 mg/L could negatively affect F− removal as observed in
Fig. 3a.
As the initial Fe in the cathode was increased, F− removal by
Coulomb forces showed an increasing trend in the cathode (Fig. 3a),
while F- in the anode showed a negative removal (Fig. 3b). As described
previously, the F- ion in the cathode was retransferred to the anode at
the second stage by coulomb forces, resulting in an increase in F-
concentration.
The data from Fig. 3c and b show that more than 99% removal in
the cathode by precipitation and 47% removal of Mg2+ from the anode
bath by transfer to the cathode to be precipitated were achieved. The
formation of Mg(OH)2 leads to Mg2+ removal in the cathode, and both
the ion concentration gradient and Coulomb forces cause the trans-
portation of Mg2+ from the anode to the cathode. Furthermore, more
than 99% of Fe removal in the cathode was observed for all levels of
initial Fe, assuring the final water quality up to the level of absence of
electrode material. Furthermore, the Coulomb transportation of Na+
and Cl− ions can be clearly seen in Fig. 3b and c. Positive Na+ transfer
to the cathode bath by causing negative removal in the cathode (Na+
increment). The negative Cl− ion was transported to the anode bath by
creating negative removal in the anode bath. However, the mobilization
from the cathode to the anode was about 40% different. The production
of Cl2(g) and its evaporation at the anode caused this difference.
3.3. Effect of initial Mg2+ concentration on F− removal in the presence of
Fe
Fig. 4a–c illustrates the effect of the initial Mg2+ concentration on
the ion removal and F− removal efficiencies at constant initial Fe
concentrations (averaged value = 51 mg/L). According to Fig. 4c, with
increasing initial Mg2+, the removal of F− showed a significant in-
crement. Distinctly, the increase of Mg2+ brought down the F- removals
by Coulomb forces, while removal by precipitation increased. Total
removal approximately remained constant throughout the Mg2+ in-
crements. As described previously (Section 3.3), a (0.24-0.94):1 molar
ratio between initial Fe and Mg2+ can be identified as the maximal F−
removing condition. By agreeing with that results, maximal F− removal
by precipitation (72%) was observed (Fig. 4c) at the initial level of Fe:
Mg molar ratio 0.43:1. Within 60 minutes of ELC time in the second
stage, total F− removal persisted higher than 68% for increasing the
Mg2+ concentration from 0 to 47 mg/L. For initial Mg2+ levels of
A.A.G.D. Amarasooriya and T. Kawakami
Fig. 4. Variation of F− removal with initial Mg2+ concentration, C/L applied at
first stage = 187.5, C/L applied at second stage = 1200, average initial Fe
concentration = 51 mg/L, initial F- concentration = 5 mg/L.
5–10 mg/L, F− removal via precipitation increased sharply (from 7 to
33%); however, for Mg2+ levels of 10–47 mg/L, F− removal via pre-
cipitation increased marginally (from 33 to 68%).
Compared to initial levels of Fe 0, Mg2+ 0 mg/L (from Section 3.1),
Fe 0, Mg2+ 20 mg/L (from Section 3.2), Fe 20, Mg2+ 20 mg/L (from
Fig. 5. Variation of F− removal with initial F− and Mg2+ concentrations, C/L applied at first stage = 75.0,C/L applied at second stage = 1200, initial Fe con-
centration = 20 mg/L.
6
Journal of Environmental Chemical Engineering 7 (2019) 103084
Section 3.2), and Fe 50, Mg2+ 46 mg/L (this section), a total of 37, 47,
71 and 77% and by precipitation 0, 37, 50 and 72% F− were removed a
respectively. Therefore, an increment of both initial concentrations of
Mg2+ and Fe, lead to the highest removal of total F-. But, increase of
initial concentrations of Fe and Mg2+ more than 20 mg/L, showed a
slight effect on the total F- removal (only 1% total increment). How-
ever, with increasing both initial concentrations of Fe and Mg2+ more
than 20 mg/L, the F- removal by precipitation shows a significant in-
crease (22%).
Compared to previous EC studies (Table 1 data), this study proposed
Mg2+ ion-assisted sacrificial Fe electrode batch system that compara-
tively showed much higher removal efficiency with lower electrode
material dissociation and sludge generation. The Coulomb transporta-
tion of Na+ and Cl− ions is also seen in Fig. 4a and b. It seems that
increasing the Mg2+ ion concentration causes a decrease in Na+
movement from the anode bath to the cathode bath. Similarly, Cl− ion
mobilization decreased from the cathode to the anode bath.
3.4. Effect of the initial F− and Mg2+ concentration on F- removal
To evaluate the initial concentrations of Mg2+ and F− on the F−
removal, a series of electrolysis experiments were performed with a
different initial level of F− and Mg2+. Initial Fe concentration was kept
constant at 20 mg/L and 1200 C/L of charge loading was applied at the
second stage. The results obtained are shown in Fig. 5a–c. According to
Fig. 5a precipitation removal, to meet the 1.5 mg/L final F− level for
the initial F− levels of 3 mg/L, the minimum level of initial Mg2+ re-
quired was 10 mg/L. By using an initial level of 5 mg/L of Mg2+, a
maximum of initial 2.25 mg/L of F− can be treated to meet the 1.5 mg/
L WHO desirable drinking water F- level. The optimum level of F−
removal to reach the minimum level of Mg2+ is an initial 10 mg/L
concentration for a 3 mg/L initial F− level (75 mg/g of F− removed by
Fe).
However, for the initial F− level of 5 mg/L, the initial Mg2+ 5, 10 or
20 mg/L was found to be insufficient for the adequate removal of F− by
precipitation. From Section 3.3 data, the initial Mg2+ 46 mg/L and Fe
50 mg/L could meet the 1.5 mg/L, F− standard (72% removal).
Therefore, to meet the desirable F- concentration in treated water initial
concentration of the Fe and Mg2+ should be increased
Fig. 5b shows that F− removal by Coulomb forces increased with
increasing F- concentration for initial Mg2+ levels of 5 and 10 mg/L.
However, for an initial Mg2+ level of 20 mg/L, Coulomb removal was
reduced with increasing F- concentration. It’s clearly the decrease F- ion
in the precipitate tended to be lower when the initial Mg2+
A.A.G.D. Amarasooriya and T. Kawakami Journal of Environmental Chemical Engineering 7 (2019) 103084

Fig. 6. Variation of F− removal with varying initial Ca2+, C/L applied at first
stage = 75.0, C/L applied at second stage = 1200, initial F− = 5 mg/L,
Fe = 20 mg/L, Mg2+ 20 mg/L. Fig. 7. Variation of F− removal with varying initial alkalinity as (CO32-
+HCO3−) μmol/L, C/L applied at first stage = 75.0, C/L applied at second
stage = 1200, initial F− = 5 mg/L, Fe = 20 mg/L, Mg2+ 20 mg/L.
concentration was reduced. Therefore, with increasing both F- and
Mg2+, the ability to co-precipitate increased. As shown in Fig. 5c, the
respectively. These removal levels were not matched with the higher
total removal of F- continued above 70% during increments of initial F-
levels of the experimental Ca2+ removals shown in Fig. 6a and b (The
and Mg2+ concentrations, adequate for meeting the 1.5 mg/L of WHO
experimental totals of Ca2+removed from the anode and the cathode
guideline. Furthermore, for all experimental levels shown in Fig. 5, the
for 17 mg/L and 148 mg/L initial Ca2+ were found to be 60%
final removal of Fe and Mg2+ from water was higher than 99%, en-
(33.4 + 26.7) and 16% (20.2–4.6), respectively). Due to the highly
suring the absence of dissociated electrode material (not shown here).
alkaline nature of the cathode bath, atmospheric CO2 (g) could dis-
sociate in the effluent. Dissolved CO2 (g) formed HCO3-and CO32-,
3.5. Effect of the Ca2+ ion on the F− removal process
which could remove Ca2+ as CaCO3 and could cause the differences
between experimental and calculated values. Furthermore, for all ex-
The effect of the Ca2+ ion on F− removal in the ELC process was
perimental levels shown in Fig. 6, the removal of Fe and Mg2+ from the
described in this section. Fig. 6 shows that F- removal by precipitation
final water was found to be higher than 99%, ensuring the absence of
was reduced to 22% from 56% for 0–150 mg/L initial Ca2+. The for-
electrode material and Mg2+.
mation of Ca(OH)2 and CaCO3 in the alkaline medium could interfere
with the Mg2+/Fe mixture that co-precipitates F-. The formation of that Ksp = 5.02 × 10 6 (mol / L)3
precipitation can be confirmed by solubility product (Ksp) calculations. Ca2 + (aq) + 2OH (aq) Ca (OH ) 2 (s )
By using the Ksp of Ca(OH)2 = 5.02 × 10-6 (mol/L)3 [23] and mea-
sured equilibrium pH 11.69–11.43 levels (initial Ca2+ 17–148 mg/L), [Ca2 +]equilibrium =
Ksp
the amount of Ca(OH)2 precipitated can be calculated with Eqs. (15) [OH ]2equilibrium (15)
and (16). For the initial levels of Ca2+ 17 and 148 mg/L, the calculated
equilibrium concentrations of Ca2+ were found 9 and 37 mg/L, re-
{[Ca2+]Initial [Ca2+]equilibrium } × 100
spectively. Accordingly, calculated Ca2+ removal percentages were [Ca2 +]removal = %
[Ca2 +]Initial (16)
47% and 18% for the initial level of Ca2+ 17 and 148 mg/L,

7
A.A.G.D. Amarasooriya and T. Kawakami
Table 2
Total individual ion removals in the (CO32−+ HCO3-) system.
(CO32−+ HCO3-) μmol/L (mmol/L)
H+/OH− Mg2+ Fe3+ C initial
0 12.4 0.8 0.4 0.0
2744 12.4 0.8 0.4 2.7
5072 12.4 0.8 0.4 5.1
7147 12.4 0.8 0.4 7.1
9567 12.4 0.8 0.4 9.6
3.6. Effect of (CO32−+HCO3-) ions on the F- removal process
In this section, the effect of (CO32−+ HCO3-) on electrolysis was
analyzed. As shown in Fig. 7a, the presence of HCO3- ions showed a
significant negative impact on F- removal by precipitation. As the initial
(CO32−+ HCO3-) increased, F- removal by precipitation rapidly de-
creased to 0.4% at 2743 μmol/L. The presence of (CO32−+ HCO3-) may
slightly reduce the formation of Mg(OH)2 and Fe(OH)3 by forming
MgCO3 and Fe2(CO3)3. The solubility product value of Mg(OH)2
5.16 × 10-11(mol/L)3, Fe(OH)3 2.79 × 10-39(mol/L)4, MgCO3
6.82 × 10-6 (mol/L)2, and Fe2(CO3)3 3.36 × 10-9 (mol/L)5 [23] con-
firms the formation of Mg(OH)2, Fe(OH)3 in the system, but the neg-
ligible formation of MgCO3 and Fe2(CO3)3. Calculated and measured
individual molar concentrations of ion removals are shown in Table 2.
The H+ and OH- (Table 2, column 1) concentrations were calculated by
assuming all electrons were contributed to form H+ and OH- by water
molecule ELC (H+/OH-=Coulomb/Faraday’s constant). Compared to
the H+ concentrations and total carbon removal, H+ alone could re-
move all of the (CO32−+ HCO3-) as CO2 (columns 2 and 8). If so, Mg2+
and Fe should be precipitated as Fe(OH)3 and Mg(OH)2, which causes F-
co-precipitation. Since F-was not removed by precipitation, it is obvious
that CO32− and HCO3- prohibited F- enmeshing, metathesis thus, co-
precipitation. Therefore, prior to using water containing CO32− and
HCO3- for de-fluoridation, ions should be removed from the solution.
The solution from the second-stage anode bath can be utilized con-
veniently for F- removal since it had pH levels of 2.62–2.89, which were
enough for removing CO32− and HCO3- by H+ ions as CO2(g).
3.7. Estimation of energy consumption and operating cost
The power consumption of the ELC system directly depends on the
diaphragm resistance and the conductivity of the solution. In this study,
much attention was not paid for decreasing the power consumption by
utilizing a commercial type diaphragm or by increasing the electrical
conductivity of the solution. In this preliminary cost analysis the cost
for power consumption, Mg and the sacrificial electrodes were taken
into consideration. Market prices of electricity (0.18 US$/kWh), 99%
pure Mg salt MgCl2.6H2O (0.60 US$/kg) and the 99% pure Fe (0.60 US
$/kg) in Japan for the year 2019 were used for the calculations. The
averaged voltage values during the constant 100 mA operation (both
first and second stages) were observed in between 17–25 V for the in-
itial Mg2+ 20, Fe 10–53 and F− 5 mg/L. The power consumption cost
was calculated by using the Eq. (17) and it was varied from 1.02 to 1.50
US$/m3 respectively for the initial Mg2+ 20, Fe 10 mg/L and Mg2+ 20,
Fe 53 mg/L. By using the Eq. (18), sacrificial Fe electrode and Mg2+
chemical costs were calculated separately. By adding power, chemical
Mg and Fe electrode costs, total cost was calculated and it was varied
from 1.13 to 1.63 US$/m3 respectively for the initial Mg2+20, Fe
10 mg/L and Mg2+ 20, Fe 53 mg/L.
Average Voltage (V ) × Current (I ) × Time (h)
Power Cost (US $/ m3) =
Volume treated (m3) × 103
US $
× Energy Price
kWh
Journal of Environmental Chemical Engineering 7 (2019) 103084
Carbon removed at the cathode Carbon removed at the anode Total carbon removal
0.0 0.0 0.0
1.9 2.7 4.6
2.7 5.1 7.8
3.6 6.3 10.0
4.5 6.8 11.3
Material Cost (US $/ m3)
=
Material Price ( ) × Dissolved amount (mg/L) × Molar fraction
US $
mg
10 3
(18)
The recent study by Thakur et al., 2019, reported that the cost for
removing F− by Al sacrificial electrodes with EC was 0.36 US$/m3 for
the industrial electricity price of 0.08 US$/kWh in India [29]. Ac-
cordingly, recalculated total cost for ELC varied from 0.56 to 0.80 US
$/m3 and found higher than the most recent Al electrode EC process.
4. Conclusion
The ELC technique was studied to optimize the removal of F− as an
2+
Mg ion assistive sacrificial Fe electrode system with a major goal of
minimizing sludge volume to pollutant removal ratio. The performance
of the ELC system for removing F− with different levels of Fe and Mg2+
was studied to minimize the concentrations of Fe and Mg2+. The effect
of co-existing ions Ca2+, CO32-, and HCO3− on the system performance
was investigated. Findings revealed that the Fe ELC system was capable
of removing F− in a form of precipitate and by Coulomb forces, even in
the absence of Mg2+. However, precipitation removal sharply increased
in the presence of Mg2+ from 14% to 72% for initial 0 Mg2+, 486 Fe
mg/L and 50 Fe, 50 Mg2+ mg/L, initial concentrations respectively.
Fluoride was removed by co-precipitating with Mg(OH)2 and Fe(OH)3
precipitates a mixture predominantly as a result of the metathesis of F−
ions with OH- ions, in the precipitate. Increasing the initial Fe con-
centration increased F− removal to a certain level in the presence of
Mg2+; however, increasing the Fe further caused a decrease in F− re-
moval. The minimal levels of the Fe and Mg2+ ratio required to meet
WHO standards for F− were found to be 5:20, 10:20, and 50:50 mg/L,
respectively, for initial levels of 2, 3, and 5 mg/L F−. The presence of
Ca2+ ions inhibited the removal of F− by precipitation, to some extent.
The presence of CO32-+HCO3− totally inhibited the removal of F− by
precipitation, even at low concentrations. Therefore, CO32-+HCO3−
should be removed prior to operation. The optimum operating cost was
calculated as 1.02 to 1.50 US$/m3 in the developed country Japan and
0.56 to 0.80 US$/m3 in developing country in Asia.
Acknowledgment
This research was supported byJapan society for the promotion of
science (JSPS)–KAKENHI Grant Number 15H05120 and 17K18910.
References
[1] H. Spencer, I. Lewln, Fluoride metabolism in man, Am. J. Med. 49 (6) (1970)
807–813.
[2] Y. Chung, D. Shin, S. Park, Y. Lim, Y. Choi, S. Cho, J. Yang, M. Hwang, Y. Park,
H. Lee, Risk assessment and management of drinking water pollutants in Korea,
Water Sci. Technol. 36 (1997) 309–323.
[3] C.B. Dissanayake, Water quality in the dry zone of Sri Lanka – some interesting
health aspects, J. Natl. Sci. Found. Sri Lanka 33 (3) (2005) 161–168.
[4] N. Mameri, A.R. Yeddou, H. Lounici, D. Belhocine, H. Grib, B. Bariou,
Defluoridation of septentrional Sahara water of north Africa by electrocoagulation
(17) process using bipolar aluminium electrodes, Water Res. 32 (5) (1998) 1604–1612.
8
A.A.G.D. Amarasooriya and T. Kawakami
[5] WHO, Guidelines for Drinking–water Quality: Fourth Edition Incorporating the First
Addendum, 4th ed., World Health Organization, 2017.
[6] H. Fu, M. Wang, Y. Ho, Mapping of drinking water research: a bibliometric analysis
of research output during 1992–2011, Sci. Total Environ. 443 (2013) 757–765.
[7] A. Bhatnagar, E. Kumar, M. Sillanpää, Fluoride removal from water by adsorption-a
review, Chem. Eng. J. 171 (3) (2011) 811–840.
[8] S.V. Jadhav, E. Bringas, G.D. Yadav, V.K. Rathod, I. Ortiz, K.V. Marathe, Arsenic
and fluoride contaminated groundwaters: a review of current technologies for
contaminants removal, J. Environ. Manage. 162 (2015) 306–325.
[9] M. Mohapatra, S. Anand, B.K.K. Mishra, D.E. Giles, P. Singh, Review of fluoride
removal from drinking water, J. Environ. Manage. 91 (1) (2009) 67–77.
[10] S. Pulkka, M. Martikainen, A. Bhatnagar, M. Sillanpää, Electrochemical methods for
the removal of anionic contaminants from water – a review, Sep. Purif. Technol.
132 (2014) 252–271.
[11] J. Shen, A. Schafer, Removal of fluoride and uranium by nanofiltration and reverse
osmosis: a review, Chemosphere. 117 (2014) 679–691.
[12] M.Y. Mollah, R. Schennach, J.R. Parga, D.L. Cocke, Electrocoagulation
(EC)—Science and Applications, J. Hazard. Mater. 84 (1) (2001) 29–41.
[13] V. Kuokkanen, T. Kuokkanen, J. Rämö, U. Lassi, Recent applications of electro-
coagulation in treatment of water and wastewater—A review, Green Sustain. Chem.
03 (2) (2013) 89–121.
[14] A. Campbell, The potential role of aluminium in Alzheimer’s disease, Nephrol. Dial.
Transplant. 36 (2) (2002) 17–20.
[15] D. Krewski, R.A. Yokel, E. Nieboer, D. Borchelt, J. Cohen, S. Kacew, J. Lindsay,
A.M. Mahfouz, V. Rondeau, Human health risk assessment for aluminium, alumi-
nium oxide, and aluminium hydroxide, J Toxic. Environ. Heal. B 10 (1) (2009)
1–269.
[16] N. Drouiche, N. Ghaffour, H. Lounici, N. Mameri, A. Maallemi, H. Mahmoudi,
Electrochemical treatment of chemical mechanical polishing wastewater: removal
of fluoride - sludge characteristics – operating cost, Desalination 223 (1–3) (2008)
134–142.
[17] N. Drouiche, S. Aoudj, H. Lounici, M. Drouiche, T. Ouslimane, N. Ghaffour, Fluoride
removal from pretreated photovoltaic wastewater by electrocoagulation: an in-
vestigation of the effect of operational parameters, Procedia Eng. 33 (2012)
385–391.
Journal of Environmental Chemical Engineering 7 (2019) 103084
[18] K. Govindan, M. Raja, S. Uma Maheshwari, M. Noel, Y. Oren, Comparison and
understanding of fluoride removal mechanism in Ca2+, Mg2+ and Al3+ ion as-
sisted electrocoagulation process using Fe and Al electrodes, J. Environ. Chem. Eng.
3 (3) (2015) 1784–1793.
[19] P.V. Nidheesh, T.S.A. Singh, Arsenic removal by electrocoagulation process:Recent
trends and removal mechanisam, Chemosphere 181 (2017) 418–432.
[20] Z. Gu, Z. Liao, M. Schulz, J.R. Davis, J.C. Baygents, J. Farrell, Estimating dosing
rates and energy consumption for electrocoagulation using iron and aluminum
electrodes estimating dosing rates and energy consumption for electrocoagulation
using iron and aluminum electrodes, Ind. Eng. Chem. Res. 48 (6) (2009)
3112–3117.
[21] L.N. Plummer, E. Busenberg, The solubilities of calcite, aragonite and vaterite in
CO2-H2O solutions between 0 and 90°C, and an evaluation of the aqueous model for
the system CaCO3-CO2-H2O, Geochim. Cosmochim. Acta 46 (6) (1982) 1011–1040.
[22] A. Stefánsson, Iron(III) hydrolysis and solubility at 25°C, Environ. Sci. Technol. 41
(17) (2007) 6117–6123.
[23] D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, Internet Version 2005,
93rd ed., CRC Press, Boca Raton, FL, 2005.
[24] M. Kobya, H. Hiz, E. Senturk, C. Aydiner, E. Demirbas, Treatment of potato chips
manufacturing wastewater by electrocoagulation, Desalination 190 (1–3) (2006)
201–211.
[25] M.M. Singh, Z. Szafran, J.G. Ibanez, Laboratory experiments on the electrochemical
remediation of environment. Part 4: color removal of simulated wastewater by
electrocoagulation-electroflotation, J. Chem. Educ. 75 (8) (1998) 1040.
[26] X. Zhao, B. Zhang, H. Liu, J. Qu, Simultaneous removal of arsenite and fluoride via
an integrated electro-oxidation and electrocoagulation process, Chemosphere 83 (5)
(2011) 726–729.
[27] R.R. Devi, I.M. Umlong, P.K. Raul, B. Das, S. Banerjee, L. Singh, Defluoridation of
water using nano-magnesium oxide, J. Exp. Nanosci. 1 (13) (2012) 512–524.
[28] N.A. Oladoja, B. Helmreich, Calcium aluminate-diatomaceous earth composite as a
reactive filter material for aqua defluoridation, J. Water Process Eng. 9 (1) (2016)
58–66.
[29] L.S. Thakur, H. Goyal, P. Mondal, Simultaneous removal of arsenic and fluoride
from synthetic solution through continuous electrocoagulation: operating cost and
sludge utilization, J. Environ. Chem. Eng. 7 (1) (2019).