CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3

available at www.sciencedirect.com

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

Preparation and characterization of charcoals that contain

dispersed aluminum oxide as adsorbents for removal
of fluoride from drinking water

Eric Tchomgui-Kamga a,b, Véronique Alonzo a, Charles P. Nanseu-Njiki b,

Nathalie Audebrand c, Emmanuel Ngameni b, André Darchen a,*
a
UMR CNRS no. 6226 Sciences Chimiques de Rennes, ENSCR, Avenue du Général Leclerc, CS 50837, 35708 Rennes Cedex 7, France
b
Laboratoire de Chimie Analytique, Faculté des Sciences, Université de Yaoundé-I, BP 812 Yaoundé, Cameroon
c
UMR CNRS no. 6226 Sciences Chimiques de Rennes, Université de Rennes-1, Avenue du Général Leclerc, 35042 Rennes Cedex, France

A R T I C L E I N F O A B S T R A C T

Article history: Charcoals adsorbents that contain dispersed aluminum and iron oxides have been synthe-
Received 5 May 2009 sized by impregnating wood with salt solutions followed by carbonization at 500 C, 650 C
Accepted 9 September 2009 or 900 C. The adsorbents were characterized and their performance for fluoride removal
Available online 13 September 2009 from aqueous solution was evaluated. Aluminum and iron oxides were well dispersed into
the porous charcoals. The carbons were amorphous and highly porous. XRD of the adsor-
bents showed crystalline iron oxide but did not show any form of crystalline aluminum oxi-
des. All the adsorbents showed acidic surface properties. The efficiency of defluoridation
was found to depend on the carbonization temperature, the pH of point of zero charge
(pHPZC), and the co-existing ions. Substrates prepared at 650 C with aluminum and iron
oxides exhibited the best efficiency with a fluoride sorption capacity of 13.64 mg g1. More
than 92% removal of fluoride was achieved within 24 h from a 10 mg L1 solution at neutral
pH. Fluoride adsorption kinetic was well fitted by a pseudo-second order model. The
amounts of residual Al and Fe in treated solution were pH dependant. At neutral pH, the
amounts of dissolved Al and Fe were found to be 0.67 and 1.8 mg L1, respectively.
 2009 Elsevier Ltd. All rights reserved.

1. Introduction
The quality of drinking water is very important for public
safety and the quality of life. The contamination of drinking
water with fluoride ions is a real health problem, especially
in arid and semi-arid areas where geology provides sources
of fluoride ions. For example, the fluoride concentration in
groundwater reaches 20 mg L1 and more, in some regions
of North Africa, India and China [1–4], this is higher than
the standard set by World Health Organization (WHO) which
is 1.5 mg L1. Excess fluoride in drinking water causes dental
and skeletal fluorosis [3,4] and decreases growth and intelli-
gence [5]. Therefore, many techniques for the removal of fluo-
ride ions from water have been developed to reduce the
acuity of their damage to human health and the environmen-
tal risks that they induce.
The most commonly used methods for defluoridation of
water can be divided into two main categories: precipitation
and adsorption [1,6]. The precipitation of fluoride by calcium
salts [7–9] and aluminum salts [10–12] has already been inves-
tigated. The residual concentration of fluoride ions in water is
greater than 2 mg L1 when calcium salts are used [1,6]. On
the other hand, Al(III) compounds have good combining affin-
ity with fluoride and reduce the fluoride concentration down

* Corresponding author: Fax: +33 223 238 199.

E-mail address: Andre.Darchen@ensc-rennes.fr (A. Darchen).
0008-6223/$ - see front matter  2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2009.09.034
334 CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3
to 2 mg L1 [4]. However, the precipitation method leads to an
increase of total dissolved solids in treated water, resulting in
a supplementary difficulty in eliminating residual chemicals
in excess [3,13]. Furthermore, a high concentration of alumi-
num in the treated water may create a danger to human
health such as Alzheimer’s disease [3,14]. Several adsorbents
have been used for removal of fluoride from water such as
amorphous alumina [15], low-cost adsorbent [16–18], metal
oxides [19,20], activated carbon [21] and especially the acti-
vated alumina [22] which is very effective and selective. How-
ever, frequent regeneration with aluminum sulfate is needed
because of its low adsorption capacity at neutral pH [13],
which results in increased difficulty for operation. Moreover,
alumina remains expensive for under developed countries
and its optimal adsorption for fluoride in acidic pH range lim-
its its use to treat drinking water [15]. Therefore, it is desirable
to seek a low-cost adsorbent material which is effective for
defluoridation at a pH value of around 7.
Due to its high electronegativity and small ionic size, the
fluoride ion is classified as a hard base, which has a strong
affinity towards metal ions including Al(III), Fe(III), La(III)
[13]. Dispersing a mixture of these metals in a protective ma-
trix would limit the total dissolved solids in treated water, and
at the same time would provide high fluoride adsorption
capacity. Porous carbons could be a good candidate such as
matrix protective. Because of their excellent properties such
as large surface area and continuous porosity, porous carbons
are used as adsorbent or metal-support [23–26]. The carbon
support provides high dispersion and inhibits sintering of
metals within its pore structure. Generally, carbon-supported
metals are prepared by impregnating carbon with the corre-
sponding metal salt solutions followed by heat treatment.
Preparation of adsorbents such as carbon-supported alumi-
num or iron is scarce in the literature [24,25]. The activated
carbon and carbon nanotubes (CNTs) are often used as a
source of carbon, which becomes quite expensive for an
industrial production. However, there are very little studies
on the use of wood instead of the activated carbon as the sup-
port or matrix [26], whereas wood is largely available, inex-
pensive and may lead to mesoporous activated carbons [27].
In this paper, we present a new approach of preparation of
new adsorbents by impregnation of wood with aluminum
and iron salts followed by a carbonization. The adsorption
of fluoride ions by these charcoals that contain aluminum
oxide adsorbents is assessed.
2. Experimental
2.1. Material preparation
Scandinavia spruce wood with a density of 0.37 was used for
the preparation of charcoals. The modified carbons were pre-
pared according to the following procedure. The wood sam-
ples of parallelepipedic form (3.8 · 3.3 · 1.3 cm) were dried
beforehand in an oven at 105 C for 24 h, and they were
impregnated with a boiling mixture of 1 M AlCl3 and 1 M FeCl3
aqueous solution for 90 min. After the heating, the samples
remained immersed in the aqueous solution for 2 h. The
impregnated woods were removed from the solution, then
dried in the open air for 2 h and finally dried in an oven at
105 C for 24 h. These dried impregnated wood samples were
placed in a steel dish in a muffle furnace and heated to 500 C,
650 C or 900 C for 1 h. It was found advantageous to restrict
the access of air during carbonization. The resulting samples
were ground to the form of particles less than 200 lm and
stored for further characterization. Samples of aluminum–
iron dispersed in carbon are denoted AlFexxx/C where xxx
refers to the carbonization temperature.
2.2. Characterization techniques
Scanning electron microscopy (SEM) images and energy dis-
persive analysis of X-rays (EDAX) were performed at the Cen-
ter of Scanning Electron Microscopy of Rennes University on
Jeol JSM 6400 equipment operating at an accelerating voltage
of 20 kV.
X-ray powder diffraction (XRD) data were collected at room
temperature with a Siemens diffractometer operating with
the parafocusing Bragg–Brentano geometry, using Cu Ka1,2
radiation (k = 1.5406/1.5444** Å) and equipped with a dif-
fracted beam graphite monochromator to limit fluorescence
due to the iron content in the samples. The diffraction pat-
terns were collected at 40 kV and 30 mA, over the angular
range 10–80 (2h) with a counting time of 14s step1 and a step
length of 0.02 (2h).
The pHPZC was measured as follows: 100 mg of AlFexxx/C
samples were contacted with 50 mL of 0.05 M NaNO3 deaer-
ated solution with a pH range of 3–11. Blanks with no carbon
were also run in the same initial pH values. After shaking for
24 h at 250 rpm at room temperature, the final pH of the solu-
tion was measured and was plotted against the initial pH. The
pHPZC was determined as the pH of NaNO3 solution that did
not change after the contact with the samples.
The concentration of total acidic and basic groups on Al-
Fexxx/C surface was determined from NaOH and HCl uptake,
respectively. 100 mg of AlFexxx/C samples were placed in
50 mL of 0.05 M NaOH or 0.05 M HCl. Sealed beakers that con-
tained samples or blanks were shaken for 24 h and filtered.
Then 10 mL of each filtrate were titrated with 0.05 M of either
NaOH or HCl solution. The difference between the NaOH or
HCl consumption by the blank and samples was calculated
and translated to the equivalent acid or base content per
gram of charcoal.
The chemical stability of the inorganic content of the char-
coals was studied by measuring the release of Al and Fe in
solution. The concentration of dissolved Al and Fe were mea-
sured with a spectrophotometer (Hach) using Aluver 3 and
Ferover reagents for Al and Fe, respectively. The inorganic
content of the charcoals was calculated from the ash weight
after an exhaustive combustion of the charcoals in an oven
at 650 C for 24 h.
2.3. Fluoride adsorption experiments
A 1000 mg L1 F stock solution was prepared by dissolved
1.1050 g of NaF (analytical grade) in 500 mL of deionized
water. All the solutions for fluoride removal experiment and
analysis were prepared by appropriate dilution from the stock
solution. The defluoridation experiments were carried out by
 CARBON 4 8 (2 0 1 0) 3 3 3–34 3
a batch method at room temperature (28 ± 2 C). The pH of the
solution was adjusted with HCl or NaOH. 25 mL of fluoride
solution and 0.1 g of AlFexxx/C were placed in a 50 mL beaker.
Sealed beakers were shaken at 300 rpm for 24 h for equilib-
rium. Then the adsorbent particles were separated immedi-
ately by filtration. The pH of the filtrate was measured, and
the concentration of the residual fluoride was determined
thanks to a fluoride ion selective electrode (Orion, number
9609, USA) connected to an ion meter (Orion, model 290A,
USA). The fluoride samples were diluted to 1:1 with a com-
mercial buffer (TISAB II, Orion). The specific amount of fluo-
ride adsorbed was calculated from Eq. (1) where Qe is the
equilibrium adsorption capacity (mg g1); C0 and Ce are
respectively the initial and the equilibrium fluoride concen-
trations (mg L1), V is the volume of the aqueous solution
(L), and W is the mass (g) of the adsorbent.
ðC0  Ce Þ
Qe ¼ V ð1Þ
W
The adsorption isotherm at pH 6.90 ± 0.10 was studied by
varying the initial fluoride concentration from 2 to 50 mg L1.
The effect of equilibrium pH on F adsorption was investi-
gated by adjusting the pH of the solution to 3–10 under an ini-
tial F concentration of 10 mg L1. A kinetic study at
pH 6.90 ± 0.10 was carried out at different time intervals with
initial fluoride concentration of 10 mg L1. In order to investi-
gate the selectivity of carbon-supported aluminum and iron
oxide charcoals for fluoride ions, the effect of some compet-
2 3 2+
ing ions (Cl, NO 
3 , HCO3 , SO4 , PO4 , Ca ) were performed
1
with 10 and 100 mg L solutions at pH 6.90 ± 0.10 during
24 h. Calcium chloride, sodium nitrate, sulfate, phosphate,
bicarbonate or chloride were used with 10 mg L1 of fluoride
concentration.
3. Results and discussion
3.1. Working hypothesis
Fe3+ and Al3+ ions have good affinity with fluoride anions, be-
cause of the hard-hard combination, in agreement with the
soft-hard acid base (SHAB) concept. Al(III) is frequently used
in defluoridation. Fe(III) has been scarcely applied in defluori-
dation [6,28,29]. To our knowledge, the dispersion of Al(III)
and Fe(III) into charcoals has never been investigated. We fol-
lowed the hypothesis that a nano-scale dispersion of the inor-
ganic adsorbent into the charcoal may be an interesting way
to an efficient fluoride scavenger. In this paper, we describe
the fact that charcoals that contain mixed Al and Fe oxides
might be good adsorbents for scavenging fluoride from the
aqueous phase. The results showed indeed that charcoals
that contained both Al and Fe oxides were more effective
for removal of fluoride ions from aqueous water than carbons
that contained Al oxide alone or Fe oxides alone. Therefore,
adsorbents with Al and Fe binary oxides were used for further
detailed experimental investigations.
3.2. Impregnation and carbonization of wood
Our studies have shown that impregnation depends on the
nature and the thickness of the wood and the acidity of the
335
solution. During the immersion of wood in the salt solutions,
the air of the wood structure was expelled during the heating
and replaced by the solution during the cooling. The colora-
tion of the solution was observed and attributed to the extrac-
tion of hydrolysable wood tannins [30]. The amount of
hydrolysable tannins was difficult to estimate because it
was pH dependent. The weight loss of dried wood samples
after impregnation in water or 0.02 M HCl was found to be
1.3% or 11.6%, respectively. The weight of solution incorpo-
rated into the wood averaged 90% of the dry wood weight.
By neglecting the weight loss due to the dissolution of wood
tannins, we can calculate the amounts of Al and Fe incorpo-
rated into the wood samples. Related to one gram of the ini-
tial dry wood, the calculated amount is 1.8 mmol of metal,
0.9 mmol of each metal. These amounts were calculated for
charcoals and they are given in Table 1 as a mol number of
metal in one gram of charcoal. After an exhaustive combus-
tion of charcoals, we calculated their carbon contents (Table
1) by assuming that inorganic matter of ashes was already
present inside the initial charcoals. With an equal mixture
of Al2O3 and Fe2O3 in these ashes, we can calculate 2–
2.3 mmol of metal in one gram of dry wood. This result corre-
sponds with the precedent calculated value of 1.8 mmol.
From the results of Table 1, it appears that when the wood
carbonization temperature increased, the carbon content in
the adsorbent decreased, while the inorganic matter content
increased. This was due to the fact that increasing carboniza-
tion temperature enhanced the combustion of cellulose,
hemicelluloses and lignin which are the main constituents
of wood. In fact, during the heat treatment, separately car-
bonization, there was also an activation treatment due to
the fact that the atmosphere was not completely inert. How-
ever, carbon remains the major constituent of the synthe-
sized charcoals. The yields observed for the preparation of
AlFexxx/C are better than those of charcoal. This may be
attributed to a flame retardant effect of alumina. The carbon-
ization of impregnated wood leads to charcoals with a high
content of inorganic compounds.
3.3. Characterizations of carbons that contained Al and Fe
oxides
3.3.1. Scanning electron microscopy
As can be seen from Fig. 1, the matrix carbon has the ex-
pected porous structure (Fig. 1a). An irregular repartition of
particles on the surface of carbon is observed. These particles
were attributed to the formation of an inorganic layer on the
charcoals during the wood carbonization. Since wood has an
anisotropic structure, it is not surprising that the wood
impregnation was not a uniform step leading to a similar
repartition of these particles on the charcoals. The number
of these particles on the surface of carbon is more significant
at a higher temperature. At 900 C, the inorganic particles clog
up the pores of carbon (Fig. 1d).
The EDAX spectrum of Fig. 2 shows the presence of Al, Fe,
O and Cl in the charcoals. The atomic percentages of these
elements, determined from EDAX analysis are gathered in Ta-
ble 2. These results confirm the increase of the inorganic ele-
ments with the carbonization temperature. Whatever the
temperature, it can be seen that the iron content was greater
336 CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3

Table 1 – Chemical contents of inorganic matter and carbon calculated after an exhaustive combustion of charcoals.
Charcoals Charcoal yield (%)a Carbon content (%)b Inorganic content (%)b Inorganic content of charcoals (mmol g1)c

Charcoal 21 99 <1 0
AlFe500/C 47 72.4 27.6 3.8
AlFe650/C 45 71.1 28.9 4.0
AlFe900/C 40 62.1 37.9 5.0
a
Yields are calculated from the initial weight of wood samples.
b
These contents are calculated from ash weights after exhaustive combustion.
c
These contents are calculated from the weight gains of impregnated wood sample.

Fig. 1 – SEM images of (a) wood charcoal (b) AlFe500/C (c) AlFe650/C and (d) AlFe900/C.

Fig. 2 – EDAX spectrum of carbon that contains Al and Fe oxides.

than the one of Al, whereas wood was impregnated with the combustion of carbon, we observed closer for the contents
same concentrations of AlCl3 and FeCl3 in solution. This of iron and aluminum in the ashes.
unexpected ratio of iron and aluminum may be in the range It is difficult to establish the chemical structure of the inor-
of error of the EDAX analysis. Indeed, after an exhaustive ganic compounds which were formed in charcoals, because of
 CARBON 4 8 (2 0 1 0) 3 3 3–34 3 337

Table 2 – Quantitative element distribution in atomic percentage determined from EDAX analysis of the charcoals.
Charcoals Atomic percentages of the elements in the charcoals (%)
Al Fe O Cl

AlFe500/C 1.62 2.49 11.62 1.93

AlFe650/C 2.27 2.75 13.37 1.74
AlFe900/C 2.68 3.82 18.20 0.96

the possible existence of some amorphous compounds in

charcoals and the complexity of the wood structure. But it ap-
pears that there would be mainly iron and aluminum oxides.
Indeed, the EDAX mapping of Fig. 3 shows an evidence of oxi-
des on the surface of carbon. From Fig. 3, aluminum oxide is
more highly dispersed inside the walls of the charcoal cellular
microstructure, while iron oxides are on the surface of these
walls. It is obvious that the residual Cl contents of charcoals
arose from the initial chloride salts AlCl3 and FeCl3. For
AlFe500/C, we can calculate more than 85% departure of Cl,
probably as HCl, during the drying and carbonization steps.
This Cl loss increased with the carbonization temperature.
For a carbonization of 900 C, the residual chloride was 4.9%
of the initial content of wood.
In fact, we are unsure of how important residual chloride
is for the fluoride removal. In order to obtain free chloride Fig. 4 – Powder XRD patterns of (a) AlFe500/C (b) AlFe650/C
charcoals, we carried out some experiments with nitrate solu- and (c) AlFe900/C.
tions instead of chloride solutions. The corresponding char-
coals were less efficient for the fluoride elimination from
aqueous solution. pared at 500 C is mainly amorphous. It is known that amor-
phous materials have good adsorbent properties due to their
3.3.2. X-ray diffraction high specific surface area and more active sites on their sur-
The powder X-ray diffraction patterns of the charcoals that faces [31]. When the carbonization temperature was in-
contain Al and Fe oxides are shown in Fig. 4. The charcoal pre- creased, we observed diffraction peaks attributed to

Fig. 3 – EDAX mapping of AlFe900/C.

338 CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3

groups such as carboxylic, phenolic or lactonic groups, and

(ii) the charcoal combustion contributes to the release of
acidic metal oxides on the charcoal surface. On the other
hand, the amounts of basic groups enhance with increasing
carbonization temperature. This may be due to basic sites of
increasing amounts of alumina on the charcoal surface. Over-
all, all the charcoals are acidic. pHPZC measurements support
this finding, as all AlFexxx/C charcoals appear to have a pos-
itively charged surface at a pH of below 5. From the results of
Table 3, pHPZC increases with the carbonization temperature.
The presence of Al and Fe oxides remarkably shifted the car-
bon pHPZC to smaller values. This is due to the amphoteric
character of the oxides of aluminum and iron, and in this case
they behave like acids.

Fig. 5 – Powder XRD pattern of product after exhaustive

3.4. Fluoride removal
combustion of carbon.

3.4.1. Adsorption kinetics

crystalline iron oxides and there was no peak due to crystal-
line aluminum oxides. The iron oxides in the charcoals were
hematite, maghemite and magnetite. The peaks of aluminum
in Fig. 4 were due to the sample support made in aluminum. It
can be seen that iron oxide peaks became better defined and
they had higher intensities when the carbonization tempera-
ture increased. From XRD results, any chloride compounds
were not observed while Cl was revealed by EDAX analysis.
It can be concluded that there are also amorphous chloride
compounds (probably as AlCl3 or FeCl3) in the charcoals.
Defluoridation experiments were performed in order to deter-
mine the rate of fluoride removal from an aqueous solution.
Fig. 6a illustrates the fluoride sorption by AlFexxx/C char-
coals. For all the substrates, fluoride uptake was fast during
the first 4 h and pseudo-equilibrium was reached in 8 h. After
2.4
2.2
a
2.0
Adsorbed fluoride (mg g-1 )

1.8
The powder X-ray diffraction of material after an exhaus- 1.6 charcoal
tive combustion of carbon is shown in Fig. 5. The diffraction 1.4 AlFe500/C
peaks are attributed to crystalline aluminum and iron oxides. 1.2 AlFe650/C
The peaks relating to Al2O3 are broader and indicate the small AlFe900/C
1.0

size of their particles. These XRD results prove that alumina is 0.8

continually present in the charcoals, even if they are under an 0.6

amorphous form. 0.4

0.2

3.3.3. Potentiometric titration 0.0

0 4 8 12 16 20 24
The pHPZC and the amounts of surface charges were assessed Time (h)
by potentiometric titration. The pHPZC of a solid is an impor-
tant characteristic that determines the pH at which the adsor- 2.5

bent surface has net electrical neutrality. For pH < pHPZC, the b Phase 2
Phase 3
Amount of fluoride adsorbed (mg g )
-1

adsorbent surface charge is positive and at pH > pHPZC the 2.0

surface charge is negative. A given value of pHPZC is related
to the presence of acidic and basic groups on the adsorbent Phase 3
1.5
surfaces. From the results of Table 3, all three charcoals ap-
pear to have more acidic groups than basic ones on their sur- Phase 2
AlFe500/C
faces. The amounts of acidic groups decrease with increasing 1.0
Phase 1 AlFe650/C
carbonization temperature. This is due to the dual fact (i) that AlFe900/C
increasing temperature enhances the loss of the main acidic 0.5

0.0
5 10 15 20 25 30 35 40
Table 3 – Values of surface acidity and pHPZC of AlFexxx/C 1/2 1/2
t (min)
charcoals.
Charcoals Acidic (mmol g1) Basic (mmol g1) pHPZC Fig. 6 – (a) Kinetics of fluoride adsorption using adsorbents
obtained at different carbonization temperature. (b) Intra-
Charcoal 1.04 0.21 6.4
particle diffusion kinetic plot for fluoride onto charcoals that
AlFe500/C 7.53 0.06 2.8
contained aluminum and iron oxides. (Initial concentration:
AlFe650/C 6.83 0.12 4.2
AlFe900/C 6.44 0.15 4.4 10 mg L1, pH 7, agitation rate: 300 rpm, adsorbent dose:
4 g L1, temperature: 28 ± 2 C.)
 CARBON 4 8 (2 0 1 0) 3 3 3–34 3
the equilibrium period, the amount of adsorbed fluoride did
not significantly change with time. The rapid fluoride uptake
is certainly due to the availability of active sites on charcoal
surfaces at the beginning of the experiments, whereas the ob-
served plateau corresponds to a slow rate of adsorption dur-
ing which there is a saturation of active sites. The charcoal
efficiency for fluoride uptake follows the order AlFe650/
C > AlFe900/C > AlFe500/C > charcoal. Thus, this efficiency
may be attributed to the double effect of charcoal porosity
and the amount of Al and Fe oxides in the charcoal. Daifullah
et al. [32] also observe that the KMnO4-modified activated car-
bon obtained by carbonization at 650 C of rice straw has the
highest capacity of fluoride uptake. Therefore, AlFe650/C was
used in the further experiments.
To understand the mechanism of fluoride adsorption onto
the surface of AlFe650/C, the Lagergren first-order and the
pseudo-second order kinetic models were used to analyze
the kinetic experimental data. The pseudo-first-order rate
equation of Lagergren is one of the most widely used for the
sorption from a liquid phase [33]. It may be represented by
the relationship (2) where Qe and Qt are the adsorption capac-
ity (mg g1) at equilibrium and time t, respectively. K1 (min1)
is the rate constant of pseudo-first-order adsorption. The
pseudo-second-order model describes the sorption of some
metal ions onto the adsorbent. This model assumed that
the adsorption follows the Langmuir equation [34]. The ki-
netic rate equation can be written by relationship (3) where
K2 (g mg1 min1) is the equilibrium rate constant of the sec-
ond order sorption.
dQ t
¼ K1 ðQ e  Q t Þ ð2Þ
dt
dQ t
¼ K2 ðQ e  Q t Þ2 ð3Þ
dt
The adsorption kinetic model parameters obtained from the
above equations are given in Table 4. From these results, it
is evident that kinetic data of the fluoride adsorption onto
charcoals containing aluminum fitted well with pseudo-sec-
ond order equation. This suggests that a chemisorption step
may be rate determining in the fluoride adsorption process.
To evaluate the rate-limiting of the fluoride adsorption
onto charcoals, the possibility of intra-particle diffusion was
explored since the particles were vigorously agitated during
the experiment. The data were analyzed by the intra-particle
diffusion Weber–Morris model (Eq. (4)) [35]:
Q t ¼ Kid t1=2 þ C ð4Þ
where Kid is the intra-particle (pore) diffusion rate constant
(mg g1 min0.5) and C is the intercept that gives an idea about
the thickness of the boundary layer. The larger the value of C,
the greater the boundary-layer effect is. From plot shown in
Fig. 6b, the data points could be connected by three straight
lines: the first linear portion represents the boundary layer
Table 4 – Comparison of the Lagergren and pseudo-second order kinetics parameters for fluoride adsorption onto AlFe650/C.
Experimental Qe (mg g1) Lagergren fisrt-order model
K1 (min1) Qe (mg g1)
2.31 0.25 · 102 0.63
339
2.4
2.3
Fluoride adsorption capacity (mg g -1)
2.2
2.1
2.0
1.9
1.8
1.7
2 3 4 5 6 7 8 9 10 11
pH
Fig. 7 – Effect of pH on the fluoride adsorption capacity using
AlFe650/C. (initial concentration: 10 mg L1, time: 24 h,
agitation rate: 300 rpm, adsorbent dose: 4 g L1,
temperature: 28 ± 2 C.)
diffusion or macropore diffusion, the second linear portion
attributes to intra-particle diffusion and the third plateau por-
tion is the equilibrium stage. The extrapolations of the first
linear portions of the curves do not pass through the origin;
this indicates that mechanism of fluoride removal onto char-
coals that contained aluminum and iron oxides is complex.
The surface adsorption as well as intra-particle diffusion con-
tributes to the rate-determining step. As evident from Fig. 6b,
the intercept C follows the sequence: AlFe650/C > AlFe500/
C > AlFe900/C. Thus, the boundary-layer thickness follows
the same sequence order for charcoals.
3.4.2. Effect of pH
The pH of the medium is one of the important parameters
that significantly affect the extent of fluoride adsorption.
Fig. 7 shows the effect of pH on the fluoride uptake using
AlFe650/C adsorbent. As shown in Fig. 7, the amount of ad-
sorbed fluoride enhances slightly with increasing pH and it
attains a maximum at pH 7, thereafter it decreases. The max-
imum fluoride adsorption capacity was found to be 2.3 mg g1
corresponding to a fluoride removal yield of 92% with a fluo-
ride residual concentration of 0.76 mg L1.
In the literature, maximum fluoride adsorption was ob-
tained in the pH range of 4.0–10.0 by using iron(III)–alumi-
num(III) mixed oxide [6], whereas it was obtained at pH 7
using activated alumina [22]. It was evident from our results
that the charcoals exhibited significantly fluoride removal in
the acidic pH range (the fluoride sorption capacity is
2.19 mg g1 at pH 3, not far from the maximal value
2.3 mg g1 at pH 7). This is due to the fact that for pH < pHpcz
(4.2), the adsorbent is positive and attracts the negative fluo-
ride ions. It can also be explained by a fluoride complexation
Pseudo-second order model
R2 K2 (g mg1 min1) Qe (mg g1) R2
0.7496 6.58 · 102 2.31 0.9994
340 CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3

of dissolved metals [13,36]. Indeed, the dissolved metal ions
may complex fluoride ions in the solution and increase the
capacity of removing fluoride. Maximum adsorption was ex-
pected to occur at pH < pHpcz, however it occurs at pH 7. Thus,
the electrostatic attraction between charcoal surfaces and
fluoride ions plays a small part in the fluoride adsorption.
The minimal fluoride removal capacity in acidic pH may be
attributed to the formation of HF, which would reduce the
coulombic attraction between fluoride and the adsorbent sur-
face [37]. Indeed, in our work, the pH of filtrated solutions
after treatment was in the acidic range, which can be attrib-
uted to HF or HCl arising from hydrolysis of residual metal
chloride. The reduction in fluoride removal capacity in pH
above pHpcz can be attributed to the electrostatic repulsion
from the negatively charged adsorbent surface or to a compe-
tition between hydroxide and fluoride anions.
The mechanism involved in the fluoride adsorption pro-
cess using AlFe650/C as adsorbent can be described by Eqs.
(5)–(7) where M is Al(III) or Fe(III) bound to carbon.
 sorb fluoride from aqueous solution is evaluated from the
M-ClðsÞ þ H2 O ! M-OHðSÞ þ Hþ
ðaqÞ þ ClðaqÞ
 adsorption isotherm curve. Fig. 9 shows that as fluoride con-
M-OHðsÞ þ F þ 
ðaqÞ ! M    F þ OH
M-OHðsÞ þ OH þ 
ðaqÞ þ NaðaqÞ þ FðaqÞ !
 þ
M-O    Na    F þ H2 O ð7Þ
The fluoride adsorption mechanism is dominated by Eq. (5)
whatever the range of pH which are well consistent with
the observe decrease of the equilibrium solution pH. The
slightly increase in adsorption capacity when increasing pH
from acid to neutral pH can be described by Eq. (6). That is
supported by the fact that the equilibrium pH increased
slightly with increasing pH. The observed decrease in adsorp-
tion capacity for pH > 7 can be described by Eq. (7). The latter
equation agrees well with the decrease of the equilibrium
solution pH.
3.4.3. Effect of some other ions on defluoridation
The fluoride-contaminated drinking water may contain sev-
eral other anions which may compete with the sorption of
fluoride. The adsorption studies were carried out in the pres-
ence of 10 and 100 mg L1 salt solutions of chloride, nitrate,
bicarbonate, sulfate, phosphate and calcium independently
at an initial fluoride concentration of 10 mg L1. The effects
of these co-existing ions on fluoride removal are shown in
Fig. 8. From the results, the chloride ions present a significant
influence on the fluoride removal. The ions at a concentration
of 100 mg L1 did not significantly affect the removal of fluo-
ride ions, except for chloride ions. This strong competition
between chloride and fluoride ions can be explained by the
radius of the chloride ion (3.32 A) which is smaller than that
of fluoride ions (3.52 A). Sulfate, bicarbonate, nitrate, phos-
phate and calcium ions did not have a significant influence
on the defluoridation. Rather, calcium ions increase the rate
of fluoride removal, which can be explained by the complex-
ation of Ca2+ with the dissolved fluoride in the form of CaF2.
Overall, it appears that increasing the concentration of co-
existing ions decreases the effect of the latter on the defluori-
dation yield.
3.4.4. Effect of fluoride concentration
The ability of Al and Fe oxides dispersed in charcoals to ad-
ð5Þ
ð6Þ
centration increases, adsorbed amounts of fluoride increase

too and then tend to reach a plateau. This plateau occurs
due to the saturation of active sites on the adsorbent surfaces.
At low fluoride concentrations, the ratio of surface active sites
to fluoride ions is high. Hence fluoride ions can occupy the ac-
tives sites of the carbon surface. From Fig. 9, at a high fluoride
concentration, the efficiency of the adsorbents for fluoride
uptake follows the order AlFe650/C > AlFe500/C > AlFe900/
C > charcoal. Compared to the kinetic studies, the efficiency
order between AlFe500/C and AlFe900/C is changed. This
can be explained by the fact that the pores of carbon in
AlFe900/C are not accessible to fluoride ions in large concen-
tration, because they are blocked by the inorganic matter, as
already suggested from the SEM observations.
3.4.5. Equilibrium adsorption isotherms
Analysis of isotherm data is very important for predicting the
adsorption capacity of adsorbent, which is one of the main
parameters required for the design of an adsorption system.

11
100
-1
10 mg L 10 Charcoal
95
100 mg L
-1
9
AlFe500/C
AlFe650/C
90 8
Fluoride Uptake (mg g )

AlFe900/C
-1
Defluoridation yield (%)

7
85
6
80
5

75 4

3
70
2
65
1

60 0
Blank Nitrate Sulfate Bicarbonate Phosphate Chloride Ca2+ 0 5 10 15 20 25 30 35 40
-1
Co-existing ions Equilibrium concentration (mg L )

Fig. 8 – Effect of co-existing ions on the fluoride adsorption Fig. 9 – Effect of initial concentration of fluoride on its
capacity using AlFe650/C. (fluoride initial concentration: removal capacity using adsorbents obtained at different
10 mg L1, pH 7; time: 24 h, agitation rate: 300 rpm, carbonization temperatures (time: 24 h, pH 7, agitation rate:
adsorbent dose: 4 g L1, temperature: 28 ± 2 C.) 300 rpm, adsorbent dose: 4 g L1, temperature: 28 ± 2 C).
 CARBON 4 8 (2 0 1 0) 3 3 3–34 3 341

Table 5 – Freundlich and Langmuir parameters for adsorption isotherms of fluoride onto adsorbents.
Adsorbents Freundlich Langmuir
n KF R2 Qmax (mg g1) KL (L mg1) R2

Charcoal 1.40 0.286 0.9907 3.77 0.073 0.9981

AF500/C 1.193 0.205 0.9927 7.92 0.023 0.9995
AF650/C 1.139 0.737 0.9944 13.64 0.058 0.9987
AF900/C 2.043 1.130 0.9033 5.67 0.26 0.9890

In this work, the Langmuir and Freundlich isotherm models
were used.
The Langmuir adsorption model is based on the assump-
tion of surface homogeneity where all adsorption sites are
equally available, and with a monolayer surface coverage
without interaction between adsorbed species. The mathe-
matical description of this model is given by Eq. (8) where
KL is Langmuir constant (L mg1) and Qm is maximum fluoride
adsorbed (mg g1 adsorbent).
The Freundlich isotherm is based on the multilayer
adsorption of an adsorbate onto the heterogeneous adsorbent
surface. This empirical equation takes the form (9) where KF is
the Freundlich constant that indicates the adsorption capac-
ity and n is a constant that expresses the adsorption intensity.
KL Q m Ce
Qe ¼ ð8Þ
1 þ KL Ce
Q e ¼ KF Ce1=n ð9Þ
The isotherm parameters obtained from these models are gi-
ven in Table 5. The applicability of a particular isotherm mod-
el is judged from the regression coefficient value (R2). From
the results of Table 4, the Langmuir model fitted well with
the isotherm data of fluoride adsorption onto elaborated
adsorbents. It is observed that the dispersion of Al and Fe in
charcoal improves (3.77 times) the maximum fluoride adsorp-
tion capacity. AlFe650/C charcoal presents the best adsorption
capacity for fluoride (13.64 mg g1). Thus the presence of
acidic sites and Al–Fe on the carbon surface plays an impor-
tant role in the efficiency of the elimination of fluoride ions
from aqueous solution. For all the sorbents, the n values are
higher than unity, which indicates a favorable adsorption. A
comparison has been made between adsorbents synthesized
in this work and previously reported adsorbents for fluoride
removal. As can be seen from Table 6, carbon-supported alu-
minum and iron oxides as adsorbent are better than many
other adsorbents in terms of defluoridation capacity. The high
fluoride adsorption capacity obtained in this work is mainly
due to the dispersion of Al and Fe in the charcoal. These both
elements have a strong affinity with fluoride ions. Therefore
synthesized charcoals that are easy to implement are effec-
tive adsorbents for removing fluoride. Moreover, these char-
coals are magnetic, so they could be used in industry for the
removal of fluoride ions from heterogeneous environments.
3.4.6. Chemical stability of Al and Fe adsorbents dispersed in
charcoal
The chemical stability of the adsorbents was examined by
measuring the concentrations of dissolved metals in solution
after adsorption. Fig. 10 gives the amounts of dissolved metal
ions under various pH values for AlFe650/C adsorbent. These
results show that pH has a significant influence on the chem-
ical stability of AlFe650/C adsorbent. It can be seen that the
acid pH promotes the Al and Fe dissolution. This dissolution
leads to a decrease of fluoride adsorption capacity, because
dissolved metals form complexes with fluoride anions. For
pH above 6, the amounts of dissolved Al are lower and slightly
constant (0.67 mg L1). The amounts of dissolved Fe are high
in the pH range of 3–5 and 8–10. However, the residual con-
centration of Fe is higher than that of Al. Overall, at pH 7
where the maximal fluoride adsorption occurs, the Al and
Fe dissolutions are relatively low. The values of residual con-
centration of Al and Fe are 0.67 mg L1 and 1.8 mg L1, respec-
tively. This Al residual concentration of 0.67 mg L1 is greater
than the standard set by WHO for drinking water (0.2 mg L1).
Further treatment is therefore necessary for the elimination
of excess dissolved aluminum.

Table 6 – Comparison between various adsorbents used for fluoride removal.

Adsorbent Adsorption capacity (mg g1) Reference
3+
La impregnated cross-linked gelatin 21.30 [38]
Iron–aluminum mixed oxide 17.73 [6]
Quick lime 16.67 [9]
KMnO4-modified activated carbon 15.90 [32]
AlFe650/C 13.64 This work
Magnesia-amended activated alumina 10.12 [36]
Iron–tin bimetal 10.05 [39]
Activated alumina 2.41 [40]
Activated carbon (Al-C300) 1.10 [41]
Laterite 0.85 [18]
Calcite 0.39 [1]
342
6 2.0
Residual concentration of Al (mg L )
-1
5 2.41 times higher than that of AlFe900/C and 3.62 times
4
3 oxides therefore have a significant amount of potential by
2 1.0
1
0
2 4 6 8
pH
Fig. 10 – Solubility of AlFe650/C adsorbent in fluoride
solution with pH. Adsorption conditions: (initial
concentration: 10 mg L1, time: 24 h, agitation rate: 300 rpm,
adsorbent dose: 4 g L1, temperature: 28 ± 2 C).
4. Conclusions
In this study, Al and Fe dispersed in charcoal as adsorbents
were prepared by the impregnation of wood with aluminum
and iron chloride salts followed by a carbonization. These
low-cost adsorbents have shown a good efficiency in fluoride
removal from aqueous solution and can be useful for waste
management and environmental protection purposes. The
main conclusions drawn from the present study are:
• Adsorbents synthesized were mainly acidic with
pHpcz < 4.5.
• Inorganic particles were present on the surface of carbon
as shown by SEM images and EDAX spectra. Crystallized
iron oxides were observed by XRD. No peaks related to alu-
minum oxides were observed. Overall, the prepared char-
coals were amorphous.
• Fluoride adsorption onto adsorbents was rapid during the
first 4 h and thereafter it slowly reached pseudo-equilib-
rium in 8 h. AlFe650/C adsorbent presented the best effi-
ciency for fluoride adsorption. The adsorption kinetic
followed the pseudo-second order model, which indicates
that fluoride removal might be a chemisorption process.
• The pH solution was an important parameter affecting
fluoride adsorption. An optimum fluoride uptake by
AlFe650/C was observed at pH 7. At that pH, the fluoride
uptake was 2.3 mg g1 corresponding to a fluoride removal
yield of 92% and a fluoride residual concentration of
0.76 mg L1.
• The chloride and phosphate ions at lower concentrations
have an adverse influence on fluoride removal by
AlFe650/C, while nitrate, sulfate, phosphate, bicarbonate
and calcium ions did not affect the defluoridation capacity.
• The amounts of dissolved Al and Fe ions in the solution
varied with the operating pH. Dissolution of Al and Fe
was very pronounced in acidic pH range. At pH 7, the
amounts of dissolved Al and Fe were respectively 0.67
and 1.8 mg L1.
• For all the charcoals, the adsorption isotherms were better
fitted by the Langmuir model. The predicted fluoride
CARBON 4 8 ( 2 0 1 0 ) 3 3 3 –3 4 3
adsorption capacity of AlFe650/C was 13.64 mg g1, which
was found to be 1.72 times higher than that of AlFe500/C,
Residual concentration of Fe (mg L )
-1
1.8
1.6 higher than that of metal free charcoal.
1.4
The elaborated charcoals with mixed aluminum and iron
1.2
promising environmental materials for fluoride removal from
water.
0.8
0.6 Acknowledgements
10
The authors thank CNRS (UMR 6226) and the French Ministry
of National Education for positions and support. This work
was supported in part by the French Ministry of Foreign Af-
fairs. One of us (E. Tchomgui) wishes to express its gratitude
to SCAC (Service de Cooperation et d’Action Culturelle) of
French Embassy in Cameroon for a doctoral fellowship.
R E F E R E N C E S
[1] Fan X, Parker DJ, Smith MD. Adsorption kinetics of fluoride on
low cost materials. Water Res 2003;37(20):4929–37.
[2] Susheela AK. Fluorosis management programme in India.
Curr Sci 1999;77(10):1250–6.
[3] Meenakshi, Maheshwari RC. Fluoride in drinking water and
its removal. J Hazard Mater B 2006;137(1):456–63.
[4] Shen F, Chen X, Gao P, Chen G. Electrochemical removal of
fluoride ions from industrial wastewater. Chem Eng Sci
2003;58(3–6):987–93.
[5] Wang SX, Wang ZH, Cheng XT, Li J, Sang ZP, Zhang XD, et al.
Arsenic and fluoride exposure in drinking water: children’s IQ
and growth in Shanyin county, Shanxi province, China.
Environ Health Perspect 2007;115(4):643–7.
[6] Biswas K, Saha SK, Ghosh UC. Adsorption of fluoride from
aqueous solution by a synthetic iron(III)–aluminum(III) mixed
oxide. Ind Eng Chem Res 2007;46(16):5346–56.
[7] Hu CY, Lo SL, Kuan WH, Lee YD. Treatment of high fluoride-
content wastewater by continuous electrocoagulation–
flotation system with bipolar aluminum electrodes. Sep Purif
Technol 2008;60(1):1–5.
[8] Sundaram CS, Viswanathan N, Meenakshi S. Defluoridation
chemistry of synthetic hydroxyapatite at nano scale:
equilibrium and kinetic studies. J Hazard Mater 2008;155(1–
2):206–15.
[9] Islam M, Patel RK. Evaluation of removal efficiency of fluoride
from aqueous solution using quick lime. J Hazard Mater
2007;143(1–2):303–10.
[10] Yang CL, Dluhy R. Electrochemical generation of aluminum
sorbent for fluoride adsorption. J Hazard Mater
2002;112(3):239–52.
[11] Tripathy SS, Bersillon JL, Gopal K. Removal of fluoride from
drinking water by adsorption onto alum-impregnated
activated alumina. Sep Purif Technol 2006;50(3):310–7.
[12] Hichour M, Persin F, Sandeaux J, Gavach C. Fluoride removal
from waters by Donnan dialysis. Sep Purif Technol
2000;18(1):1–11.
[13] Wu X, Zhang Y, Dou X, Yang M. Fluoride removal
performance of a novel Fe–Al–Ce trimetal oxide adsorbent.
Chemosphere 2007;69(11):1758–64.
[14] Walton JR. Aluminum in hippocampal neurons from humans
with Alzheimer’s disease. Neurotoxicology 2006;27(3):385–94.
 CARBON 4 8 (2 0 1 0) 3 3 3–34 3
[15] Li YH, Wang S, Cao A, Zhao D, Zhang X, Xu C, et al.
Adsorption of fluoride from water by amorphous alumina
supported on carbon nanotubes. Chem Phys Lett 2001;
350(5–6):412–6.
[16] Kagne S, Jagtap S, Dhawade P, Kamble SP, Devotta S, Rayalu
SS. Hydrated cement: a promising adsorbent for the removal
of fluoride from aqueous solution. J Hazard Mater 2008;
154(1–3):88–95.
[17] Meenakshi S, Sundaram CS, Sukumar R. Enhanced fluoride
sorption by mechanochemically activated kaolinites.
J Hazard Mater 2008;153(1–2):164–72.
[18] Sakar M, Banerjee A, Pramanick PP, Sarkar AR. Use of laterite
for the removal of fluoride from contamined drinking water.
J Colloid Interf Sci 2006;302(2):432–41.
[19] Maliyekkal SM, Sharma AK, Philip L. Manganese-oxide-
coated alumina: a promising sorbent for defluoridation of
water. Water Res 2006;40(19):3497–506.
[20] Nagappa B, Chandrappa GT. Mesoporous nanocrystalline
magnesium oxide for environmental remediation.
Microporous Mesoporous Mater 2007;106(1–3):212–8.
[21] Abe I, Iwasaki S, Tokimoto T, Kawasaki N, Nakamura T,
Tanada S. Adsorption of fluoride ions onto carbonaceous
materials. J Colloid Interface Sci 2004;275(1):35–9.
[22] Ghorai S, Pant KK. Equilibrium, kinetics and breakthrough
studies for adsorption of fluoride on activated alumina. Sep
Purif Technol 2005;42(3):265–71.
[23] Hared AI, Dirion JL, Salvador S, Lacroix M, Rio S. Pyrolysis of
wood impregnated with phosphoric acid for the production
of activated carbon: kinetics and porosity development
studies. J Anal Appl Pyrol 2007;79(1–2):101–5.
[24] Weifang C, Parette R, Jiying Z, Cannon FS, Dempsey BA.
Arsenic removal by iron-modified activated carbon. Water
Res 2007;41(9):1851–8.
[25] Petit C, Bandosz TJ. Role of aluminium oxycations in
retention of ammonia on modified activated carbons. J Phys
Chem C 2007;111(44):16445–52.
[26] Sakata Y, Uddin MA, Muto A, Imaoka M. Carbon-supported
well-dispersed Cu–ZnO catalysts prepared from sawdust
impregnated with [Cu(NO3)2, Zn(NO3)2] solution: catalytic
activity in CO2 hydrogenation to methanol. Microporous
Mater 1997;9(3–4):183–7.
[27] Salame II, Bandosz TJ. Comparison of the surface features of
two wood-based activated carbons. Ind Eng Chem Res
2000;39(2):301–6.
343
[28] Dey S, Goswami S, Ghosh UC. Hydrous ferric oxide (HFO) – a
scavenger of fluoride from contaminated water. Water Air
Soil Pollut 2004;158(1):311–23.
[29] Eskandarpour SK, Bando Y, Ikuta H, Iwai K, Okido M, Asai S.
Creation of nanomagnetite aggregated iron oxide hydroxide
for magnetically removal of fluoride and phosphate from
wastewater. ISIJ Int 2007;47(4):558–62.
[30] Bru K, Blin J, Julbe A, Volle G. Pyrolysis of metal impregnated
biomass: an innovative catalytic way to produce gas fuel.
J Anal Appl Pyrol 2007;78(2):291–300.
[31] Zhang G, Qu J, Huijuan L, Liu R, Wu R. Preparation and
evaluation of a novel Fe–Mn binary oxide adsorbent for
effective arsenite removal. Water Res 2007;41(9):1921–8.
[32] Daifullah AAM, Yakout SM, Elreefy SA. Adsorption of fluoride
in aqueous solutions using KMnO4-modified activated carbon
derived from steam pyrolysis of rice straw. J Hazard Mater
2007;147(1–2):633–43.
[33] Lagergren S. Zur theorie der sogenannten adsorption gelöster
stoffe. K Sven Vetenskapsakad Handl 1898;24:174–5.
[34] Ho YS, Mckay G. The kinetics of sorption of divalent metal
ions onto sphagnum moss peat. Water Res
2000;34(3):735–42.
[35] Weber WJ, Morris JC. Kinetics of adsorption on carbon
solution. J Sanit Eng Div Am Soc Civ Eng 1963;89:31–59.
[36] Maliyekkal SM, Shukla S, Philip L, Nambi IM. Enhanced
fluoride removal from drinking water by magnesia-
amended activated alumina granules. Chem Eng J
2008;140(1–3):183–92.
[37] Raichur AM, Basu MJ. Adsorption of fluoride onto rare earth
oxides. Sep Purif Technol 2001;24(1–2):121–7.
[38] Zhou Y, Yu C, Shan Y. Adsorption of fluoride from aqueous
solution on La3+-impregnated cross-linked gelatin. Sep Purif
Technol 2004;36(2):89–94.
[39] Biswas K, Gupta K, Ghosh UC. Adsorption of fluoride by
hydrous iron(III)–tin(III) bimetal oxide from the aqueous
solutions. Chem Eng J 2009;149(1–3):196–206.
[40] Ghorai S, Pant KK. Investigations on the column performance
of fluoride adsorption by activated alumina in a fixed-bed.
Chem Eng J 2004;98(1–2):165–73.
[41] Ramos RL, Ovalle-Turrubiartes J, Sanchez-Castillo MA.
Adsorption of fluoride from aqueous solution on aluminum-
impregnated carbon. Carbon 1999;37(4):609–17.