You are on page 1of 15

Chemical Engineering Journal 281 (2015) 345–359

Contents lists available at ScienceDirect

Chemical Engineering Journal


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

Selective removal of Hg(II) from aqueous solution by functionalized


magnetic-macromolecular hybrid material
Khalid Z. Elwakeel a, Eric Guibal b,⇑
a
Environmental Science Department, Faculty of Science, Port-Said University, Port-Said, Egypt
b
Ecole des mines d’Alès, Centre des Matériaux des Mines d’Alès (C2MA), 6, avenue de Clavières, F-30319 Alès cedex, France

h i g h l i g h t s

 Magnetic hybrid material for mercury recovery from slightly acidic solutions.
 Diethylenetriamine grafting on chitosan for improved sorption performance.
 Pseudo-second order rate equation for modeling uptake kinetics.
 Selectivity for mercury recovery against other base metals.
 Mercury can be easily desorbed from sorbent for efficient recycling.

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

Article history: A new hybrid material was prepared using chitosan, glycidyl methacrylate and magnetite microparticles.
Received 17 April 2015 The concentration of amine groups on the sorbent was increased by grafting diethylenetriamine. These
Received in revised form 28 May 2015 materials were tested for the sorption of Hg(II) metal ions. These materials showed high affinity and
Accepted 29 May 2015
selectivity for Hg(II) uptake from aqueous solutions: maximum sorption capacity reached 2.6 mmol g1
Available online 18 June 2015
at pH 4.0. The influence of pH was tested on the selective separation of Hg(II) from a mixture of Hg(II),
Co(II), Cu(II), Fe(II), Ni(II), Zn(II) and Mg(II). The effect of counter anions was also studied at different
Keywords:
pH values. Sorption may occur by chelation of cationic mercury species on amino groups or by ion
Hg(II) sorption
Magnetic hybrid sorbent
exchange of chloroanionic mercury species on protonated amino groups. Uptake kinetics and sorption
Diethylenetriamine-grafted chitosan isotherms were modeled using the pseudo-second order rate equation and the Langmuir equation,
Uptake kinetics respectively. The distribution coefficient was obtained at different temperatures and the thermodynamic
Sorption isotherms parameters have been calculated: the sorption is endothermic, spontaneous and contributes to increase
Thermodynamics the entropy of the system. Potassium iodide was used for Hg(II) desorption from loaded sorbents: desorp-
tion yields 99%, and the sorbent could be efficiently recycled for a minimum of three sorption/desorption
cycles.
Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction The Agency for Toxic Substances and Disease Registry ranked
mercury as the third priority hazardous substance, after arsenic
The regulations at international level are becoming more and and lead [1]. Water pollution by Hg(II) is thus a serious environ-
more drastic concerning the discharge of heavy metal ions into mental issue, though many national or international directives
the environment. Therefore controls on industrial wastewater are are recommending as far as possible substituting other metal to
stressed because of their potential threat on human health and mercury or changing the production processes to avoid using mer-
on the biosphere. In addition, due to the rarefaction of some metal cury. This metal is non-biodegradable and can threat the ecosys-
resources (critical, strategic and precious metals, for example) tem by being accumulated in the food chain; the dramatic
strongly incentive politics of material reuse and recycling have Minamata intoxication of local population clearly illustrates this
been recently elaborated. hazardous effect. Hg(II) may be discharged into the environment
through wastewaters issued from different industries such as:
electroplating, leather tanning, metal finishing and petroleum
⇑ Corresponding author. Tel.: +33 (0)466782734; fax: +33 (0)466782701. refining.
E-mail address: Eric.Guibal@mines-ales.fr (E. Guibal).

http://dx.doi.org/10.1016/j.cej.2015.05.110
1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
346 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

Hg(II) removal from contaminated water is imperative to make have developed an alternative for solving the problem of filtration
water fit for the human uses (drinking, agriculture, etc.). The or recovery of fine (micro- or nano) particles: the incorporation of a
allowed concentrations for discharge into surface water and for magnetic core (or its in situ production, with simultaneous coating
drinking water are 10 and 1 lg/L, respectively [2]. The aforemen- with biopolymer) in the particle makes easier the handling and
tioned threats of mercury require developing treatment processes recovery of these fine sorbent particles. These methods are cheap
capable to face, alone or in combination with other processes, the and often highly scalable. Moreover, these techniques employing
target levels for discharge into the environment. Generally, metal external magnetic fields are more amenable to automation [8].
ions can be recovered from solutions through conventional The reactivity of the sorbent can be improved by grafting new
processes such as precipitation, solvent extraction, membrane functional groups such as polyamine compounds [9–13]. The
techniques, ion-exchange and chelating resins. Frequently these objective may consist in increasing the density of sorption sites,
techniques fail to fit with target regulations (technical limitations), in improving the sorption selectivity, in changing the sorption
or economic constraints (solvent extraction is not appropriate for mechanism (ion-exchange vs. chelation or the reciprocal), or in
dilute effluents; membrane processes may be expensive for extending the field of application (especially regarding pH range).
large-scale applications) or produce huge amounts of contami- The objective of this work consists in synthesizing a magnetic
nated sub-products (with potential complementary hazards for composite made of glycidyl methacrylate (GMA) and chitosan
the environment, as it may occur with precipitation processes). (coating the magnetic Fe3O4 core) that is functionalized by grafting
Sorption is one of the preferred methods for the removal of metal diethylenetriamine (DETA) through GMA (to increase the density
ions from the wastewater as it is effective as well as cost-effective of sorption sites and their potential selectivity). These composite
[3]. In the field of sorption processes, several alternative materials materials are being tested for Hg(II) recovery through the study
have been investigated for the last decades making profit of mate- of pH effect, the investigation of competition effects (composition
rials of biological origin for metal binding: biosorption consists in of the matrix: competitor metal ions, counter anions), the determi-
using functional groups held on materials of biological origin for nation of sorption isotherms and thermodynamic characteristics
metal recovery with mechanisms similar to those found in and the identification of controlling steps in uptake kinetics.
ion-exchange and chelating resins. These materials can be issued Finally the desorption of Hg(II) is studied with the objective of ver-
from agriculture, fishing or as by-product of other industries: fun- ifying the possibility to recycle the sorbent.
gal, algal, bacterial biomass, agriculture or food industry residues,
for example. Chitosan ((1,4)-2-amino-2-deoxy-b-D-glucan) is an
emblematic biopolymer that was abundantly studied for metal 2. Materials and methods
binding. Being produced at the commercial scale from the shells
of crustacean, it is obtained by alkaline deacetylation from chitin 2.1. Materials
(the primary resource) and bears many hydroxyl groups (which
bring hydrophilic behavior to the biopolymer, and also possible All chemicals used were of analytical grade and demineralized
sites for chemical modification and grafting) and amino groups. water was used for the preparation of all aqueous solutions.
This biopolymer has unique properties among polysaccharide hav- Glycidyl methacrylate (GMA) was provided by Riedel-de Haën
ing cationic behavior in acid solutions. Amino groups are involved (Germany), while N,N0 -methylenebisacrylamide (MBA) and ben-
in metal binding through different mechanisms including (a) metal zoyl peroxide (Bz2O2) were supplied by Fluka AG (Switzerland).
cation chelation in near neutral solutions (through the free elec- Chitosan, epichlorohydrin (chloromethyloxirane) and diethylene-
tron doublet of nitrogen), and (b) binding of metal complex anion triamine (DETA) were obtained from Sigma-Aldrich (Switzerland)
by ion-exchange/electrostatic attraction on protonated amino while isopropyl alcohol was provided by Carlo Erba (France).
groups (in acidic solutions). With a pKa that depends on the degree Chitosan was provided by Aber Technologies (France); the deacety-
of acetylation (incomplete deacetylation of chitin) and the degree lation degree of the biopolymer was 87% and the molecular weight
of neutralization but ranges between 6.3 and 6.7 for most common (MWn) was 125,000 g mol1. All other chemicals were Prolabo
commercial samples [4] the protonation of chitosan leads, in most products (France) and were used as received. HgCl2 salt was used
cases (with the remarkable exception of sulfuric acid) to the dis- for the preparation of the stock solution, except in the case of the
solving of chitosan in acid media. This is the base of the mecha- study of counter anions (for which the relevant salt of mercury
nisms used for changing the conditioning of the biopolymer but was used).
this is also a serious drawback for potential applications in the field The principle of the synthesis of Fe3O4 particles consisted in the
of metal sorption. A cross-linking treatment or a chemical modifi- co-precipitation of ferric and ferrous salts following a procedure
cation is thus required for extending the pH field of application (for derived from the so-called Massart method [14].
both sorption and desorption steps). On the other hand, glycidyl
methacrylate (GMA) is an attractive vinyl monomer because of
its low toxicity, lower cost compared with other acrylic monomers, 2.2. Preparation of the magnetic-macromolecular hybrid material
versatile properties and especially due to the presence of epoxy
group in its molecule [5], which makes it very reactive for chemical Step 1: Magnetic chitosan particles were prepared by the
modification or for reacting with other materials. These polymers co-precipitation method, 4 g chitosan powder was dissolved in a
are potent sorbents for various transition metals [6,7]. Various volume of 150 mL of acetic acid solution (10%, w/v). The magnetite
studies on metal sorption on chitosan-based materials have precursors were mixed with 150 mL of demineralized water (i.e.,
pointed out the limitations in uptake kinetics due to the resistance 6.479 g of FeCl36H2O and 3.334 g of FeSO47H2O) using a volume
to intraparticle diffusion (which may be explained by the residual of 2 mL of HCl (concentrated) for improving salt dissolution under
crystallinity of the biopolymer and by poor porous and surface continuous stirring for 30 min. After complete dissolution of iron
properties). This drawback was partially solved changing the con- salts, the solution was mixed with the chitosan solution. Then
ditioning of the biopolymer (manufacturing gel beads, membranes the chemical precipitation of magnetic chitosan material was per-
and so on). Another possibility consists in reducing the size of sor- formed at 30 °C under vigorous stirring by dropwise addition of
bent particles but at the expense of serious drawbacks concerning 20 mL of NaOH solution (50%, v/v) for 30 min. The pH was roughly
particle recovery at the end of the process (in batch systems) or controlled around pH 11. The solid product that was recovered by
head loss (in fixed-bed applications). Recently, a number of studies centrifugation was freeze-dried using a freeze-dryer (Bioblock
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 347

Scientific, Christ) at 223 K and 0.01 mbar; this intermediary pro- Scanning Electron Microscope (ESEM) Quanta FEG 200, equipped
duct was called MC (magnetic chitosan). with an OXFORD Inca 350 Energy Dispersive X-ray microanalysis
Step 2: The magnetic-chitosan-glycidyl-methacrylate macro- (EDX) system). The magnetic property of the sorbent was mea-
molecular hybrid material was prepared through the polymeriza- sured on a vibrating-sample magnetometer (VSM) (Lake Shore
tion of GMA in the presence of MC particles (Step 1). GMA (5.0 g) 730T, Westerville OH, USA) at room temperature.
corresponding to a mass ratio of 50% (w/w, referred to MC amount)
was used for GMA grafting; 0.2 g of MBA was used as the
2.5. Sorption experiments
cross-linking agent and 0.1 g of Bz2O (acting as the initiator of
the polymerization reaction) was added under agitation before
Demineralized water was used for the preparation of metal ion
adding 1.0 g of magnetite. Three milliliter of isopropyl alcohol
solutions. A stock solution (20 mM) of HgCl2 was prepared in dem-
and 25 mL of cyclohexane were mixed and added to the former
ineralized water. The other solutions were obtained by dilution of
solution. The solution was then poured into a flask containing
the stock solution with demineralized water just prior experi-
100 mL (1%) polyvinyl alcohol and heated on a water bath at
ments. HCl (0.5 M) and NaOH (0.5 M) were used to control the
75–80 °C under continuous stirring for 3 h. The product was fil-
pH of the solutions.
tered off and washed repeatedly with demineralized water and
Sorption experiments were performed in batch systems using
acetone before being air-dried. The product was called MCGMA.
polyethylene flasks and the temperature was set to 20 ± 1 °C
The preparation and modification of magnetic-macromolecular
(unless specified). For the study of pH effect 20 mL of 10 mM
hybrid material is described in Scheme 1, and the suggested chem-
Hg(II) solutions at different pH values (in the range 1–6) were
ical structure of MCGMA is shown in Scheme 2.
mixed with 50 mg of sorbent (dried weight) for 5 h, and the stirring
speed was maintained at 100 rpm using a reciprocal agitator
2.3. Chemical modification of MCGMA Rotabit, J.P. Selecta (Spain). Samples were collected and filtrated
through 1.0 lm pore size filtration membrane and/or magnetic
2.3.1. Diethylenetriamine grafting – First method separation and the filtrate was analyzed for residual Hg(II) concen-
MCGMA was suspended in dioxane (100 mL) and then treated tration using ICP-AES. The pH was not controlled during the sorp-
with 5 mL of diethylenetriamine (DETA). The suspension was stir- tion but the final pH was systematically recorded.
red at 70 °C for 12 h. The product was successively washed with For sorption isotherms 50 mg of sorbent (m) were mixed with
water and acetone. The product was then air-dried and called 20 mL (V) of Hg(II) solutions at different initial concentrations
MCGMA-I. (C0, ranging between 3 and 10 mmol Hg L1) for 5 h. The pH of
the solutions was initially set at 4. After solid/liquid separation,
2.3.2. Diethylenetriamine grafting – Second method (via the residual concentration (Ceq, mmol Hg L1) was determined by
epichlorhydrin) ICP-AES and the sorption capacity (qeq, mmol g1) was determined
MCGMA was suspended in 70 mL of isopropyl alcohol. Then by the mass balance equation: qeq = (C0  Ceq) V/m.
7 mL of epichlorohydrin (62.5 mmol) dissolved in 100 mL ace- For uptake kinetics 300 mg of sorbent were mixed with 120 mL
tone/water mixture (1:1 v/v) was added to the suspension. The of Hg(II) solutions (C0: 10 mmol Hg L1) at pH 4. Samples (a volume
reaction was performed under continuous stirring for 24 h at of 4 mL) were collected (the sorbent was magnetically separated) at
60 °C. The solid product was filtered off and washed several times fixed times and the residual concentrations were determined by
with water. The obtained product was treated with DETA according ICP-AES. The agitation speed was set at 100 rpm while the temper-
to the method reported in Section 2.3.1. The product was finally ature was maintained at 20 ± 1 °C. The sorbed amount of Hg(II) per
air-dried; it was sieved and the fraction below 1 mm was retained unit weight of the sorbent at time t (q(t), mmol Hg g1), was calcu-
for experiments. This sorbent was called MCGMA-II. lated from the mass balance equation (taking into account the
decrement in the volume of the solution) as:
2.3.3. Estimation of the amine content  
The amine content in the obtained resin was estimated using a X
n CðtÞði1Þ  CðtÞðiÞ  VðtÞði1Þ
volumetric method previously described by Atia et al. [15]. Fifty qðtÞ ¼ ð2Þ
i¼1
m
milliliter of HCl (0.05 M) was added to 0.5 g of MCGMA-II or
MCGMA-II and conditioned for 15 h on a shaker at 20 °C. The resid-
where C(t)(i) (mmol Hg L1) is the Hg(II) concentration of the with-
ual concentration of HCl was measured by the titration against
drawn sample number i at time t and C(t)(0) = C0, V(t)(i) (mL) is the
0.05 M standardized NaOH (using phenolphtalein as the titration
volume of the solution in the flask at sample number i and time t,
indicator). The number of moles of HCl that reacted with the amino
and m is the amount of resin added into the flask. Here
groups was determined and the concentration of amino group was
V(t)(i)  V(t)(i1) equals 4 mL.
calculated by the following equation
Regeneration experiments were performed by contact of
Concentration of ðNH2 Þ groups 250 mg of the sorbent with 100 mL of 10 mmol Hg L1 solution
at pH 4 Hg(II) for 5 h. The amount of metal sorbed (and the sorp-
ðM 1  M 2 Þ  50
¼ ðmmol=g resinÞ ð1Þ tion capacity) was determined by the mass balance equation. The
0:5
solution was magnetically decanted and the sorbent was washed
where M1 and M2 are the initial and final concentrations of HCl. by distilled water. The loaded sorbent was mixed with 100 mL of
100 mM KI for 50 min. Samples were collected at different time
2.4. Analytical methods intervals and the residual concentration of Hg(II) was determined
by ICP-AES after solid/liquid separation. The regenerated sorbent
Mercury was analyzed using inductively coupled plasma atomic was carefully washed by distilled water for reuse in the second
emission spectrometer ICP-AES (Jobin-Yvon Activa M, Horiba-Jobin run. The regeneration efficiency (RE, %) was calculated according
Yvon, France). The morphology and the elemental distribution of to the following equation:
Hg in the magnetic-macromolecular material were analyzed with
a Scanning Electron Microscope coupled with an Energy
Amount of desorbed HgðIIÞðmmolÞat runðnÞ
RE % ¼  100 ð3Þ
Dispersive X-ray analysis system (SEM-EDX; Environmental Amount of desorbed HgðIIÞðmmolÞat runðn  1Þ
348 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

O
NH2 NH2
HO HO
OH n polymerization O CH2 CH CH2
O OH O
H2N H2N
OH OH
O
O
(MC) + (GMA) + (magnetite) (MCGMA)
NH
NH2 NH2
CH2
HO CH
NH2
HO
NH
(MCGMA) NH2 NH2 O CH2 CH CH2

H2N OH
OH
HO CH
CH2
NH
NH2 NH2

( M C G MA - I)

CH2 HC CH2
O
NH O
HO

(MCGMA) Cl CH2 HC CH2 O CH2 CH CH2


O OH O
HN
OH
H2C CH CH2 O
O
( M C G M A - E CH )

NH
OH NH2 NH2
NH
NH2 NH H2C CH CH2 HO CH
NH
HO
NH
(MCGMA-ECH) NH2 NH2 O CH2 CH CH2

HN OH O
NH OH
NH2 NH H2C CH CH2
HO CH
OH
CH2
NH
NH2 NH2

(MCGMA-II)

Scheme 1. Scheme for the synthesis and modification of magnetic-macromolecular hybrid material.

3. Results and discussion a quantitative analysis of the Hg load on the sorbent by SEM-EDX
would not be accurate, the levels of Hg are high enough to demon-
3.1. SEM and SEM-EDX analysis strate qualitatively the homogeneous distribution of the metal on
the surface of the sorbent. The mass percentages of Hg element
Fig. 1 shows the SEM and SEM-EDX analyses of the sorbents are found close to 27.5% and 32.9% for MCGMA-I and MCGMA-II,
MCGMA-I and MCGMA-II before and after Hg(II) sorption. The sur- respectively. This is a first evidence of the good affinity of the sor-
face of MCGMA-I and MCGMA-II is generally smooth with several bents for mercury. The presence of Cl element is due to the binding
small pores inside and interconnected macropores can be identi- of mercury under the form of chloroanionic species: the sample
fied within the sorbents (Fig. 1a and b). However, after Hg(II) sorp- used for analysis was collected after metal sorption at pH 3
tion, differences in the surface structure of the sorbents clearly (controlled with HCl).
appear: agglomerated spherical particles are formed on the surface Fig. AM1 (see Additional Material Section) shows the SEM
and the porous structure disappears (Fig. 1c and d). These conclu- images of the two sorbents at different magnitudes. There is a
sions are consistent with observations made on SEM-EDX analysis large dispersion in the size of sorbent particles (in the range
(Fig. 2): the typical signals of Hg element are appearing at 1.7, 2.2, 10–200 lm) and the particles are characterized as
2.6 and 10 keV on both MCGMA-I and MCGMA-II sorbents. Though irregularly-shaped flakes.
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 349

H2C CH3 CH3


C H2C CH
CH CH2 C
C O O C
O O C C O
NH
NH O
H2C
CH2
CH2 H2C
HC OH NH
NH HC OH
O C
CH2
O C
H2C CH CH2
O CH2 CH
O O
O O
O O
HO NH2 O
HO NH2

Chitosan-GMA (MBA) Macromolecular hybrid material

(MCGMA)

Scheme 2. Proposed chemical structure of chitosan–glycidyl methacrylate – N,N0 -methylene-bis-acrylamide (macromolecular hybrid material).

(a) (b)

(c) (d)
Fig. 1. Scanning electron micrographs of; (a) unloaded MCGMA-I, (b) unloaded MCGMA-II, (c) Hg(II)-loaded MCGMA-I and (d) Hg(II)-loaded MCGMA-II.

3.2. Amine content magnetite fraction in the sorbents was characterized by weight
loss at 700 °C: the materials contain 36% and 34% of magnetite
The amine content in MCGMA, MCGMA-I and MCGMA-II (which for MCGMA-I and MCGMA-II, respectively. On the basis of similar
was determined by titration) are 3.12, 5.92 and 6.5 mmol g1, magnetite fraction and chitosan (without chemical modification)
respectively. The chitosan used in this study was characterized the theoretical nitrogen content would be close to
by a deacetylation degree of 87% (this means 6 mmol N g1). The 3.9 mmol N g1. This confirms the efficient grafting of additional
350 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

(a) (b)

(c) (d)
Fig. 2. Energy dispersive X-ray analysis (EDX) of MCGMA-I and MCGMA-II sorbents: (a) unloaded MCGMA-I, (b) unloaded MCGMA-II, (c) Hg(II)-loaded MCGMA-I and (d)
Hg(II)-loaded MCGMA-II.

amino groups on the reference material (MCMA). The 25


semi-quantitative analysis of nitrogen content obtained by 20 MCGMA-I
SEM-EDX analysis confirms this trend. Indeed, the nitrogen mass
Moment/Mass (emu g-1)

15 MCGMA-II
percentages were close to 8.6% and 9.1% for MCGMA-I and
10
MCGMA-II, respectively (this means 6.15 and 6.5 mmol N g1,
5
respectively). These observations and analyses confirm the suc-
cessful experimental procedure described in Scheme 1 for the 0
grafting of additional amino groups on chitosan backbone. -5
-10
3.3. Magnetic properties
-15

The magnetic properties of the materials were determined -20


using VSM (vibrating sample magnetometry). Fig. 3 shows their -25
-2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500
typical magnetization loop. There was no remanence and coerciv-
Field (G)
ity, contrary to certain supported-magnetite materials [16]. This
means that MCGMA sorbents are superparamagnetic. The satura- Fig. 3. Magnetization curves for MCGMA-I and MCGMA-II sorbents.
tion magnetization of MCGMA-I and MCGMA-II were found to be
about 19.1 and 17.9 emu g1, respectively. These values are much
smaller than the levels reported for bulk phase magnetite (i.e.,
92 emu g1) and also smaller than the values obtained for mag- can be readily separated with the help of an external magnetic
netite nanoparticles [17]. The slight decrease in the magnetization field. This may be very helpful for solid/phase separation or for
in case of MCGMA-II may be due to the grafting of DETA via handling the material in hazardous environment.
epichlorhydrin, which increases the mass of organic material and
decreases the relative percentage of magnetite in the sorbent. 3.4. Infra-red spectrometry
The reported decrease in saturation magnetization can be
explained by several factors including size effect and particle crys- FT-IR spectrometry was used to characterize the structure of the
tallization [17], and obviously by the fact that only a fraction of the sorbents (Fig. 4). The two sorbents have very similar FT-IR profiles:
sorbent (about 65%) is constituted by the magnetic core. Similar the same bands are appearing on the two spectra, the only small
decrease in magnetization was observed for other differences observed are small shifts in some of these bands. The
chitosan-magnetite composites [18]. Therefore, MCGMA sorbents large band around 3410 cm1 is generally attributed to the
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 351

combination of stretching vibrations of OH and NH2 groups, while properties of the different amino groups present on the sorbent
a series of –CH vibrations are reported around 2940 and 2850– are different to those on the MCGMA-II sorbent due to differences
2860 cm1. These bands are present in a wide number of com- in their chemical environment. The decrease in amino group proto-
pounds and they are bringing poor information on the specific nation at pH above 2.5 (which contributed to bind anionic mercury
properties of the sorbents. More interesting and representative chloroanions) is not compensated by the chelation properties of
are the different bands identified in the range 1800–400 cm1. amino groups and by the speciation of mercury in solution (metal
The bands at 1264, 1162 and 986/990 cm1 are representative of speciation is shifted from chloroanions to positively charged mer-
the saccharide ring. The band at 1653–1657 cm1 is associated to cury species with low affinity for protonated amino groups). The
the CONH2 group [19], while the band at 1561–1573 cm1 is prob- two sorbents have comparable sorption capacities at pH 2.3: below
ably assigned to the in-plane bending vibration of free amine this value MCGMA-I is more efficient for Hg(II) binding while
groups and/or the asymmetric carboxylate stretching vibration above this value the best results were obtained by MCGMA-II
[19,20]. The grafting of diethylenetriamine, via glycidylmethacry- sorbent.
late, can be identified by the strong peak observed around Above pH 5 the formation of colloidal mercury species
1724 cm1 (assigned to C@O stretching vibration) [20,21]. (Hg(OH)2) may occur and their precipitation may overestimate
Additional methylene (–CH2–) groups can probably be associated sorption performance [25]. This phenomenon should be taken into
to the band observed at 1466–1470 cm1 [22]. The band at 586– account. For further experiments the pH was set to 4–4.2.
596 cm1 was attributed to c-Fe2O3 [23].
3.5.2. Kinetics
3.5. Hg(II) sorption properties The uptake kinetics of Hg(II) using MCGMA-I and MCGCMA-II is
shown in Fig. 6. Fig. 6a shows the relative decay of concentration
3.5.1. Effect of pH with time, which allows comparing the two sorbents both in terms
The pH of the aqueous solution is a key parameter that controls of equilibrium performance and relative decay slopes. Fig. 6b
the sorption of the Hg(II). This effect can be associated to metal shows the evolution of the concentration of Hg(II) in the sorbent
speciation, to chemical properties of the polymer (protona- (i.e., q(t) vs. time): the approach of equilibrium is another method
tion/deprotonation of amino groups). Fig. 5 shows the effect of for comparing the kinetic profiles. Experimental conditions have
pH on Hg(II) sorption (varying the initial pH from 0.9 to 6), and been selected to be able to illustrate the effect of resistance to dif-
the pH variation observed during sorption operation. Fig. 5a con- fusion. Indeed, an excess of sorbent (compared to metal) would
firms that the pH influences sorption capacity: the sorption lead to fast and complete removal of the metal bound on the sorp-
increases with the pH up to pH 4–4.2 before beginning to stabilize tion sites at the surface of the sorbent and minimizing the impact
and even reducing when pH exceeds 5. It is remarkable that the pH of mass transfer mechanisms. Under these conditions, the solu-
variation during metal sorption is not varying significantly tions is not fully decontaminated (Fig. 6a); this issue (complete
(Fig. 5b): at low pH (below pH 4) a slight increase of the equilib- metal recovery) is addressed below (see Section 3.5.7).
rium (below 0.4 pH unit) is observed, while above pH 4–4.5 the The kinetic profiles (both in terms of relative concentration
equilibrium pH tended to slightly decrease (variation being less decay; i.e., Fig. 6a, and metal concentration in the sorbent; i.e.,
than 0.5 pH unit). This is remarkably stable compared to other Fig. 6b) are characterized by two main phases in the uptake: (a) a
chitosan-based systems where a kind of ‘‘buffering effect’’ is first initial step that lasts for about 60 min and counts for more than
obtained around 5–6 due to the acid-base properties of the 60% of total sorption, (a) a second step that takes about 6 h and cor-
biopolymer (pKa in the range 6.3–6.7) that binds protons. It is sus- responds to a much slower metal accumulation. The initial section
pected that the grafting of new functional groups (including other of the curve corresponds to a great availability of reactive groups
amino groups with different acid-base properties, associated to (surface coverage is progressively increasing) and a large concentra-
primary and secondary amino groups) contributes to smooth this tion gradient between the solution and both the surface and the
pH variation effect. internal sorption sites. These two conditions may explain the fast
As the pH increases the protonation of amine groups decreases, initial accumulation of Hg(II). The sorption mainly occurs on the
which, in turn, contributes to make free the amino groups for inter- reactive groups covering the surface of the sorbent. The second step,
acting with metal cations by chelation. For MCGMA-II sorbent the much slower, is controlled by the decrease of the concentration gra-
sorption capacity increases from pH 1 to pH 4 and then tends to dient and by the resistance to intraparticle diffusion and requires
stabilize. At higher pH (above pH 5) the formation of hydrolyzed much longer time for reaching the equilibrium (i.e., about 6 h).
species (Hg(OH)+ or soluble Hg(OH)2, before precipitation) that Actually the binding kinetics is controlled by a series of mecha-
have less affinity for amino groups, may explain the slight decrease nisms including: (a) the bulk diffusion, (b) the resistance to film
in sorption capacity (which represents less than 0.2 mmol Hg g1). diffusion (or external diffusion), (c) the resistance to intraparticle
This effect is completed by the impact of deprotonation of amine diffusion, and (d) the proper reaction rate (chemical reaction rate)
groups. On the other hand, at low pH (below 2–3), the sorption [26]. Usually a sufficient agitation allows neglecting the resistance
capacity remains quite high (up to 1.5 mmol Hg g1), especially to bulk diffusion and minimizes the resistance to film diffusion
compared to the levels reached with pure chitosan [24]. This can whose contribution in the control of uptake kinetics is mainly sig-
be explained by either (a) differences in the acid-base properties nificant within the first minutes of contact. The modeling of such a
of amino groups grafted on chitosan (lower pKa compared to chi- complex system, which is beginning even more complex when the
tosan amino groups), and/or (b) the binding of chloro-anionic mer- sorbent is heterogeneous, when the solution is subject to changes
cury species on protonated amino groups. Indeed, in the presence in the speciation of the metal ions, etc., requires complex numeri-
of HCl, at low pH, metal speciation can be displaced to the forma- cal tools. Experimental data have been modeled using simplified
tion of HgCl2 
4 or HgCl3 species which may be bound to protonated conventional equations to fit kinetic profiles and make possible
amino groups through an electrostatic attraction mechanism of an the comparison of kinetic parameters for the two sorbents.
ion-exchange mechanism: chloride ions bound to protonated Hence, the kinetics of Hg(II) sorption on modified MCGMA sorbents
amino groups may be exchanged with anionic chloroanions. The were analyzed using the pseudo-first order rate equation (PFORE)
sorbent MCGMA-I has a slightly different behavior: the sorption [27], the pseudo-second order rate equation (PSORE) [28], the sim-
capacity tends to increase with pH increasing from 0.9 to 2.5 and plified resistance to intraparticle diffusion equation [29] and the
progressively decreases when pH increases: the acid-base Elovich equation [30]. These models and their linear forms are
352 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

Transmittance

MCGMA-I
MCGMA-II

4000 3600 3200 2800 2400 2000 1600 1200 800 400
Wavenumber (cm-1)

Fig. 4. FT-IR spectra of MCGMA-I and MCGMA-II sorbents.

3.5 0.9

3.0 MCGMA-I

0.8 MCGMA-II
2.5
q e (mmol g-1)

2.0 0.7
C(t)/C0

1.5
0.6
1.0
MCGMA-I
0.5
0.5 MCGMA-II

0.0 0.4
(a) 0 1 2 3 4 5 6 7 8 9 10 (a) 0 50 100 150 200 250 300 350 400
Equilibrium pH t (min)

7 2.1

1.8
6
q (t) (mmol g-1)

1.5
5
1.2
Final pH

4 MCGMA-I
0.9
MCGMA-II
3 0.6 Simulated from PSORE (MCGMA-I)

2 0.3 Simulated from PSORE (MCGMA-II)


MCGMA-I

1 MCGMA-II 0.0
0 50 100 150 200 250 300 350 400 450 500
(b) t (min)
0
0 1 2 3 4 5 6 7
(b) Initial pH
Fig. 6. Hg(II) uptake kinetics using MCGMA-I and MCGMA-II (a: relative concen-
tration decay of Hg(II) in the solution, b: Hg(II) concentration in the sorbent (i.e.,
q(t)) vs. time) (C0: 10 mmol Hg L1; pH 4; T: 20 °C; sorbent dosage, SD: 2.5 g L1).
Fig. 5. pH effect on Hg(II) sorption using MCGMA-I and MCGMA-II: (a) sorption
capacity, (b) pH change (T: 20 °C; C0: 10 mmol Hg L1).

sorption rate (mmol g1 min1) and b the desorption constant


reported in Table 1, where k1 is the pseudo-first order rate constant (g mmol1). The validity of each model is checked by the correla-
(min1) of sorption and qeq and q(t) (mmol Hg g1) are the tion coefficient associated to the linear fits. Table 2 reports the
amounts of Hg(II) sorbed at equilibrium and time t, respectively, parameters of the different models for both MCGMA-I and
k2 is the pseudo-second order rate constant (g mmol1 min1), Ki MCGMA-II sorbents. Systematically, the best correlation coeffi-
is the intraparticle diffusion rate (mmol g1 min0.5), a the initial cients were found for the PSORE model; this is confirmed by the
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 353

plot of experimental data according the linearized forms of these linearization as well the plots to be used for determining the
models: Fig. AM2a and AM2b (see Additional Material Section) parameters (which are reported in Table 2). The values of a and
for PFORE and PSORE, respectively, show a best fit of kinetic pro- b were determined from the intercept and slope, respectively, of
files by PSORE. In addition, the comparison of equilibrium sorption the linear plot of qt vs. ln t (Fig. AM3b, see Additional Material
capacities for the calculated values and the experimental values Section). The values of a for the sorption of Hg(II) ions on the mod-
are only consistent for the PSORE model: the equilibrium sorption ified sorbents are 0.46 and 0.73 (mmol g1 min1) for MCGMA-I
capacities are found close to 1.4 and 2.0 mmol Hg(II) g1 for and MCGMA-II, respectively. These values are higher than the val-
MCGMA-I and MCGMA-II, respectively. PSORE modeling gave val- ues cited in literature [34,35], which may be attributed to the high
ues of 1.49 and 2.09 mmol Hg(II) g1 closer from experimental val- concentration of active sites on the sorbent surface allowed for
ues than PFORE (0.91 and 1.27 mmol Hg g1, respectively). It was reacting with Hg(II) ions. The values of b (desorption constant)
more likely to reflect that the rate-determining step might be are found to be 4.54 and 3.54 g mmol1 for MCGMA-I and
chemical sorption and that the sorption behavior might involve a MCGMA-II, respectively. These values are smaller compared with
chelation mechanism through coordination between Hg(II) ions the levels cited in the literature [34,35]. This is another confirma-
and the reactive groups of the sorbents [25]. The Hg(II) ions may tion of the high affinity of the sorbents for Hg(II) ions.
form complexes with amino groups of the modified MCGMA sor-
bents (with possible contribution of hydroxyl groups in the stabi- 3.5.3. Equilibrium sorption isotherm
lization of metal binding) [31]. The sorption isotherms represent the distribution of the sorbate
However, the PSORE describes kinetics data through a global between the solid phase (sorption capacity or sorbate concentra-
approach, and does not take into account the contribution of diffu- tion in the solid) and liquid phase (residual concentration of the
sion mechanisms in the control of the kinetics. Under these condi- sorbate in the solution), at equilibrium. This distribution is only
tions, the kinetic parameters should be considered as apparent rate controlled by the temperature and is independent of the size of
coefficients. The influence of resistance to intraparticle diffusion sorbent particles, on the experimental procedure (batch,
has been approached using a simplified model: the so-called fixed-bed column) providing a sufficient time has been given to
Weber and Morris plot (Table 1). The intraparticle diffusion model the process for reaching equilibrium. The distribution of the sor-
provides a more comprehensive approach for defining sorption bate between the two phases can be modeled using different equa-
mechanism, and the plot generally allows identifying different suc- tions; however, the models the most frequently used for describing
cessive steps in the global process [32]. The Weber and Morris solid/liquid equilibrium systems have been established by
shows multi-linear sections (Fig. AM3a, see Additional Material Langmuir [36], Freundlich [37], and Dubinin–Radushkevich [38].
Section), i.e., three linear sections (on the plot q(t) vs. t0.5) with fast These models and their linear forms are reported in Table 3, where
kinetics in first step followed by the gradual attainment of equilib- qeq the sorbed value of Hg(II) at equilibrium concentration
rium for both sorbents, and a pseudo saturation plateau. The (mmol Hg g1), qm,L is the maximum sorption capacity (corre-
multi-linear plot does not pass through the origin suggesting that sponding to the saturation of the monolayer, mmol Hg g1) and
the resistance to intraparticle diffusion is not the sole KL is the Langmuir binding constant, which is related to the energy
rate-limiting step: other steps, e.g., resistance to film diffusion of sorption (L mmol1), Ceq is the equilibrium concentration of
and/or reaction rate, are probably involved in the control of uptake Hg(II) in solution (mmol Hg L1). KF (mmol g1) (L mmol1)1/n
kinetics (Table 2, Fig. AM3a, see Additional Material Section). We and n are the Freundlich constants related to the sorption capacity
can assume that the first linear section corresponds to a regime and intensity, respectively. KDR (J2 mol2) is a constant related to
controlled by the resistance to film diffusion and that the binding the sorption energy, QDR (mmol g1) is the theoretical saturation
is limited in this stage at the sorption on external sorption sites capacity, e (J2 mol2) is the Polanyi potential.
or on the macropores in the first external layers of the material. At low initial concentrations (1.0–3.0 mM) the sorption of Hg(II)
The second section is characterized by a much lower kinetic rate was almost quantitative. Brunauer et al. [39] divided the sorption
and leads to a slow approach to equilibrium with the control by isotherms into five types. Type I isotherm represents unimolecular
the resistance to intraparticle diffusion (into internal macroporous sorption and applies to non-porous, microporous and sorbents
and mesoporous network). The last step is very slow and repre- with small pore sizes (not significantly greater than the molecular
sents only a few percentage of the total sorption: this phase can diameter of the sorbate). The isotherm curves (Fig. 7) are following
be associated to the resistance to diffusion in the microporous net- the typical shape of I-type systems, according to the BET classifica-
work of the sorbent. In addition the progressive saturation of avail- tion [39]: they are characterized by a high degree of sorption at low
able and accessible sorption sites influences the local equilibrium concentrations. At higher concentrations, the sorption sites becom-
on the surface between surface sorption and desorption. The low ing progressively occupied the sorption tends to stabilize and
values of intraparticle rate constants (Ki) appearing in Table 2 with forms the saturation plateau.
values available in the literature indicates that the sorbents (both The Langmuir model is the simplest theoretical model that is
MCGMA-I and MCGAM-II) are significantly affected by the resis- used for describing monolayer sorption onto a surface with a finite
tance to intraparticle diffusion, at least compared to the fast reac- number of identical sites. It was originally developed to represent
tion rate of chemical sorption of Hg(II) ions at the surface of the gaz/solid sorption before being extrapolated to liquid/solid sorp-
sorbent [33]. The analysis of the rate constants for the Weber tion. The values of qm,L and KL were determined from the slope
and Morris model shows that except for the first step, the param- and intercept, respectively, of the linear plot of Ceq/qeq vs. Ceq
eters were systematically higher for MCGMA-II compared with (Fig. AM4a, see Additional Material Section); parameters of the
MCGMA-I. In the initial stage, MCGMA-I seems to be faster in sorp- Langmuir equation are reported in Table 3. The maximum sorption
tion than MCGMA-II: this first stage corresponds to a kinetic con- capacities (qm,L) are in good agreement with the experimental val-
trol by external film diffusion and/or diffusion in the macropores ues. The correlation coefficients reported in Table 4 confirm the
present at the external surface of sorbent particles. While in the better fit of experimental data by the Langmuir model compared
next steps the uptake kinetics is controlled by the diffusion in with the other two models. Though the fact that the mathematical
meso- and micro-pores, which appears to be a little more accessi- fit of the sorption equation fits well experimental data does not
ble for MCGMA-II material. necessarily mean that the hypotheses associated to the model
The Elovich equation was developed for modeling chemisorp- are verified this is indicative of the relative homogeneity of the sur-
tion processes [30]. Table 1 displays the model equation and its face; or at least that the active sites are energetically equivalent.
354 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

Table 1
Kinetics models and their linear forms.

Kinetic model Non-Linear form Linear form Plot References


 
Pseudo-First order qt ¼ qe ½1  e  k1 t
 logðqe  qt Þ ¼ log qe  K1
t log (qe  qt) vs. t [27]
2:303
 
Pseudo-Second order qt ¼ k2 t t 1 1 (t/qt) vs. t [28]
1 þ k2 qe t q ¼ k q2 þ q t
t 2 e e

Intraparticle diffusion – qt = Kit0.5 + X qt vs. t0.5 [29]


Elovich equation dqt
¼ aebq qt ¼ 1b ln ab þ 1b ln t qt vs. ln t [30]
dt

Table 2
Kinetic parameters for Hg(II) sorption.

PFORE PSORE Weber & Morris model Elovich equation


Sorbent qe,exp k1 qe,calc R2 k2 qe, calc R2 Ki X R2 a b R2
MCGMA-I 1.43 0.0130 0.91 0.942 0.035 1.49 0.9976 Ki1 0.124 0.247 0.953 0.470 4.55 0.993
Ki2 0.042 0.751 0.942
Ki3 0.013 1.178 0.889
MCGMA-II 2.01 0.0095 1.27 0.979 0.017 2.10 0.9899 Ki1 0.091 0.664 0.991 0.736 3.54 0.951
Ki2 0.079 0.689 0.989
Ki3 0.035 1.342 0.996

Units: qe,exp/calc: mmol g1; k1: min1; k2: g mmol1 min1; Ki: mmol g1 min0.5; a: mmol g1 min1; b: g mmol1.

Table 3
Sorption isotherms: equation in their linear and non-linear forms.

Isotherm Non-Linear form Linear form Plot References


Langmuir qeq ¼
qm;L K L C eq Ceq
¼
Ceq
þ 1 Ceq
vs: Ceq [36]
1 þ K L C eq qeq qm;L K L qm;L qeq
Freundlich qeq ¼ K F C 1=n ln qeq ¼ ln K f þ 1n ln C eq ln qeq vs. ln Ceq [37]
eq
Dubinin–Radushkevich qeq ¼ Q DR expK DR e
2
lnqe ¼ lnQ DR  K DR e2 ln qeq vs. e2 [38]

2.4 The Freundlich isotherm model is applicable to highly heteroge-


neous surfaces, and a sorption isotherm lacking to form a satura-
2.1
tion plateau indicates a multi-layer sorption. The values of n and
1.8 KF were determined from the slope and intercept, respectively, of
the linear plot of ln qeq vs. ln Ce (Fig. AM4b, see Additional
q eq (mmol g-1)

1.5 Material Section) and the calculated parameters of the Freundlich


equation are summarized in Table 4. Both the saturation plateau
1.2
formed on Fig. 7 at high residual mercury concentration and the
0.9 MCGMA-I
relatively low correlation coefficients confirm that the Freundlich
MCGMA-II
poorly fit experimental data.
0.6 Another popular equation for the analysis of isotherm was pro-
Simulated Langmuir isotherm (MCGMA-I)
posed by Dubinin and Radushkevich (Table 3). This isotherm was
0.3 Simulated Langmuir isotherm (MCGMA-II)
developed taking into account the effect of the porous structure
0.0 of the sorbent, and the energy involved in the sorption process.
0 1 2 3 4 5 6 7 8 9 10 The Polanyi potent (e) is given by Eq. (5) [38]:
Ceq (mM)  
1
e ¼ RTln 1 þ ð5Þ
Fig. 7. Hg(II) sorption isotherms using MCGMA-I and MCGMA-II (pH 4; T: 20 °C). C eq
R is the universal gas constant (8.314 J mol1 K1) and T is the
absolute temperature (K). The slope of the plot of ln qeq vs. e2
Furthermore, the Langmuir parameters can be used to predict the (Fig. AM4c, see Additional Material Section) gives KDR and the
affinity between the sorbate and sorbent using the dimensionless intercept yields QDR (Table 4). The D–R constant (KDR) can give
separation factor RL valuable information regarding the mean energy of sorption (Ea,
J mol1) by Eq. (6):
1
RL ¼ ð4Þ
1 þ K LCo 1
Ea ¼ ð6Þ
ð2K DR Þ0:5
where KL is the Langmuir equilibrium constant and Co is the initial
concentration of Hg(II) ions. Values of 0 < RL < 1 indicates the ‘‘suit- The results of D–R isotherm are reported in Table 4. Fig. AM4c
ability’’ of the process. In this study, the values of RL for the resin for (see Additional Material Section) shows that the experimental data
the sorption of Hg(II) ions lie between 0.002 and 0.088 for each sor- are poorly fitted to the model (consistently with the low value of
bents and all concentrations at 20 °C. These values confirmed the correlation coefficients): the values of the parameters of the model
affinity of modified MCGMA sorbents for Hg(II) ions. should be taken as indicative values (order of magnitude). The
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 355

Table 4
Parameters of the sorption isotherm models.

Sorbent Langmuir model Freundlich model Dubinin–Radushkevich (D–R) model


2 2
qmax,exp qm,L KL R n KF R QDR KDR Ea R2
(mmol g1) (mmol g1) (L mmol1) (mmol g1) (mmol g1) (J2 mol2) (kJ mol1)
(L mmol1)1/n
MCGMA-I 1.43 1.55 27.78 0.999 12.87 1.37 0.965 1.50 4.72  109 10.3 0.892
MCGMA-II 2.02 2.28 16.48 0.999 7.98 1.89 0.945 2.16 6.48  109 8.78 0.894

Table 5
values of the mean energy of sorption range between 8.7 and Effect of temperature on Hg(II) sorption capacity.
10.3 kJ mol1: this is consistent with the proposed mechanism of
chemisorption. Indeed, it is generally admitted that 8 kJ mol1 is Temperature (°C) Sorption capacity (mmol Hg g1)

the limit energy for distinguishing physical (below 8 kJ mol1) MCGMA-I MCGMA-II
and chemical sorption. A comparison of the correlation coefficient 20 1.48 2.12
values obtained from the Langmuir, Freundlich and D–R isotherm 30 1.74 2.40
models in Table 4 reveals that the correlation coefficients for the 40 1.91 2.58
50 2.21 2.65
Langmuir isotherm are somewhat higher than those for the
Freundlich and D–R isotherm. This result suggests that the binding
of Hg(II) may occur as a monolayer on the surface of the sorbent
and that the uptake occurs on a homogenous surface by monolayer The values of standard enthalpy change (DH°) and standard
sorption. This should be confirmed by experimental observation entropy change (DS°) for the sorption process are thus deter-
(as reported above the mathematical fit of isotherm curve by a mined from the slope and intercept of the plot of ln Kc vs. 1/T:
given model equation does not mean that the relevant hypotheses the values of thermodynamic parameters are reported in
of the model are verified). The uptake can be described in terms of Table 6. The positive values of DH° confirm the endothermic nat-
chemisorption as the formation of ionic or covalent bonds between ure of sorption process. The negative values of DG° indicate that
the sorbent (free mercury species or chloroanionic species, the sorption reaction is spontaneous. The increase in the negativ-
depending on the pH and the sorption mechanism) and the sorbate ity of DG° with increasing temperature confirms that the ‘‘favor-
(free amine groups or protonated amino groups depending on the ability’’ increases with temperature. In addition, the values of
pH). standard free energy change for MCGMA-II are more negative
The coexistence of different metal species that could be sorbed than these of MCGMA-I: Hg(II) sorption is more favorable on
in function of the pH as well as the different mechanisms, and the MCGMA-II than on MCGMA-I, especially at low temperature.
simultaneous presence of different types of amino groups are not The density of sorption sites and the spatial arrangement of func-
comforting the hypothesis of homogeneous surface (or homoge- tional groups are more favorable for Hg(II) binding on MCGMA-II
neous energies of sorption). sorbent. It is noteworthy that in the case of MCGMA-I, the stan-
dard free Gibbs energy becomes positive only at higher tempera-
3.5.4. Influence of temperature ture (i.e., above 40 °C): the spontaneous nature of sorption
The sorption performance may be controlled by the tempera- process is only possible when the system is ‘‘activated’’ by tem-
ture both in terms of kinetics (depending on the activation energy) perature. With MCGMA-II more amine groups are available; in
and equilibrium: the thermodynamics of the process influences the addition their distribution along a free long chain makes them
instantaneous kinetics of sorption and desorption, and conse- more accessible. This could explain the better ‘‘spontaneity’’ of
quently the equilibrium. The sorption capacities were compared sorption process for MCGMA-II compared to MCGMA-I (which
at different stabilized temperatures (under identical experimental possibly requires an increase in temperature for the enhancement
conditions) for both MCGMA-I and MCGMA-II (Table 5): the sorp- of metal ion or polymer chains ‘‘mobility’’ that, in turn, improves
tion capacity increases with temperature and the sorption is an sorption performance).
endothermic process. The sorption equilibrium constant, Kc was In industry and water purification plants the optimum tem-
determined (Eq. (7)) and used with the van’t Hoff equation (Eq. perature at which the sorption is highly feasible and sponta-
(8)) and conventional thermodynamic equation (Eq. (9)) for evalu- neous is essential. The sorption of metals onto sorbent surfaces
ating the thermodynamic constants of the sorbents (i.e., the stan- may be either spontaneous or non-spontaneous in function of
dard enthalpy change, DH°, the standard free Gibbs energy, DG°, temperature. The limit temperature value corresponding to a
and the standard entropy change, DS°). null value of standard free energy can thus be deduced from
Eq. (11). The range of temperature can be predicted from the
qeq value of temperature at which the standard free energy is zero
Kc ¼ ð7Þ (T0), and then the minimal temperature for the process to being
C eq
spontaneous.
where qeq and Ceq are equilibrium concentrations of Hg(II) on the
sorbent and in the solution, respectively. 
DH
T0 ¼  ð11Þ
DG ¼ RTlnK c ð8Þ DS

Here, the calculated values of zero standard free energy temper-


DG ¼ DH  T DS ð9Þ
ature (T0) are 312.8 and 283.8 for MCGMA-I and MCGMA-II,
Therefore the van’t Hoff equation becomes: respectively. The low T0 for MCGMA-II indicates the feasibility of
  Hg(II) removal at ambient temperature, while MCGMA-I sorbent
DH DS requires a thermal activation: the spontaneous reaction occurs at
lnK C ¼ þ ð10Þ
RT R around 40 °C.
356 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

Table 6
Standard enthalpy, entropy and free energy changes for Hg(II) sorption.

Adsorbent DH° (kJ mol1) DS° (J mol1 K1) T0 (K) DG° (kJ mol1)
293 K 303 K 313 K 323 K
MCGMA-I 19.2 61.4 312.8 1.22 0.60 0.01 0.62
MCGMA-II 15.3 54.0 283.8 0.50 1.04 1.58 2.12

Table 7
Effect of counter ion on Hg(II) sorption.

Initial pH Sorption capacity (mmol Hg g1)


HgCl2 HgNO3 HgSO4
MCGMA-I MCGMA-II MCGMA-I MCGMA-II MCGMA-I MCGMA-II
0.95 1.51 2.07 0.15 0.16 0.048 0.054
2.12 1.45 2.07 1.00 1.41 0.153 0.164
3.2 1.36 1.95 1.33 2.14 – –

3.5.5. Effect of counter anions and metal speciation (containing Hg(II), Co(II), Cu(II), Fe(II), Ni(II), Zn(II) and Mg(II)) at
The effect of the counter anion on the uptake of Hg(II) by mod- different pH values (0.4–1.6). Table 8 shows that the recovery of
ified MCGMA sorbents was investigated at different pH values Hg(II) is almost quantitative in the presence of equimolar concen-
using different mercury salts. The pH of the solution was adjusted trations of other base metal cations, regardless of the pH (at least in
using HCl for HgCl2 solution, HNO3 for Hg(NO3)2 solution and the selected pH range). Whatever the pH, the maximum sorption
H2SO4 for HgSO4 solution. Table 7 shows that at low pH (i.e., capacities of competitor metal ions by the modified MCGMA
0.95) an appreciable sorption was obtained in the case of HgCl2 sorbents is less than 5.0% of sorbed amounts of Hg(II) ions. The sor-
for both sorbents, while with Hg(NO3)2 and HgSO4 salts the sorp- bents appears to have a selective recognition pattern for Hg(II) ions
tion was negligible. This result can be explained by the possibility against other selected heavy metal ions. The reported negative val-
for the sorbents to bind mercury by an ion-exchange between the ues for the uptake of Fe(II) can be attributed to the release of iron
counter anions (chloride) bound to amino groups and the Hg(II) from magnetite at low pH: when the pH increases this release
2
chloroanions (HgCl 3 or HgCl4 ) [25]. In the case of other decreases. The stability of the composite material is an important
counter-anions, mercury does not form stable anionic complexes issue to take into account and it sounds preferable managing the
that could be bound to protonated amine groups by hybrid material in solution whose pH is higher than pH 1.5–2.
ion-exchange and the sorption strongly decreases. The limited The selectivity of the sorbents for Hg(II) against other base metals
uptake in the case of Hg(NO3)2 or HgSO4 salts may be also attribu- may be explained by the sorption mechanism which is attributed
ted to the higher stability of (MCGMA-NH+)NO 3 or the to ion-exchange of Hg(II) chloro-anions on protonated amine
(MCGMA-NH+)2SO2 4 that retards the ion exchange mechanism. groups in acid conditions: the other selected metal ions are not
At pH 2.12 for both MCGMA-I and MCGMA-II Hg(II) was signifi- supposed to form stable chloro-anions under selected experimen-
cantly sorbed when using HgCl2 or Hg(NO3)2 salts, while the sorp- tal conditions.
tion was negligible with HgSO4 salt. Sulfate salts have more affinity
for protonated amine groups than other mono-anions making
more difficult metal anion exchange. The composition of the solu- 3.5.7. Effect of sorbent dose
tion may thus have a significant impact on the sorption perfor- The sorption of Hg(II) on modified MCGMA sorbents was stud-
mance, especially at pH where the ion-exchange is the ied by changing the sorbent dosage (in the range 1.5–10 g L1),
predominant sorption mechanism. metal concentration being fixed to 9.8 mM (pH 4 and T: 20 °C).
Fig. AM5a (see Additional Material Section) shows the variation
in sorption capacity with sorbent dosage: the sorption capacity
3.5.6. Selectivity of modified MCGMA for Hg(II) sorption decreases from 1.52 to 0.97 mmol Hg g1 and from 2.07 to
In order to evaluate the possible selectivity of the sorbents 0.97 mmol Hg g1 for MCGMA-I and MCGMA-II, respectively when
Hg(II) sorption was carried out in multicomponent solutions the SD increases from 1.5 to 10 g L1. Fig. AM5b (see Additional

Table 8
Multicomponent sorption – effect of pH on the sorption capacity for Hg(II) Co(II), Cu(II), Fe(II), Ni(II), Zn(II) and Mg(II) (initial concentration of 1.0 mmol metal L1 for each metal
ion).

Adsorbent Initial pH Final pH Uptake (mmol g1)


Metal ion
Hg(II) Co(II) Cu(II) Fe(II)* Ni(II) Zn(II) Mg(II)
MCGMA-I 0.4 0.82 0.280 0.016 0.001 0.209 0.002 0.034 0.000
0.8 1.12 0.342 0.006 0.000 0.118 0.011 0.021 0.001
1.2 1.26 0.350 0.016 0.005 0.008 0.015 0.008 0.004
1.6 1.80 0.301 0.014 0.004 0.005 0.016 0.000 0.000
MCGMA-II 0.4 0.90 0.310 0.014 0.000 0.192 0.004 0.027 0.000
0.8 1.22 0.372 0.000 0.003 0.109 0.005 0.008 0.000
1.2 1.29 0.380 0.001 0.006 0.007 0.000 0.000 0.005
1.6 1.86 0.331 0.000 0.013 0.004 0.000 0.000 0.002
*
Negative values for Fe(II) are associated to metal release due to magnetite partial dissolving.
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 357

Table 9
Comparison of sorption capacity for Hg(II) ions with various sorbents.

Adsorbent material Initial Contact time Temperature Initial Sorbent dosage Sorption capacity References
pH (min) (°C) concentration (g L1) (mmol Hg g1)
(mM)
Thiourea modified Hg(II) ion-imprinted 5.00 180 30 1.99 0.30 0.54 [42]
cellulosic cotton fibers
Schiff-base modified guar gum 5.00 120 25 0.24 1.00 0.20 [43]
Hydrolyzed acrylamide-grafted PET films 4.5 40 25 0.49 4.00 0.07 [44]
ZnCl2-MCM-41 6.00 30 20 0.24 0.33 0.43 [45]
Chitosan foam 4.00 2880 20 0.49 0.45 1.74 [40]
Thiol modified activated coke 5.00 480 25 0.49 0.10 2.29 [41]
Glycidylmethacrylate grafted on cellulose 5.00 50 – 0.24 20.00 0.18 [6]
Sheep bone charcoal 3.0 360 25 0.39 4.00 0.06 [46]
MCGMA-I 4.0 300 20 5.00 2.50 1.43 This work
MCGMA-II 4.0 300 20 5.00 2.50 2.02 This work

Material Section) shows the effect of sorbent dosage on the relative 4.0
equilibrium concentration (C/C0). As expected when the sorbent
3.8
dosage increases, the equilibrium concentration of Hg(II)

Released Hg(II) (mM)


decreased. The results shown indicate that the residual concentra- 3.6
tion C/C0 may decrease up to 0.0014 for SD = 10 g L1. At low sor-
bent dosage, all sorption sites are exposed and the can be easily 3.4
saturated. With an increase of sorbent dosage the amount of sorp-
tion sites becomes in excess compared to present metal ions and 3.2
the saturation of the sorption sites is not achieved. According to
3.0 Cycle I
these results, the sorbent dosage of 1.5 g L1 allows achieving the
Cycle II
saturation of the sorbent and makes possible the concentration
2.8
effect and the enrichment of mercury after desorption step. On Cycle III
the opposite hand with a sorbent dosage of 10 g L1 it is possible 2.6
removing about 99.6% of mercury from the solution (achieving
environmental target). The sorbent dosage to be used depends on (a) 0 10 20 30 40 50 60
the target of the sorption process: concentration effect or maxi- t (min)
mum decontamination.
6.0
3.5.8. Comparison of sorption capacity for Hg(II) ions with various 5.8
sorbents
5.6
Released Hg(II) (mM)

Table 9 shows the comparison of maximum sorption capacities


of MCGMA-I and MCGMA-II with a series of values found in the lit- 5.4
erature (together with the best operating conditions reported by 5.2
respective authors). A direct comparison of sorption performance
5.0
is difficult due to different experimental conditions; however, this
is a useful criterion for roughly evaluating the potential of these 4.8 Cycle I
materials. The modified MCGMA sorbents have a sorption capacity 4.6 Cycle II
of the same order of magnitude as other sorbents; although chi-
4.4
tosan foam [40] and thiol-modified activated coke [41] showed Cycle III
better sorption capacity. It is noteworthy that the modified 4.2
MCGMA sorbents have an important advantage related to their fast 4.0
kinetics. The high sorption capacity of the modified MCGMA sor- 0 10 20 30 40 50 60
bents towards Hg(II) ions reveals that sorbents could be promising
(b) t (min)
for practical application in Hg(II) ions removal from wastewater.
Fig. 8. Desorption kinetics for successive sorption/desorption cycles for MCGMA-I
(a) and MCGMA-II (b).
3.5.9. Regeneration
Three successive sorption/desorption cycle runs were per-
formed using 100 mL of 0.1 M KI solutions for 50 min. Samples
were collected at different time intervals and the residual concen- MCGMA-II for the first three sorption/desorption cycles,
tration of Hg(II) was determined. Fig. 8 shows the time course of respectively.
released Hg(II) ions. About 97% of the sorbed Hg(II) ions is released
within 15 min. The desorption is even faster than the sorption.
Though a slight and progressive decrease in the amount of 4. Conclusion
desorbed Hg(II) ions is observed along the successive cycles the
desorption efficiencies remained higher than 90% after three sorp- Novel magnetic macromolecular hybrid materials were
tion/desorption cycles. The sorbent MCGMA-I seems to be slightly prepared by grafting glycidyl methacrylate (synthetic polymer) to
more efficient than MCGMA-II sorbent in terms of desorption with chitosan (biopolymer) and functionalizing the intermediary com-
desorption yield of 97.2, 98.2 for MCGMA-I and 99.6%, 89.0% for pound with diethylenetriamine (directly, or via epichlorhydrin
358 K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359

spacer, MCGMA-I and MCGMA-II, respectively). The two sorbents [12] J.-S. Wang, R.-T. Peng, J.-H. Yang, Y.-C. Liu, X.-J. Hu, Preparation of
ethylenediamine-modified magnetic chitosan complex for adsorption of
are characterized by efficient and selective sorption towards
uranyl ions, Carbohydr. Polym. 84 (2011) 1169–1175.
Hg(II) ions from aqueous medium at approximately pH 4. The [13] L. Zhou, J. Xu, X. Liang, Z. Liu, Adsorption of platinum(IV) and palladium(II)
uptake kinetics are well fitted by the PSORE, while the distribution from aqueous solution by magnetic cross-linking chitosan nanoparticles
of the metal at equilibrium between the solid and the liquid is modified with ethylenediamine, J. Hazard. Mater. 182 (2010) 518–524.
[14] R. Massart, Preparation of aqueous magnetic liquids in alkaline and acidic
modeled by the Langmuir equation. The maximum capacity media, IEEE Trans. Magn. 17 (2) (1981) 1247–1248. IEEE Trans. Magn., 17
reached 1.43 mmol g1 for MCGMA-I and 2.01 mmol g1 for (1981) 1247–1249.
MCGMA-II at 20 °C. The tests performed in the presence of other [15] A.A. Atia, A.M. Donia, K.Z. Elwakeel, Adsorption behaviour of non-transition
metal ions on a synthetic chelating resin bearing iminoacetate functions, Sep.
heavy metals (multicomponent solutions) demonstrated the selec- Purif. Technol. 43 (2005) 43–48.
tivity of the sorbents for Hg(II). Depending on the pH two sorption [16] X. Wang, Z. Zhao, J. Qu, Z. Wang, J. Qiu, Fabrication and characterization of
mechanisms can be involved: ion-exchange on protonated amine magnetic Fe3O4-CNT composites, J. Phys. Chem. Solids 71 (2010) 673–676.
[17] B. Wang, Q. Wei, S. Qu, Synthesis and characterization of uniform and
groups in acidic solutions, or chelation on amine groups at mild crystalline magnetite nanoparticles via oxidation-precipitation and modified
acidic pH. However, this interpretation of sorption mechanisms is co-precipitation methods, Int. J. Electrochem. Sci. 8 (2013) 3786–3793.
affected by the composition of the solution and more specifically [18] D.H.K. Reddy, S.-M. Lee, Application of magnetic chitosan composites for the
removal of toxic metal and dyes from aqueous solutions, Adv. Colloid Interface
the nature of counter-anions present in the solution. The presence Sci. 201–202 (2013) 68–93.
of counter anions that form anionic complexes in acidic solutions [19] L.F. Qi, Z.R. Xu, X. Jiang, C.H. Hu, X.F. Zou, Preparation and antibacterial activity
makes possible Hg(II) sorption while those counter anions that of chitosan nanoparticles, Carbohydr. Res. 339 (2004) 2693–2700.
[20] A.T. Paulino, A.G.B. Pereira, A.R. Fajardo, K. Erickson, M.J. Kipper, E.C. Muniz,
do not form stable complexes with mercury reduce the affinity of
L.A. Belfiore, E.B. Tambourgi, Natural polymer-based magnetic hydrogels:
the sorbent for metal ions. The investigation of temperature effect potential vectors for remote-controlled drug release, Carbohydr. Polym. 90
on Hg(II) sorption isotherm confirms that the sorption mechanism (2012) 1216–1225.
is endothermic. Mercury ions bound to the sorbents can be suc- [21] S.H. Huang, D.H. Chen, Rapid removal of heavy metal cations and anions from
aqueous solutions by an amino-functionalized magnetic nano-adsorbent, J.
cessfully desorbed using 0.1 M potassium iodide solutions and Hazard. Mater. 163 (2009) 174–179.
the sorbent can be effectively re-used. [22] J. Coates, Interpretation of infrared spectra, a practical approach, in: R.A.
Meyers (Ed.), Encyclopedia of Analytical Chemistry, John Wiley & Sons Ltd,
Chichester, U.K., 2000, pp. 10815–10837.
Acknowledgements [23] Y.-T. Zhou, H.-L. Nie, C. Branford-White, Z.-Y. He, L.-M. Zhu, Removal of Cu2+
from aqueous solution by chitosan-coated magnetic nanoparticles modified
with alpha-ketoglutaric acid, J. Colloid Interface Sci. 330 (2009) 29–37.
This study was supported by the French Government through a [24] P. Miretzky, A.F. Cirelli, Hg(II) removal from water by chitosan and chitosan
fellowship granted from the French Embassy in Egypt (Institut derivatives: a review, J. Hazard. Mater. 167 (2009) 10–23.
Français d’Égypte). The authors would like to thank Thierry [25] A.A. Atia, A.M. Donia, K.Z. Elwakeel, Selective separation of mercury (II) using a
synthetic resin containing amine and mercaptan as chelating groups, React.
Vincent, André Brun and Jean-Marie Taulemesse for their technical
Funct. Polym. 65 (2005) 267–275.
and scientific contributions to this work. [26] X. Li, Z. Liu, J.-Y. Lee, Adsorption kinetic and equilibrium study for removal of
mercuric chloride by CuCl2-impregnated activated carbon sorbent, J. Hazard.
Mater. 252–253 (2013) 419–427.
Appendix A. Supplementary data [27] S. Lagergren, About the theory of so-called adsorption of soluble substances,
Kungliga Svenska Vet. 24 (1898) 1–39.
[28] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process
Supplementary data associated with this article can be found, in Biochem. 34 (1999) 451–465.
the online version, at http://dx.doi.org/10.1016/j.cej.2015.05.110. [29] W.J. Weber, J.C. Morris, Kinetics of adsorption on carbon from solutions, J.
Sanitary Eng. Div., ASCE 89 (1963) 31–60.
[30] J. Zeldowitsch, The catalytic oxidation of carbon monoxide on manganese
References dioxide, Acta Physicochim. URSS 1 (1934) 364–449.
[31] A.M. Donia, A.A. Atia, K.Z. Elwakeel, Selective separation of mercury(II) using
[1] ATSDR, Prioritary List of Hazardous Substances, <http://www.atsdr.cdc.gov/ magnetic chitosan resin modified with Schiff’s base derived from thiourea and
SPL/index.html> (accessed 16.04.2015). glutaraldehyde, J. Hazard. Mater. 151 (2008) 372–379.
[2] C. Namasivayam, K. Periasamy, Bicarbonate-treated peanut hull carbon for [32] J.S. Markovski, V. Ðokić, M. Milosavljević, M. Mitrić, A.A. Perić-Grujić, A.E.
mercury (II) removal from aqueous solution, Water Res. 27 (1993) 1663–1668. Onjia, A.D. Marinković, Ultrasonic assisted arsenate adsorption on
[3] H. Demey, T. Vincent, M. Ruiz, A.M. Sastre, E. Guibal, Development of a new solvothermally synthesized calcite modified by goethite, a-MnO2 and
chitosan/Ni(OH)2-based sorbent for boron removal, Chem. Eng. J. 244 (2014) goethite/a-MnO2, Ultrason. Sonochem. 21 (2014) 790–801.
576–586. [33] P.R. Grossl, M. Eick, D.L. Sparks, S. Goldberg, C.C. Ainsworth, Arsenate and
[4] P. Sorlier, A. Denuzière, C. Viton, A. Domard, Relation between the degree of chromate retention mechanisms on goethite. 2. Kinetic evaluation using a
acetylation and the electrostatic properties of chitin and chitosan, pressure-jump relaxation technique, Environ. Sci. Technol. 31 (1997) 321–
Biomacromolecules 2 (2001) 765–772. 326.
[5] T.S. Anirudhan, L. Divya, J. Parvathy, Arsenic adsorption from contaminated [34] B. Dou, V. Dupont, W. Pan, B. Chen, Removal of aqueous toxic Hg(II) by
water on Fe(III)-coordinated amino-functionalized synthesized TiO2 nanoparticles and TiO2/montmorillonite, Chem. Eng. J. 166
poly(glycidylmethacrylate)-grafted TiO2-densified cellulose, J. Chem. Technol. (2011) 631–638.
Biotechnol. 88 (2013) 878–886. [35] F. Raji, M. Pakizeh, Kinetic and thermodynamic studies of Hg(II) adsorption
[6] A.S.K. Kumar, M. Barathi, S. Puvvada, N. Rajesh, Microwave assisted onto MCM-41 modified by ZnCl2, Appl. Surf. Sci. 301 (2014) 568–575.
preparation of glycidyl methacrylate grafted cellulose adsorbent for the [36] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and
effective adsorption of mercury from a coal fly ash sample, J. Environ. Chem. platinum, J. Am. Chem. Soc. 40 (1918) 1361–1402.
Eng. 1 (2013) 1359–1367. [37] H.M.F. Freundlich, Uber die adsorption in lasungen, Z. Phys. Chem. 57 (1906)
[7] G.Z. Kyzas, N.A. Travlou, E.A. Deliyanni, The role of chitosan as nanofiller of 385–470.
graphite oxide for the removal of toxic mercury ions, Colloids Surf. B 113 [38] M.M. Dubinin, E.D. Zaverina, L.V. Radushkevich, Sorption and structure of
(2014) 467–476. active carbons. I. Adsorption of organic vapors, Zh. Fiz. Khim. 21 (1947) 1351–
[8] K.Z. Elwakeel, A.A. Atia, E. Guibal, Fast removal of uranium from aqueous 1362.
solutions using tetraethylenepentamine modified magnetic chitosan resin, [39] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of gases in multimolecular
Bioresour. Technol. 160 (2014) 107–114. layers, J. Am. Chem. Soc. 60 (1938) 309–319.
[9] X.-Y. Huang, X.-Y. Mao, H.-T. Bu, X.-Y. Yu, G.-B. Jiang, M.-H. Zeng, Chemical [40] F.-N. Allouche, E. Guibal, N. Mameri, Preparation of a new chitosan-based
modification of chitosan by tetraethylenepentamine and adsorption study for material and its application for mercury sorption, Colloids Surf. A 446 (2014)
anionic dye removal, Carbohydr. Res. 346 (2011) 1232–1240. 224–232.
[10] P.D. Chethan, B. Vishalakshi, Synthesis of ethylenediamine modified chitosan [41] Z. Li, L. Wu, H. Liu, H. Lan, J. Qu, Improvement of aqueous mercury adsorption
and evaluation for removal of divalent metal ions, Carbohydr. Polym. 97 on activated coke by thiol-functionalization, Chem. Eng. J. 228 (2013) 925–
(2013) 530–536. 934.
[11] G. Dodi, D. Hritcu, G. Lisa, M.I. Popa, Core–shell magnetic chitosan particles [42] M. Monier, I.M. Kenawy, M.A. Hashem, Synthesis and characterization of
functionalized by grafting: Synthesis and characterization, Chem. Eng. J. 203 selective thiourea modified Hg(II) ion-imprinted cellulosic cotton fibers,
(2012) 130–141. Carbohydr. Polym. 106 (2014) 49–59.
K.Z. Elwakeel, E. Guibal / Chemical Engineering Journal 281 (2015) 345–359 359

[43] S. Thakur, S. Kumari, P. Dogra, G.S. Chauhan, A new guar gum-based adsorbent [45] F. Raji, M. Pakizeh, Study of Hg(II) species removal from aqueous solution
for the removal of Hg(II) from its aqueous solutions, Carbohydr. Polym. 106 using hybrid ZnCl2-MCM-41 adsorbent, Appl. Surf. Sci. 282 (2013) 415–424.
(2014) 276–282. [46] A. Dawlet, D. Talip, H.Y. Mi, MaLiKeZhaTi, Removal of mercury from aqueous
[44] N. Rahman, N. Sato, M. Sugiyama, Y. Hidaka, H. Okabe, K. Hara, Selective Hg(II) solution using sheep bone charcoal, Procedia Environ. Sci. 18 (2013) 800–808.
adsorption from aqueous solutions of Hg(II) and Pb(II) by hydrolyzed
acrylamide-grafted PET films, J. Environ. Sci. Health., Part A 49 (2014) 798–806.

You might also like