Study of the adsorption of mercury (II) on lignocellulosic materials under static and dynamic conditions

Accepted Manuscript Study of the adsorption of mercury (II) on lignocellulosic materials under static and dynamic conditions Fabian E. Arias Arias, Amerigo Beneduci, Francesco Chidichimo, Emilia Furia, Salvatore Straface PII:

S0045-6535(17)30522-2

DOI:

10.1016/j.chemosphere.2017.03.137

Reference:

CHEM 19056

To appear in:

ECSN

Received Date: 24 August 2016 Revised Date:

12 December 2016

Accepted Date: 31 March 2017

Please cite this article as: Arias Arias, F.E., Beneduci, A., Chidichimo, F., Furia, E., Straface, S., Study of the adsorption of mercury (II) on lignocellulosic materials under static and dynamic conditions, Chemosphere (2017), doi: 10.1016/j.chemosphere.2017.03.137. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT 1

Study of the adsorption of mercury (II) on lignocellulosic materials under static and dynamic

2

conditions

3 Fabian E. Arias Arias,a Amerigo Beneduci,b,c* Francesco Chidichimo,a,c Emilia Furiab and

5

Salvatore Strafacea,c

6 7

a

8

Bucci, Cubo 41B, 87036 Arcavacata di Rende (CS), Italy

9

b

SC

Department of Environmental and Chemical Engineering, University of Calabria, Via P.

M AN U

Department of Chemistry and Chemical Technologies, University of Calabria, Via P. Bucci,

10

Cubo 15D, 87036 Arcavacata di Rende (CS), Italy

11

c

12

Calabria

SIRiA S.r.l. - Servizi Integrati e Ricerche per l’Ambiente, Spin-off of the University of

13

TE D

*Corresponding author: e-mail: [email protected]

15 16

EP

14

RI PT

4

ABSTRACT

18

WHO has declared mercury as one of the most dangerous pollutants for human health.

19

Unfortunately, several cases of rivers and aquifers contaminated by mercury inevitably poses

20

the problem on how to remediate them. Considerable efforts are being addressed to develop

21

cost-effective methodologies, among which the use of low-cost adsorbing materials. In this

22

paper, the adsorption performances of an alternative lignocellulosic material derived from the

23

Spanish broom plant, are presented. This plant is widely diffused in the world and its usage

24

for Hg(II) removal from water in real working conditions requires only minimal pretreatment

25

steps. A thoroughly investigation on the kinetics and thermodynamics of Hg(II) adsorption on

AC C

17

1

ACCEPTED MANUSCRIPT Spanish broom is presented, by using Hg(II) polluted aqueous solutions specifically prepared

2

in order to simulate typical groundwater conditions. Several batch experiments, under static

3

conditions, were carried out in order to evaluate the effect of pH, contact time, adsorbent

4

dosage, initial concentration, temperature. A maximum adsorption capacity of 20 mg L-1 can

5

be obtained at pH 5, following a pseudo second order kinetics. Moreover, adsorption

6

experiments in dynamic conditions were carried out using Spanish broom filters.

7

Interestingly, a systematic, unconventional double S-shape breakthrough curve was observed

8

under different experimental conditions, revealing the occurrence of two adsorption processes

9

with different time scales. This behavior has been fitted by a bimodal Thomas model which,

SC

RI PT

1

unlike the single Thomas fitting, gives satisfactory results with the introduction of a new

11

parameter related to the fraction of surface active sites involved in the adsorption processes.

M AN U

10

12 Keywords

14

Mercury Pollution, Mercury speciation, Water remediation, Adsorption, Spanish broom

15

lignocellulosic sorbent, Double Thomas model

16

TE D

13

1. Introduction

18

The World Health Organization has declared Hg as one of the most dangerous pollutants for

19

human health (WHO, 1991). It primarily affects the central and peripheral nervous system as

20

well as respiratory and renal systems.

21

Thanks to its chemical-physical characteristics the Hg is very important for the industrials

22

process such as, pharmaceutical, oil refinery, electroplating, battery manufacturing, and

23

mining activities (Manohar et al., 2002). Specifically, gold extraction processes are

24

responsible of an emission in the environment of about 730 tons per year of mercury, which

AC C

EP

17

2

ACCEPTED MANUSCRIPT is, according to the estimates of the United Nations Environment Program (UNEP), 35% of

2

the global amount of mercury released in the environment (UNEP, 2013).

3

Organomercury compounds, such as methylmercury and ethylmercury, are readily absorbed

4

by human organism causing damage of the central nervous system and, in severe cases,

5

blindness, deafness or even coma (Al-Damluji, 1976; Ratcliffe et al., 1996).

6

Several cases of rivers and aquifers contaminated by Hg are ascertained to date and their

7

remediation would be quite expensive. Therefore, considerable efforts are being addressed to

8

develop cost-effective methodologies for the removal of Hg from superficial water and

9

groundwater to prevent sea pollution, fish contamination and hence man poisoning through

SC

RI PT

1

the food chain.

11

Electrochemical reduction and precipitation (Esalah et al., 2000; Lin S.H. et al., 1994;

12

Cantrell K.J. et al., 1995), electrodialysis, coagulation (Mrozowski and Zielinski., 1983;

13

Kumar et al., 2004; Adhoum et al., 2009), flotation, adsorption on surfaces and pores (Naiya

14

et al., 2009; Zhang et al., 2005; Rao et al., 2009), ion-exchange (Visa, 2016 ; Alyüz and Veli

15

2009), are among the developed techniques used for heavy metals removal from aqueous

16

solutions.

17

Adsorption through chemical-physical mechanisms using materials based on zero valent iron

18

(Cantrell et al., 1995; Weisener et al., 2005; Liu et al., 2014; Wilkin and Neil, 2003), zeolites

19

(Visa, 2016, Murthy et al., 2013; Alver and Metin, 2012; Wang and Peng, 2010), active

20

carbons (Zhang et al., 2005; Rao et al., 2009, Zabihi et al., 2010; Juang et al., 2002), and ion

21

exchange resins (Chiarle et al., 2000), have shown good efficacy in the removal of various

22

heavy metals, including Hg(II) (Gupta et al., 2015; Zhang et al., 2005; Rao et al., 2009; Liu et

23

al., 2014; Murthy et al., 2013; Zabihi et al., 2010, Namasivayam and Kadirvelu, 1999).

24

Synthetic porous organic polymers-based mercury nano-traps, have also been shown to have

25

high efficiency and efficacy in the removal of Hg(II) from water (Li et al., 2014). However,

AC C

EP

TE D

M AN U

10

3

ACCEPTED MANUSCRIPT all of these methods require the preliminary activation or the synthesis of the adsorbent

2

materials through several steps, leading to rather costly and/or poor green technologies.

3

Therefore, attention has been focused on several potentially low-cost sorbents (Bailey et al.,

4

1999) among which lignocellulosic materials derived from plants as by-products or as waste

5

from other production chains (agroindustry, paper, sugar production, etc.) (Namasivayam and

6

Kadirvelu, 1999; Ho and Wang, 2008; Mahajan and Sud, 2013; Khoramzadeh et al., 2013)

7

seem very promising. Lignocellulose is the most abundant raw material on Earth and is

8

composed of cellulose, lignin, hemicellulose and pectin (Khachatourians and Arora, 2001)

9

thus containing either carboxyl or hydroxyl functional groups at its surface (Henriksonn,

SC

RI PT

1

2009), which are in principle able to bind various heavy metals through chemical-physical

11

adsorption (Lv et al., 2012).

12

Here we study the performances of the plant Spanish Broom as adsorbent lignocellulosic

13

material for mercury removal from water. The common broom, also called fragrant broom or

14

Spanish broom (Spartium junceum) is a plant belonging to the Febacaea family. It is a native

15

species of the Mediterranean area, which goes from South Europe to North Africa up to the

16

Middle East zones, also reaching some areas of the central and South America. It grows in

17

sunny areas up to an altitude of 1200 m asl, characterized by arid and sandy soils. It is a

18

shrubby (from 0.5 to 3.00 m high) and perennial plant with long stems. The stems are green

19

and cylindrically shaped, compressible, but tensile resistant, straight and highly branched. The

20

sporadic foliage is almost absent, while the bright yellow flowers are often located in the

21

terminal part of the branches. Historically, the Spanish broom was a highly exploited plant

22

and even nowadays it is used for several purposes: traditionally used in the ornamental trade,

23

for yellow dye extraction, as well as for cellulose fibers production (Gabriele et al., 2010),

24

reforestation of degraded areas, consolidation of high slope terrains, extraction of essences

AC C

EP

TE D

M AN U

10

4

ACCEPTED MANUSCRIPT from flowers for perfumes creation, extraction of fibers for fabrics production or building

2

materials reinforcement. (Chidichimo et al., 2015).

3

In spite of the multiple uses for which this plant was employed, this study focuses on the

4

investigation of the broom adsorption properties with respect to mercury(II). SB has many

5

practical advantages as raw adsorbent material, due to its widespread diffusion, the very easy

6

as well as economic pretreatment steps for its operation and the exceptional mechanical

7

properties needed for setting up filters such as permeable reactive barriers. In addition, it

8

served us to thoroughly investigate the adsorption performances of Hg(II) on a typical raw

9

lignocellulosic material under either static or dynamic experimental set-ups and using Hg(II)

SC

polluted aqueous solutions under conditions typically found in groundwater.

11

M AN U

10

RI PT

1

2. Materials and Methods

13

2.1. Spanish broom characterization

14

2.1.1 Fourier Transform Infrared Spectroscopy (FTIR) analysis

15

The chemical composition of the broom used in the present study has been previous

16

determined (Gabriele et al., 2010): 44.5 (0.2)% of Cellulose; 18.5 (0.3)% of Lignin; 16.3

17

(0.1)% of Pentosans; 13.3 (0.1)% of Pectins and 4.0 (0.2)% of Ash.

18

To characterize the SB fibers FTIR spectra between 375 and 4000 cm-1 were directly

19

collected on dry samples by a Bruker ALPHA FT-IR spectrometer equipped with a A241/D

20

reflection module.

EP

AC C

21

TE D

12

22

2.1.2. Scanning Electron Microscopy (SEM) analysis

23

The structure of the Spanish broom fibers was analyzed by a SEM LEO 420 (LEO Electron

24

Microscopy Ltd., Cambridge, England). Fiber samples were gold sputter-coated with an Auto

25

Sputter Coater, AGAR. Images were acquired under an accelerating voltage of 15.00 kV.

5

ACCEPTED MANUSCRIPT 2.1.3. X-ray Fluorescent Spectroscopy (XRF) analysis

2

In order to show the presence of mercury on the lignocellulosic fibers, XRF analysis was

3

performed with a Bruker Tracer III, equipped with a Rh/Pd source, at a voltage of 40kV, a

4

current of 11µA and an acquisition time of 15sec. Spanish broom samples,

5

before and after suspension in Hg(II) solution, were dried in a desiccator, and pelleted to form

6

disks that were subsequently placed directly onto the probe of the instrument for the

7

measurement.

RI PT

1

9

SC

8 2.2. Spanish broom pretreatment and solutions preparation

Before its usage in the experimental tests, the Spanish broom was first washed with distillated

11

water and dried for five hours at 105 °C, to remove its water content, then it was grinded by

12

an industrial mill, up to an average size of about 5 mm.

13

All chemicals and reagents used for the experiments had an analytical grade of purity.

14

Ultrapure water (18.3 MΩ cm, Arioso-Human Corporation) was used for all the solutions. A

15

stock solution of Hg(II) with a concentration of 1000 mg L-1 was prepared starting from

16

Hg(NO3)2 •H2O (Sigma-Aldrich, 99.99 % purity). Chloride ions at a concentration of 150 mg

17

L-1 were also added to the stock solution through the addition of concentrated HCl. This was

18

done in order to have Cl- ions concentration in the range of the groundwater samples collected

19

in important aquifers involved by Hg pollution (INIGEMM, 2013). The working standard

20

solutions, adopted in the tests, were prepared by means of serial dilutions of the Hg stock

21

solution. HNO3 and NaOH solutions were used to adjust the pH of these solutions.

22

Hg concentration in the samples collected during the experimental activities, was measured by

23

an Inductively Coupled Plasma Mass Spectrometer (ICP/MS, iCapQ Thermofischer). The

24

instrument was calibrated using different analytical standards concentrations (TraceCERT®

25

certified reference materials (CRM)): 1, 5, 10 and 20 ppb. A pH meter (HI221, Hanna

AC C

EP

TE D

M AN U

10

6

ACCEPTED MANUSCRIPT 1

Instruments), calibrated with two different buffer solutions (pH = 4.01 and 7.00 ) was used.

2

All samples were purified with a 0.45µm filter and then acidified with HNO3 (pH = 2) before

3

ICP/MS analysis.

4 2.3. Potentiometric titrations

6

The potentiometric titration experiments were carried out through a Hanna 22001

7

potentiometer with temperature control. Titrations were performed on suspensions of the

8

absorbent material (2 g L-1) or on blank control samples not containing the absorbent. The

9

suspensions were acidified using HNO3 at pH 3 and then the suspensions were slowly back-

10

titrated with standard 0.1 M NaOH up to about pH 10. The Gran’s plot methods was used for

11

precise end points determination of Spanish broom acidic groups (Gran, 1952; Rossotti and

12

Rossotti, 1965). The ionic strength was maintained constant throughout the titration

13

experiments by the addition of NaNO3 (0.1 M), according to the procedures previously

14

described by Guo et al. (2008) and Niu et al. (2009).

SC

M AN U

TE D

15

RI PT

5

16 17 18

2.4. Speciation study of Hg(II) in the presence of chloride

19

standardized as described previously (Porto et al., 2005; Khalil and Radalla, 1998). Mercury

20

(II) nitrate stock solution was prepared and standardized according to Esteban et al. (1986).

21

The hydrogen ion concentration in the metal cation stock solution was determined

22

potentiometrically with a glass electrode using Gran’s method (Gran, 1952; Rossotti and

23

Rossotti, 1965). All solutions were prepared by using ultrapure water with a resistivity of 18.2

24

MΩ cm, obtained from a Milli-Q plus system (Millipore, Bedford, MA, USA).

25

The complexation equilibria were studied in 0.1 M NaNO3 and at 25°C by measuring with a

26

glass electrode the competition of chloride for H+ and the metal cation. The cell arrangement

AC C

EP

Hydrochloric acid, sodium hydroxide and sodium nitrate stock solutions were prepared and

7

ACCEPTED MANUSCRIPT and the electrodes employed were described in a previous work (Furia et al., 2009). A slight

2

flow of nitrogen gas was passed through three bottles (a–c) containing: (a) 1 M NaOH, (b) 1

3

M H2SO4 and (c) 0.1 M NaNO3, and finally into the test solutions, continuously stirred during

4

the potentiometric measurements. The metal (CM) and ligand (CL) concentrations were ranged

5

from 0.5 to 5 mM, the ligand−to−metal ratio was varied between 1 and 10 and pH was ranged

6

from 2 to 9. The test solution’s acidity was varied stepwise by adding CB M NaOH stock

7

solution.

RI PT

1

2.5. Adsorption Experiments

11

The effect of pH on Hg adsorption, was studied by performing batch tests in which 0.5 g of

12

Spanish broom were suspended in 50 mL of 80 mg L-1 Hg(II) solution at room temperature.

13

The pH was varied in the range 2-9 by adding HNO3 or NaOH solution.

14

Adsorption kinetics and the influence of the contact time were evaluated in similar batch tests

15

using the following conditions: 1) initial Hg(II) concentration = 100 mg L-1, solution volume

16

= 0.400 L, mass of sorbent = 40.00 g; 2) initial Hg(II) concentration = 200 mg L-1, solution

17

volume = 0.400 L, mass of sorbent = 20.00 g; 3) initial Hg(II) concentration = 100 mg L-1,

18

solution volume = 0.400 L, mass of sorbent = 2.00 g, at pH 5.02 ± 0.3.

19

Adsorption isotherms at different temperatures were obtained from similar batch experiments

20

in which 0.5 g of Spanish broom were suspended in 50 mL of Hg(II) solutions with different

21

concentrations in the range 1-180 mg L-1. The suspensions, contained in 50 mL Falcons, were

22

immersed in a water bath and incubated in an oven at the set temperature for eight hours.

23

During the incubation time, each suspension was subjected to vigorous agitations by inversion

24

for 30 s every hour.

25

These experiments were also useful to assess the influence of the Hg(II) initial concentration

26

on the Spanish broom adsorption capacity and removal efficiency.

AC C

EP

TE D

M AN U

SC

8 9 10

8

ACCEPTED MANUSCRIPT Another set of batch experiments were also carried out in order to test the effect of the

2

increasing adsorbent dosage (mass of Spanish broom) on the adsorption capacity and mercury

3

removal efficiency. In each experiment, the Hg(II) solution concentration was fixed at 80 mg

4

L-1 and gradually increasing the amounts of Spanish broom, in step of 0.5 g, from 0.5 g to 7.0

5

g. Each amount was suspended in 50 mL of the above Hg(II) solution.

6

To analyze the results, the adsorption capacity, i.e., the amount of adsorbed mercury per unit

7

mass of adsorbent (qt) was determined:

8


 =

9

where C0 and Ct are the initial and mercury ion concentrations (mg L-1) at time t, respectively,

(1)

M AN U



SC

 

RI PT

1

V is the volume of the solution (L) and W is the adsorbent mass (g). When the adsorption

11

equilibrium condition holds, Ct = Ce and qt = qe where, Ce is the equilibrium concentration

12

and qe is the equilibrium adsorption capacity.

13

The Hg(II) removal efficiency (RE%), defined by the equation 2, was also used to assess the

14

adsorption performances of the fiber.

15

% =



 100

(2)

EP

16

 

TE D

10

2.6. Flow cell experiments

18

Adsorption experiments were also carried out by pumping the Hg(II) solution within a

19

Spanish broom filter. The grinded plant was packed into a clear acrylic column (flow cell) of

20

15 cm length and an inner diameter of 56 mm. Packing was conducted taking care to create an

21

homogeneous porous medium. At both ends of the column, the broom was retained by a non-

22

woven membrane, a metal plate 2-mm thick with 2.5-mm wide openings and a closing flange.

23

The aforementioned configuration ensures that the entire cross section of the filtering material

24

is affected by the flow of the solution. The inlet flow rate (at the bottom of the column) was

AC C

17

9

ACCEPTED MANUSCRIPT fixed by a peristaltic pump (model PD 5101, Heidolf). The outlet flow rate (at the top) was

2

sampled in time to trace the breakthrough curves describing the adsorption performances of

3

the broom in the different conditions adopted for each experiment (Figure S1). For this

4

purpose different amounts of adsorbent material were used (80 g and 115 g) and subjected to

5

different flow rates (19, 15, and 10 mL min-1) having the same mercury concentration of 100

6

mg L-1.

7

The following tests were designed and performed (mass of adsorbent material – flow rate): a)

8

80 g – 19 mL min-1; b) 115 g - 19 mL min-1; c) 115 g - 15 mL min-1 and d) 115 g – 10 mL

9

min-1. Sampling was realized every 30 minutes.

M AN U

10

SC

RI PT

1

2.7 Curve-Fitting procedure and statistical analysis.

12

The curve-fitting procedures were performed by the Levenberg-Marquardt nonlinear least

13

squares method. It involves an iterative improvement to parameter values in order to reduce

14

the sum of the squares of the errors between the function and the measured data points. At

15

each iteration, the parameters used in the model were varied in order to minimize the chi-

16

square (  ):

17

  =   ∑   −  ; " , " , … %

(3)

18

&' () 

(4)

19

where, y is the dependent variable, x the independent variables, neff is the total number of

20

experimental points used in the fitting, p is the total number of adjustable parameters used in

21

the fitting, p1 , p 2 ,... are the model parameters to be varied in order to minimize χ 2 .

22

As a measure to determine the goodness of fit the adjusted r-square (R2) was used. The closer

23

the fit is to the data points, the closer the adjusted r-square will be to the value of 1.

EP

AC C



TE D

11

10

ACCEPTED MANUSCRIPT Standard error on each fitted parameters was calculated in the fitting procedure by equation 4

2

where, Cii is the diagonal element of the variance-covariance matrix.

3

Each experiment has been replicated at least three times. The data points on each graph

4

represent the average value calculated over the replicates and the error bars are the

5

corresponding standard deviations.

6 7 8

SC

RI PT

1

3. Results 3.1. Spanish broom characterization

11

Figure S2 shows the typical FTIR spectrum of Spanish broom (SB). The peaks picked are

12

characteristic of lignocellulosic materials (Casas et al., 2012). Clearly distinguishable are the

13

peaks related to the cellulose fraction and those of the lignin one. Thus, the peaks at 1169 cm-1

14

and at 898 cm-1 are attributed to the C – O – C asymmetric stretching vibration and C1 – H

15

deformation in cellulose. On the other hand, the intense peak at 1741 cm-1 is related to the C =

16

O stretching vibrations of the carboxyl groups of lignin. Moreover, vibrations at 1330 cm-1

17

and 830 cm-1 are typical of the syringyl unit, while the guaiacyl ring vibration occurs at 1511

18

cm-1. Other characteristic peaks of lignin are that at 1383 cm-1 attributed to the phenolic

19

hydroxyls and the small peak at 1511 cm-1 related to the aromatic skeletal vibration.

20

The ultrastructure of SB has been analysed by SEM. A representative images of the structure

21

of the lignocellulosic fiber is reported in Figure S3. It shows that the SB fibers are organized

22

in parallel bundles of cylindrical microfibers glued together by lignin. Each microfibers has a

23

diameter of about 4 µm.

AC C

EP

TE D

M AN U

9 10

24 25 26 11

ACCEPTED MANUSCRIPT 3.2. Effects of various parameters on Hg(II) removal

2

3.2.1 pH Influence

3

Figure 1 shows the influence of initial pH on the adsorption capacity of the Spanish broom.

4

As can be seen, adsorption increases with pH starting from removal percentages of about 67

5

% at pH 2 up to about 85% for pH ≥ 6. There is a step-wise increase of the removal efficiency

6

% above pH 4 and another stepwise increase above pH 7 that can be only seen at its initial

7

stage in Figure 1. This is in agreement with the titration curve of the Spanish broom, which

8

shows two equivalent points at pH 4.25 and 7.48, as determined by the Gran plot’s method

9

(Rossotti and Rossotti, 1965; Guo et al., 2008) (Figure S4). These points can be attributed to

10

the ionization of the carboxyl and phenolic hydroxyl groups of the lignin, respectively (Lv et

11

al., 2012; Guo et al 2008). It is clear that, as the acidic functional groups present on the broom

12

are ionized due to the increase of the pH, the adsorption efficiency increases.

13

This effect can be partly explained by the increase of the negative charge on the surface of the

14

material that is therefore able to more efficiently attract positive ions and partly by the ability

15

of the oxygen atoms to make stable complexes with the mercury (II) species present in

16

solution. The adsorption properties of the broom are due to its composite material nature,

17

where cellulose and lignin play the major role. Indeed, metal cations can interact with the

18

hydroxyl groups of cellulose and lignin as well as with the carboxyl functional groups of

19

lignin. The chelating action of hydroxyl and carboxyl groups toward several metals, is known

20

(Porwal et al., 2015; Beneduci, 2011; Furia and Sindona, 2012; Adamo, et al. 2009; Furia et

21

al., 2013). Specifically, the monomers constituting the lignin polymer, p-Coumaryl alchol,

22

Conferyl alchol, and their derivatives (Henriksonn, 2009), provide exceptional moieties for

23

heavy metal binding under different pH and other solution conditions (Furia and Sindona,

24

2012; Furia et al., 2013; Porwal et al., 2015).

AC C

EP

TE D

M AN U

SC

RI PT

1

12

ACCEPTED MANUSCRIPT Another important aspect to be considered when the influence of the pH is analyzed with

2

respect to the mercury adsorption, is the Hg(II) speciation within the solution. Previous

3

studies have shown that, in the absence of chelating agents, Hg+2 is the dominant species in

4

solution at pH < 3, Hg(OH)2 is the dominant one at pH > 5, while for pH values between 3

5

and 5, both species coexist (Zhang et al., 2005). Moreover, small traces of HgOH+ ranging

6

from 1% to 13% of the total mercury Hg(II), were also detected for pH values between 2 and

7

6. When agents able to complex the Hg(II) ions are present in solution, the speciation of

8

Hg(II) significantly changes. As shown in Figure 1b, Hg(II) forms several stable products

9

related either to the equilibrium of hydrolysis of Hg2+ (Hg(OH)+, Hg(OH)2, Hg(OH)3-) or to

SC

RI PT

1

the complexation equilibrium with the chloride (HgCl+, HgCl2, HgCl3- and HgCl42-), together

11

with the mixed species Hg(OH)Cl (Supplementary Information). It can be seen that, for pH

12

values between 3.5 up to 5.5, the predominant species is the HgCl2 complex.

13

AC C

EP

TE D

M AN U

10

14

Figure 1. (a) Hg(II) adsorption as a function of the initial pH (C0 = 80 mg L-1; W = 0.5 g, T =

15

298 K). (b) Distribution diagram of Hg2+ in the presence of chloride (CM = 5.10-4 M and CL =

16

4.10-3 M).

17

13

ACCEPTED MANUSCRIPT It is worth noting that all the adsorption experiments were carried out at pH = 5 and in the

2

presence of chloride. These conditions served to approximate those of typical groundwater

3

(INIGEMM, 2013), where the predominant species is HgCl2 (Figure 1b). Therefore, all the

4

results reported in this work mainly reflect the adsorption of HgCl2 and represent a lower limit

5

of adsorption performances for SB. At increasing pH, the RE% increases (Figure 1a), but

6

other species come into play in the adsorption mechanism.

7

RI PT

1

3.2.2. Effect of the contact time

9

Figure 2 shows the trend of the Hg(II) concentration as a function of the contact time between

M AN U

SC

8

the solution and the broom. There is a fast Hg concentration reduction during the first 60

11

minutes (Figure 2a), which leads to a removal percentage of over 70% in both tests (Figure

12

2b). The rate of mercury removal becomes subsequently much slower, until it reaches a

13

plateau value, corresponding to the equilibrium condition. As can be seen from Figure 2, the

14

equilibrium time, i.e. the time required to reach the maximum Hg(II) adsorption, it is not

15

influenced by the Hg(II) concentration or by the amount of adsorbent material. Indeed, in both

16

cases the equilibrium was reached after about 480 min (8 hours), resulting in an adsorption of

17

more than 86%. Mercury can be easily detected on the dry fiber after the adsorption

18

experiment by XRF (Figure 2c), thus confirming the adsorption of mercury by the

19

lignocellulosic material.

EP

AC C

20

TE D

10

14

1

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

2

Figure 2. Removal efficiency (RE%) (a) and Hg(II) concentration (Ct) as a function of the

3

contact time (b), at two different Hg(II) initial concentration (Hg 100 and 200 mg L-1). XRF

4

spectra of the Spanish broom before (black line) and after adsorption (red line) (c).

5 6

15

ACCEPTED MANUSCRIPT 3.2.3. Effect of the initial concentration and of the adsorbent dosage

2

The adsorption capacity (qe) of the Spanish broom increases quite linearly with the initial

3

concentration of Hg(II) in solution (C0) almost in the whole range of initial concentrations

4

considered, though at high concentrations a deviation from linearity does occur (Figure S5).

5

This suggests that the Spanish broom has a limited number of adsorbent sites, which is fixed

6

by its amount and by the experimental conditions adopted (mainly pH, temperature, solution

7

volume/adsorbent mass ratio, etc.). Thus, qe increases as long as free active sites are available

8

on the broom. When all the active sites are involved by the adsorption process, the saturation

9

and therefore the maximum adsorbent capacity per unit mass of adsorbent material (qm), is

M AN U

SC

RI PT

1

achieved.

11

Due to the above trend, the adsorption efficiency of the broom, defined as the percentage of

12

mercury removal from the solution, is almost independent of C0, assuming an average value

13

of (67 ± 5) %, (Figure 3a). In contrast, the equilibrium concentration of Hg(II) in solution

14

monotonically increases with C0 (Figure 3a). However, it is worth noting that at low initial

15

concentrations (< 10 mg L-1), the equilibrium concentration falls in the ppb range (Figure 3a).

AC C

EP

TE D

10

16 17

Figure 3. (a) Effect of the initial concentration on the adsorption process (W = 0.5 g, V= 50

18

mL); (b) effect of the adsorbent dosage on the efficiency of the adsorption process (C0 = 80

19

mg L-1, V= 50 mL).

16

ACCEPTED MANUSCRIPT 1 It is expected that the percentage of mercury removal depends on the adsorbent dosage, being

3

the other conditions the same (Rao et al., 2009). This is actually the case, as illustrated in

4

Figure 3b. Adsorption efficiencies as high as 87 % were found at 7 g of Spanish broom

5

suspended in 50 mL of a 80 mg L-1 solution of Hg(II).

6

It is also evident from the plot in Figure 3b that the efficiency increases with the increase of

7

the mass of Spanish broom. On the other hand, the adsorption capacity has an opposite trend

8

and decreases with the adsorbent mass. Both these trends strongly indicate a situation far from

9

the saturation, due to the large excess of available active sites. Moreover, it is interesting to

10

note that the above trends are nonlinear and both the observables seem to tend to a plateau

11

value that is determined by the thermodynamic of the adsorption processes.

12

M AN U

SC

RI PT

2

3.3. Adsorption kinetics

14

The mechanism of adsorption of ions and other substances such as dyes on solid porous

15

surfaces (particle), generally involves several steps characterized by different rates, in which

16

the solute diffuses into the adsorbent material. The complexity of the adsorption process can

17

be fully displayed in a qt vs. t1/2 plot, according to equation (3) (Weber and Morris, 1962),

18

generally showing a multilinearity in the set of kinetic data in any adsorption experiment

19

where intraparticle diffusion (IPD) occurs (Wu et al., 2009). Figure 4a shows the IPD plot for

20

the Hg(II) on the Spanish broom fiber. As can be observed, three linear regions were

21

identified corresponding to different diffusion steps. The initial, most rapid one, is related to

22

the external surface adsorption during which the Hg(II) species moves from the bulk of the

23

solution to the fiber surface. The second linear region may be related to the gradual diffusion

24

of the solute into the pores of the lignocellulosic matrix. The final equilibrium process may

AC C

EP

TE D

13

17

ACCEPTED MANUSCRIPT involve a very slow diffusion of the Hg(II) species from larger pores to smaller ones or from

2

lignin rich to cellulose rich regions, being the adsorbent material a lignocellulosic composite.

TE D

M AN U

SC

RI PT

1

3

Figure 4. Adsorption kinetics. (a) Intraparticle diffusion plot showing three regions of

5

linearity (C0 = 200 mg L-1, V = 0.100 L, W = 5 g, T = 298 K). (b) Pseudo second order plots

6

showing the fitting between the experimental data points, obtained for three different

7

adsorbent material masses, and the linear function described by eq. (9). The working

8

conditions for each set of data are reported in Table 1.

AC C

9

EP

4

10

Each linear region can be fitted by eq. (5) giving the kinetic constant kp for each process.

11


 = * +/ + )

(5)

18

ACCEPTED MANUSCRIPT 1

where, * is the rate constant of intraparticle diffusion expressed in mg g-1-min-1/2 and C (mg

2

g-1), the intercept at qt = 0, is a constant for the process.

3

The contribution of the first diffusion step can be evaluated by the characteristic ratio or initial

4

adsorption ratio Ri given by equation (6):

5

 =

6

where,
012 is the adsorption capacity at the longest time and C can be related to the boundary

7

layer thickness (Wu et al., 2009).

8

The values of kp and Ri are reported in Table 1 for three batch experiments which only differ

9

for the employed mass of fiber. The calculated Ri ratio is less than 0.5 (Table 1) indicating

10

that the adsorption kinetics has a strong extent of surface adsorption that accounts for more

11

than 80% increase of qt (Juang et al., 2002; Tunali et al., 2006), that is to say that most of the

12

adsorption of Hg(II) occurs on the surface of the Spanish broom fiber.

13

The IPD model provides a fractional description of the adsorption kinetics of Hg(II) species

14

on the fiber, highlighting that a variety of processes takes place according to the existence of

15

active binding sites either on the surface of the fiber or at its interior. However, it does not

16

provide a whole description of the entire set of kinetic data. A complementary approach is to

17

calculate the number of free active sites per unit mass of fiber at time t (Sf) by assuming a

18

pseudo first order or a pseudo second order kinetics (Ho, 2006). These models rely on the

19

assumption that the following adsorption equilibrium holds:

20

Sf + Hg(II) = S-Hg(II)

21

where, S-Hg(II) is the mercury-fiber complex formed upon adsorption, corresponding to the

22

number of occupied active sites per unit mass of fiber (So).

23

The pseudo first order kinetics (Lagergren, 1898), assumes that the rate of change of the

24

adsorption capacity is proportional to the concentration of available active sites per unit mass

25

of adsorbent material (eq. 7);

./ 

AC C

EP

TE D

M AN U

SC

RI PT

(6)

./

(equilibrium 1)

19

ACCEPTED MANUSCRIPT 1

3. 3

= * 
1 −


(7)

where,
 is the adsorption capacity at time t and
1 is the equilibrium adsorption capacity.

3

This model failed to reproduce the entire set of data (Figure S6), in agreement with similar

4

observation on different adsorbent materials (Zabihi et al., 2010). In contrast, very good

5

fitting results were obtained by a pseudo second order rate model (Table 1). It assumes that

6

the rate of change of the concentration of occupied active sites per unit mass of the adsorbent

7

material, is proportional to the square of the concentration of free active sites per unit mass of

8

sorbent (Ho and McKay, 1999; Kumar, 2006). In terms of adsorption capacity, the pseudo

9

second order rate equation can be written as (Ho, 2006):

SC

RI PT

2

3. 3

11

where, k2 is the pseudo-second order rate constant (g mg-1min-1) scaled by the number of

12

active sites per unit mass of fiber,
1 and
 are the concentrations of Hg(II) species per unit

13

mass of fiber (mg g-1) at equilibrium and at time t, respectively. It can be shown that equation

14

(6) can be derived from a first order kinetics of the rate of the adsorption process with respect

15

to both the concentration of Hg(II) in solution and the concentration of free active sites on the

16

adsorbent material. Integration of equation (8) and linearization of the integrated equation,

17

leads to equation (9) which is one of the linear forms that can be advantageously used to fit

18

the data by plotting the t/qt ratio as a function of t (Kumar, 2006):

19

 .

20

The key parameters k2 and
1 can then be directly extracted from the linear fitting (Figure

21

4b), whose results are reported in Table 1.

22

In agreement with the results discussed before in Figure 3b, the equilibrium adsorption

23

capacity decreases with the increase of the adsorbent mass. In contrast, the pseudo second

M AN U

10

= * 
1 −
 

AC C

EP

TE D

(8)

=4



5 5 .



+. + 

(9)

20

ACCEPTED MANUSCRIPT 1

order rate constant decreases with the reduction of the mass of fibers due to the decrease of

2

the total number of active sites.

3

Fitting results of the pseudo second order kinetic modela IPD model Kp

Pseudo second order model Ri

R2

(mg g-1 min-1/2)

qe (mg g-1)

k2 g mg-1 min-1

SC

W (g)

RI PT

Table 1

R2

0.040 (0.004)

0.36 0.9493

0.137 (0.031)

0.908 (0.003)

0.9998

20.00c

0.15 (0.04)

0.31 0.7643

0.0280 (0.007)

3.45 (0.02)

0.9995

2.00d

0.10 (0.02)

0.22 0.9025

0.0193 (0.006)

5.29 (0.04)

0.9997

a

M AN U

40.00b

Values in parentheses are the standard deviation of the data. bC0 = 100 mg L-1, V= 0.400 L, W

= 40.00 g; cC0 = 200 mg L-1, V= 0.400 L, W = 20.00 g; dC0 = 100 mg L-1, V= 0.400 L, W =

4

TE D

2.00 g;

The above results support a chemisorption mechanism in which the interaction between the

6

functional groups of the sorbent and HgCl2, leads to the formation of mercury-lignocellulosic

7

complexes (Sf-Hg(II)), as supposed with the equilibrium (1). This interaction may involve

8

either the hydroxyl groups of cellulose and lignin or the carboxyl groups of lignin. However,

9

it is known that cellulose is a poor sorbent of heavy metal cations and that it should be

AC C

EP

5

10

activated by chemical modification in order to improve its adsorption capacity (Demirbas,

11

2008). This is due to the unfavorable interaction between the soft Hg(II) metal (Pearson,

12

1963) and the hard hydroxyl groups of cellulose. On the other hand, Hg(II) may better interact

13

with the more acidic carboxyl groups of lignin. This hypothesis is in agreement with the fact

14

that lignin can form both monodentate and bidentate complexes with HgCl2 through the

15

involvement of both carboxyl and phenol binding sites (Lv et al., 2012). This mechanism 21

ACCEPTED MANUSCRIPT should lead to a pH decrease following adsorption due to the concomitant release of chloride

2

acid into the aqueous medium. We verified this mechanism by observing that the final pH of

3

the solution after adsorption is lower than the initial one (pH = 5), ranging from 4.3 to 4.8 and

4

systematically decreases with the increase of the initial Hg(II) concentration (Table S2). This

5

indicates that the above reaction do effectively occurs with the release of hydrochloric acid in

6

solution. Thus, the complexation of HgCl2 by the lignocellulosic fibers, evidently involves the

7

acidic functional groups of the fibers, mainly located on lignin.

RI PT

1

SC

8 9 3.4. Asdorption isotherm

11

The plot of the adsorption capacity (qe) versus the equilibrium concentration of a solution (Ce)

12

at constant temperature is called adsorption isotherm. The mercury adsorption isotherms at

13

several temperatures are reported in Figure 5. They show a rather rapid increase of the

14

adsorption capacity for low Ce values and a decrease of the slope at higher equilibrium

15

concentrations, suggesting the existence of plateau values for the adsorption capacity.

AC C

EP

TE D

M AN U

10

16 17

Figure 5. Experimental adsorption isotherms for mercury at different temperatures and

18

nonlinear curve fitting with the Langmuir (a) and Freundlich model (b).

19

22

ACCEPTED MANUSCRIPT This trend should correspond to an L-type isotherm according to Giles et al. (1974), which

2

can be suitably described by the Langmuir model (eq. 10) (Limousin et al., 2007; Atkins et

3

al., 2006). This model assumes that the adsorption process is restricted to a monolayer surface

4

coverage on a uniform surface exhibiting equivalent binding sites and that there are not

5

interactions among the adsorbed solute species that may affect the adsorption mechanism

6

(Atkins et al., 2006). Due to the limited number of active sites on the surface of the adsorbent

7

material, the maximum adsorption capacity, qm, can be predicted.

8


1 =

9

where, 9: is the Langmuir adsorption constant related to the equilibrium distribution constant

.6 47 

SC

(10)

847 

RI PT

1

93 at very low qe far from the saturation (eq. 11):

11

9: = 

12

The results of the nonlinear curve fitting with the Langmuir model are summarized in Table

13

2.

14

Good results were obtained also with the Freundlich model (Table 2 and Figure 5b), which

15

does not introduce any limitation on the mechanism of surface coverage (eq. 12) (Limousin et

16

al., 2007; Haghsererth and Lu, 1998):

17


1 = 9? )1 /

18

where, KF is the Freundlich constant related to the distribution constant and n is a parameter

19

related to the surface heterogeneity. Values of n close to zero indicate strong surface

20

heterogeneity (Haghsererth and Lu, 1998). It is also evident, from eq. 10 and eq. 12, that the

21

Freundlich and the Langmuir models become equivalent if n = 1 and 9: )1 ≪ 1.

22

The mechanism of monolayer adsorption is further supported by the fact that a multilayers

23

model, such as that of Elovich (Hamdaoui and Naffrechoux, 2007), is not able to reproduce

24

the experimental set of data (Supplementary Information, Figure S7).

.

.

 .6

4

.

= . < , in the limit
1 ≪
> ; with 93 =  6



(11)

(12)

AC C

EP

TE D

 .6 .

≅

M AN U

10

25 23

ACCEPTED MANUSCRIPT Table 2 Langmuir and Freundlich adsorption isotherms parameters Langmuir model qm

3 -1

KF

R2

T (K) -1

1/n

(mg g )

g L )

294

315 (6)

20 (4)

0.96038

0.6 (0.2)

301

307 (5)

20 (1)

0.99752

0.58 (0.07)

311

223 (3)

20 (4)

0.97901

0.43 (0.09)

328

213 (4)

28 (11)

0.95729

0.5 (0.1)

1.5 (0.2)

0.93719

1.46 (0.07)

0.99090

1.4 (0.1)

0.97997

1.4 (0.1)

0.96730

SC

(mg

-1

(cm g )

Calculated with eq. (10); 93 becomes dimensionless if the density of the solution is

M AN U

(a)

R2

n (1-n)

RI PT

Kd (a)

Freundlich model

assumed equal to that of water. Values in parentheses are the standard deviation of the data. 1 2

3.4.1. Thermodynamic of the adsorption

4

The Langmuir model has been used to extract the thermodynamic quantities related to the

5

adsorption isotherms. The distribution constant Kd has been used to calculate the Gibbs free

6

C C energy ( ∆@A3B ) using equation 11. The value of ∆@A3B = -14.2 (0.3) kJ mol-1, averaged over

7

the temperature range considered, clearly indicates that the adsorption process is spontaneous

8

(Mondal et al., 2013; Gupta and Nayak, 2012). By applying the van’t Hoff equation, the

9

adsorption enthalpy was also derived from the slope of the plot of DE93 versus T-1 (Figure

AC C

EP

TE D

3

10

C S8), with the assumption that ΔFA3B is independent on T in the restricted temperature range

11

considered.

12

C ∆@A3B = − GDE93

(13)

24

ACCEPTED MANUSCRIPT C The negative enthalpy value ΔFA3B = -10 (3) kJ mol-1, indicates that the adsorption of

2

mercury on Spanish broom fiber is exothermic, in agreement with the decreases of the

3

equilibrium distribution constant with T (Table 2 and Fig. S8).

4

Finally, the maximum adsorption capacity does not change significantly with T, indicating

5

that the adsorption efficiency of the Spanish broom is quite independent of the temperature.

6

The Spanish broom shows a good adsorption capacity compared to other materials (Zabihi et

7

al., 2009; Gupta et al., 2015; Lv et. al, 2012; Yee et al., 2013; Bag et al., 2007; Shin et al.,

8

2007; Feng et al., 1997; Liu et al., 1998; MacKay et al., 1989; Li et al., 2014) which,

9

however, must be activated with chemical or physical pretreatments or even prepared by

11 12

SC

chemical synthesis (Table S2).

M AN U

10

RI PT

1

3.5. Dynamic study in column

14

Figure 6 shows the breakthrough curves obtained by plotting the ratio between the outlet (Ct)

15

and inlet (C0) concentration of Hg(II) as a function of time, under different sorbent mass (m)

16

and different flow rates (Q). It can be seen that, for a given mass, as the flow rate decreases

17

the exhaustion point, i.e., the point at which the adsorbent bed is fully saturated, shifts

18

forward in time.

19

According to the literature (Bear, 1972; Dagan, 1989; Neuman, 1990), this is a typical

20

behavior and is due to the fact that at decreasing flow rates the concentration of mercury per

21

unit time flowing through the column is lower, being the other conditions the same.

22

Therefore, the exhaustion point occurs later.

23

On the other hand, at the same flow rate, as the adsorbent mass increases the exhaustion point

24

occurs later. This can be explained by taking into account the increase of the overall number

25

of active sites as the m/C0 ratio increases.

AC C

EP

TE D

13

25

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

1

Figure 6. Breakthrough plots showing the Ct /C0 ratio as a function of time with an inlet

3

Hg(II) concentration C0 = 100 mg L-1. Solid lines are the fitting curves modeled by a Single

4

Thomas (a) and a Double Thomas model (b). The two breakpoints, common to all the S-shape

5

curves, are indicated by arrows.

EP

AC C

6

TE D

2

7

However, in contrast to a typical S-shape trend generally observed for a breakthrough curve

8

collected during constant injection tracer tests (De Marsily, 1986; Pugliese, 2013), here we

9

observed a systematic double S shape behavior. Indeed, we can clearly identify two

10

breakpoints, one intermediate and a final pseudo-exhaustion point on each curve reported in

11

Figure 6. This suggests that mercury adsorption occurs in at least two processes with different

12

timescales. According to the batch results regarding the adsorption kinetics (Figure 4), it can

13

be advanced the hypothesis that, during column operation, the adsorption mechanism involves 26

ACCEPTED MANUSCRIPT the surface active sites in the first events until they become totally occupied, and then the

2

inner ones which, however, contribute to a lesser extent to the overall adsorption. As

3

highlighted in section 3.3, in which the adsorption kinetics of the batch tests have been

4

discussed, Spanish broom fibers are responsible of a diffusion process involving different

5

solute transport steps (IPD). This process is probably the same observed during column

6

dynamic tests and causing the occurrence of the double S shape curves. The hypothesis is that

7

the adsorption mechanism should first involve all the external superficial active sites of the

8

fibers, and then the ones settled within the fibers which, however, give a minor contribute to

9

the overall adsorption process. When a solution flows within a porous medium, the solute is

10

transported according to three main processes: advection which is originated from the fluid’s

11

flow which in turn is determined by the fluid’s physical properties and by the geometry and

12

physicochemical properties of the bounding solid, kinematic dispersion causing macroscopic

13

mixing because of local velocity variations, and diffusion resulting from the molecules

14

thermal motion (De Marsily, 1986). The flow rates adopted for the column tests were

15

calibrated to reproduce a velocity field typical of groundwater systems. In such systems, the

16

relative importance of the three aforementioned solute transport phenomena is indicated by a

17

dimensionless parameter known as the Peclet Number (Pe). It is defined as

18

HI =

19

where q (m s-1) is Darcy’s velocity, k0 (m2) is the intrinsic permeability of the porous medium,

20

n is the porosity, an D0 (m2 s-1) is the diffusion coefficient. When Pe >> 1 transport by

21

advection is faster than transport by diffusion, and the system is advection dominated. As the

22

Peclet Number decreases the three processes become comparable. For small values (Pe < 1),

23

diffusion dominates (De Marsily, 1986). Table 3 shows the Peclet Numbers calculated for all

24

the column tests performed in this study. The results demonstrate that the transport process

.JK

(14)

AC C

L

EP

TE D

M AN U

SC

RI PT

1

27

ACCEPTED MANUSCRIPT 1

occurred in the experiments is mainly driven by diffusion, confirming the theory on how the

2

double S curves have been generated.

3 In the hypothesis of a Langmuir adsorption isotherm and a second order adsorption kinetics,

5

the breakthrough curves can be modeled by the Thomas model (Mustafa and Ebrahim, 2010),

6

given by equation 15:

7

Ct / C0 =

8

where, kTh (mL min-1 mg-1) and q0 (mg g-1) are the Thomas constant and the maximum

9

adsorption capacity, respectively.

1

(15)

SC

e

M AN U

1+ e

− k Th C 0 t k Th q 0 m / Q

RI PT

4

However, the fitted curves reported in Figure 6a, clearly evidence that the Thomas model is

11

not able to reproduce the experimental trend. Actually, strong deviations from the Thomas

12

model occur in the intermediate portion of the breakthrough curves. This is due to the fact that

13

this model does not take into account the two processes mentioned above, both contributing to

14

the adsorption. Since the experimental points describe a double S-shape curve, we tried to fit

15

them with a double Thomas model given by equation 16:

16

Ct / C 0 =

17

where, the new parameter f, was named Active Surface Sites Fraction (ASSF). Obviously, if f

18

assumes a value equal to 1 the equation 15 (Double-Thomas Model) becomes identical to

19

equation 14 (Single-Thomas Model). This parameter can be mathematically described as the

20

difference between the Ct/C0 ratio at the beginning and at the end of the first breakthrough

21

curve while 1-f is the analogous difference for the second breakthrough curve; kTh_1; kTh_2 and

22

q0_1; q0_2 are the Thomas constants and the maximum adsorption capacities for the first and

23

second process, respectively. The parameter f also takes a clear physical meaning from which

24

its name derives. Being a number in the 0-1 range, it discriminates the fraction of Hg(II)

+

EP

− kTh _ 1C0t kTh _1 q0 _ 1m / Q

1+ e

1− f e

− kTh _ 2C0t kTh _ 2 q0 _ 2m / Q

(16)

AC C

1+ e

f e

TE D

10

28

ACCEPTED MANUSCRIPT species adsorbed by the superficial active sites from that adsorbed by the inner ones (1-f). For

2

a given adsorbent mass, f is a function of the contact time between the aforementioned mass

3

and the solution flowing through it. As can be seen in Table 3, the ASSF decreases with

4

decreasing injected flow rates (row 2-4). A lower flow rate results in a higher contact time

5

which allows the ions to reach and involve, in the adsorption process, more inner active sites.

6

Considering hence the experiments performed with the same flow rate and different adsorbent

7

mass, f decreases with the increase of the latter (row 1 and 2). The increase in mass, for a

8

given volume, of the adsorbent material causes a density variation of the filter which in turn

9

results in a porosity decrease. The lower porosity reduces the effective transversal area

SC

RI PT

1

available for the solution flux, resulting in a reduced involvement of the superficial active

11

sites. The Double-Thomas Model fitting lines are displayed in Figure 6b and the related

12

fitting parameters are collected in Table 3. This model better fits the experimental data with

13

respect to the Single-Thomas Model, with R2 values always larger than 0.996. Interestingly,

14

the Thomas constants associated to the two processes have quite similar values, suggesting

15

that the kinetic mechanism undergoing the two processes is the same. Notably, the

16

breakthrough step of the S-shape curve (f) related to the first process has a tendency to

17

decrease with decreasing flow rate, and tends to be vanishingly small for very low flow rates.

18

It can be argued that, in such a case, the second process is quite indistinguishable from the

19

first (the two are quite overlapped) because the system has enough time to equilibrate under

20

low flux conditions.

21

As Table 3 shows, the maximum adsorption capacities increases as the flow rate decreases,

22

being the other conditions the same. As a matter of fact, the overall adsorption capacity, q0_2,

23

assumes values close to the batch ones (Table 2).

AC C

EP

TE D

M AN U

10

24 25

29

ACCEPTED MANUSCRIPT Table 3 Double-Thomas model fitting parameters m

Q

kTh_1 ×10-4

kTh_2 ×10-4

q0_1

q0_2

(g)

(mL min-1)

(mL min-1 mg-1)

(mL min-1 mg-1)

(mg g-1)

(mg g-1)

f

R2

19

2.0 (0.2)

2.3 (0.1)

4.4 (0.2)

11.95 (0.07)

0.33 (0.01)

115

19

0.98 (0.01)

0.99 (0.03)

4.5 (0.2)

11.22 (0.06)

0.23 (0.01)

115

15

1.0 (0.2)

0.54 (0.02)

4.9 (0.3)

9.6 (0.1)

0.20 (0.01)

115

10

0.6 (0.1)

1.41 (0.08)

7.7 (0.5)

14.02 (0.04)

0.15 (0.02)

Values in parentheses are the standard deviation of the data

EC1/EC2

Double-

Single-

Thomas

Thomas

0.637

0.9971

0.9652

RI PT

80

0.577

5.1

0.9980

0.9953

0.455

3.4

0.9961

0.9964

0.303

1.4

0.9975

0.9764

SC

1

R2 Pe

The contribution of the two processes to the overall adsorption can be estimated by the ratio

3

of the exhaustion capacities between the first (EC1) and the second process (EC2). The

4

exhaustion capacity is defined as the mass of the adsorbate removed per unit mass of

5

adsorbent at the saturation point (Gupta et al., 2016). It can be estimated by calculating the

6

area overlaying the breakthrough curve up to the exhaustion point (Gupta et al., 2000; Gupta

7

et al., 1997; Medvidovic´et al., 2008) (Appendix A). If we compare the experiments

8

performed under identical conditions (Table 3 row 2-4), we find that this ratio is always

9

higher than 1 indicating that the first adsorption process contributes mainly to the overall

10

adsorption. However, the EC1/EC2 ratio decreases in time with the flow rate (Table 3),

11

indicating that, if the adsorbent bed is allowed to stay in longer contact with the solution, the

12

second process becomes more important.

14

TE D

EP

AC C

13

M AN U

2

15

4. Conclusions

16

In conclusion, here we have shown that the Spanish broom, that is taken as a typical

17

lignocellulosic material, can be used as adsorbent material for mercury removal from

18

contaminated water with removal percentages as high as 86%. Batch experiments revealed 30

ACCEPTED MANUSCRIPT that the adsorption occurs due to the active carboxyl and hydroxyl binding sites available on

2

its lignocellulosic surface. Interestingly, the adsorption dynamic study performed in column

3

under different experimental conditions, revealed, for the first time, an unconventional

4

breakthrough behavior displaying a double S-shape curve. This has been modeled by a

5

Double-Thomas equation with the introduction of a new parameter, the active surface sites

6

fraction (f), which discriminates the fraction of Hg(II) species adsorbed by the superficial

7

active sites from that adsorbed by the inner ones (1-f). The parameter f is significantly

8

affected by the contact time. In fact, the greater is the time in which the Hg(II) remains in

9

contact with the adsorbent material, the greater is the efficiency of the filter because the inner

SC

RI PT

1

sites come into play. This finding has therefore important implications for the design of

11

filtering devices based on lignocellulosic materials. In general, this double mechanism, due to

12

the intraparticle diffusion, suggests that this kind of filter is suitable for the remediation of

13

aquifers contaminated by Hg(II), because in this context the flow rate and consequently the

14

Peclet number, are very low.

15

The use of the Spanish Broom has, however, additional advantages with respect to other

16

potential lignocellulosic adsorbent sources: its high diffusion in many areas of the Earth, and

17

the very simple and cheap pretreatment process (drying and milling).

18

The experimental results are very encouraging, indicating that the Spanish broom could be a

19

promising material for aquifer remediation, in those cases in which Hg pollution is observed.

20

This natural material, thanks to its mechanical properties, could be adopted, during the

21

implementation of "on site" techniques, inside filtering devices included within pumping

22

systems or, thanks to its extremely availability and affordability, by "in situ" techniques such

23

as Permeable Reactive Barriers (PRB).

24

Considering the encouraging results obtained in this study, new experiments involving the

25

remediation of Hg contaminated water flowing within Spanish brooms filtering devices are

AC C

EP

TE D

M AN U

10

31

ACCEPTED MANUSCRIPT 1

planned, together with the implementation of interpretative models dealing with the

2

aforementioned working conditions. The need for such an approach is required by the actual

3

operating situations in which this material must be used, which are, in most cases, water

4

flowing conditions through or on the soil

RI PT

5 6 Appendix A

8

Exhaustion capacity calculation

9

The total mass (m) of mercury (II) retained in the column is given by the following integral:

SC

7

11

M AN U

10 

M = N OC )C − ) P+

(A1)

12

where, te is the time at the exhaustion point.

14

Equation A1 can be re-written as the mass of solute retained in the column normalized with

15

respect to the initial mercury concentration (C0) at the volume of exhaustion (Vec):

16 >







= OC R1 −  PQ

EP

17

TE D

13



AC C

18

(A2)

19

In order to estimate the contribution of the first and second adsorption process to the overall

20

Hg(II) adsorption in the column, the following ratio was calculated, that is the ratio between

21

the exhaustion capacity relative to the first adsorption process (EC1) and to the second one

22

(EC2):

23 24

6S T 65 T

>S

=> = 5

U U

=

T V O RS T  3

V RS

 T

OV R5T  3

(A3)



32

ACCEPTED MANUSCRIPT 1 where, m1 and m2 are the mass of Hg(II) retained in the column due to the first and the second

3

process respectively; Vec1 and Vec2 are the volume at the intermediate and final exhaustion

4

points, respectively.

5

As shown in Table 3, this ratio is larger than one, in agreement with the condition M > M ,

6

i.e., the mass adsorbed in the first adsorption process is larger than that adsorbed in the second

7

one.

RI PT

2

SC

8

0,8

1-Ct/C0

0,6

0,4

m1/C0 m2/C0

0,0 0

2000

4000

TE D

0,2

M AN U

1,0

6000

8000

10000

12000

14000

Volume (mL)

9

Figure A1. Illustration of the graphical integration of the breakthrough curve for the

11

exhaustion capacities calculation.

EP

10

13 14

AC C

12

15

Acknowlegments

16

F. E. A. A. is grateful to the Secretaría de Educación Superior, Ciencia, Tecnología e

17

Innovación of the Ministerio de Educación of the Government of Ecuador for funding his

18

Ph.D studies.

19 33

ACCEPTED MANUSCRIPT References

2

Adhoum, N., Monser, L., Bellakhal, N., Belgaied, J.E., 2004. Treatment of electroplating

3

wastewater containing Cu2+, Zn2+ and Cr(VI) by electrocoagulation, J. Hazard. Mater.,

4

B112, 207–213.

5

Al-Damluji SF., 1976, Intoxication due to alkylmercury-treated seed-1971-72 outbreak in

6

Iraq: clinical aspects, Bull World Health Organization 53, 65-81.

7

Alver, E., Metin, A.Ü., 2012. Anionic dye removal from aqueous solutions using modified

8

zeolite: Adsorption kinetics and isotherm studies, Chem. Eng. J., 200–202, 59–67.

9

Alyüz, B., Veli, S., 2009. Kinetics and equilibrium studies for the removal of nickel and zinc

M AN U

SC

RI PT

1

from aqueous solutions by ion exchange resins, J. Hazard. Mater., 167, 482–488.

11

Atkins, P., De Paula, J., 2006 Physical Chemistry, Eighth ed., W. H. Freeman and Company,

12

New York, 918-922.

13

Baes, C. F. and Mesmer R. E., 1976. The Hydrolysis of Cations. A Wiley-Interscience

14

Publication.

15

Bag, S., Trikalitis, P. N., Chupas, P. J., Armatas, G. S. & Kanatzidis, M. G., 2007. Porous

16

semiconducting gels and aerogels from chalcogenide clusters. Science 317, 490–493.

17

Bailey S. E., Trudy, J. O., Bricka, R. M., and Dean A., 1999. A review of potentially low-

18

cost sorbent for heavy metals. Water Res., 33, 2469-2479.

19

Bear, J., 1972. Dynamics of Fluids in Porous Media, American Elsevier, New York.

AC C

EP

TE D

10

34

ACCEPTED MANUSCRIPT Beneduci, A., Chidichimo, G., Dardo, G., Pontoni, G. 2011. Highly routinely reproducible

2

alignment of 1H-NMR spectral peaks of metabolites in huge sets of urines. Analytica Chimica

3

Acta, 685, 186-195.

4

Cantrell, K. J., Kaplan, D. I., Wietsma, T. W, 1995. Zero-valent iron for the in situ

5

remediation of selected metals in groundwater: J. Hazard. Mater., 42, 201-212.

6

Casas A., Alonso M. V., Oliet M., Rojo E., and Rodríguez F., 2012. FTIR analysis of lignin

7

regenerated from Pinus radiate and Eucalyptus globulus woods dissolved in imidazolium-

8

based ionic liquids. J. Chem. Technol. Biotechnol., 87, 472-480.

9

Chiarle, S., Ratto, M., and Rovatti, M., 2000. Mercury removal from water by ion exchange

M AN U

SC

RI PT

1

resins adsorption. Water Res., 34, 2971-2978.

11

Chidichimo, G., Aloise, A., Beneduci, A., De Rango, A., Pingitore, G., Furgiuele, F.,

12

Valentino, P., 2015. Polyurethanes reinforced with Spartium Junceum fibers, Polym.

13

Composites, doi: 10.1002/pc.23501.

14

D’Adamo, P., Ulivi, S., Beneduci, A., Pontoni G., Capasso G., Lanzara C., Andrighetto G.,

15

Hladnik U., Nunes V., Palacin M., and Gasparini P., 2010. Metabonomics and population

16

studies: age-related amino acids excretion and inferring networks through the study of urine

17

samples in two Italian isolated populations, Amino Acids, 38, 65-73.

18

Dagan, G., 1989. Flow and Transport in Porous Formations, Springer, Berlin, 465.

19

De Marsily, G., 1986. Quantitative hydrogeology: groundwater hydrology for engineers.

20

Academic Press, Orlando, FL.

AC C

EP

TE D

10

35

ACCEPTED MANUSCRIPT Demirbas, A., 2008. Heavy metal adsorption onto agro-based waste materials: A review, J.

2

Hazard. Mater., 157, 220–229.

3

Esalah, J.O., Weber, M.E., Vera, J.H., 2000. Removal of lead, cadmium and zinc from

4

aqueous solutions by precipitation with sodium di-(n-octyl) phosphinate, Can. J. Chem. Eng.,

5

78, 948–954.

6

Esteban, M., Casassas E. and Fernandez L., 1986. Formation constants of some mercury(II)

7

complexes determined from their anodic polarographic signals. Talanta, 33, 843-846.

8

Feng, X., Fryxell, G. E., Wang, L.-Q., Kim, A. Y., Liu, J., Kemner, K. M., 1997.

9

Functionalized monolayers on ordered mesoporous supports. Science, 276, 923–926.

M AN U

SC

RI PT

1

Furia, E. and Sindona, G., 2012. Interaction of Iron(III) with 2-Hydroxybenzoic Acid in

11

Aqueous Solutions. J. Chem. Eng. Data, 57, 195-199.

12

Furia, E., Falvo, M. and Porto, R., 2009. Solubility and Acidic Constants of l-Cystine in

13

NaClO4 Aqueous Solutions at 25 °C. J. Chem. Eng. Data, 54(11), 3037-3042.

14

Furia, E., Napoli, A., Tagarelli, A, Sindona, G., 2013. Speciation of 2‐Hydroxybenzoic Acid

15

with Calcium(II), Magnesium(II), and Nickel(II) Cations in Self-Medium. J. Chem. Eng.

16

Data, 58, 1349-1353.

17

Gabriele, B., Cerchiara, T., Salerno, G., Chidichimo, G., Vetere, M.V., Alampi, C., Gallucci,

18

M.C., Conidi, C., Cassano, A., 2010. A new physical–chemical process for the efficient

19

production of cellulose fibers from Spanish broom (Spartium junceum L.), Bioresour.

20

Technol. 101, 724–729.

AC C

EP

TE D

10

36

ACCEPTED MANUSCRIPT Gans, P., Sabatini, A. and Vacca, A., 1985. SUPERQUAD: An Improved General Program

2

for Computation of Formation Constants from Potentiometric Data. J. Chem. Soc. Dalton, 6,

3

1195-1200.

4

Giles, C.H., Smith, D., Huitson, A., 1974. A general treatment and classification of the solute

5

adsorption isotherm. I. Theoretical, J. Colloid Interf. Sci,, 47, 755–765.

6

Gran, G., 1952. Determination of the equivalence point in potentiometric titrations. Part II.

7

Analyst, 77, 661-671.

8

Guo, X., Zhang, S., Shan, X., 2008. Adsorption of metal ions on lignin, J. Hazard. Mater.,

9

151, 134–142.

M AN U

SC

RI PT

1

Gupta, V. K., Nayak, A., 2012. Cadmium removal and recovery from aqueous solutions by

11

novel adsorbents prepared from orange peel and Fe2O3 nanoparticles. Chem. Eng. J., 180, 81-

12

90.

13

Gupta, V. K., Nayak, A., Agarwal, S., 2015. Bioadsorbents for remediation of heavy metals:

14

current status and their future prospects. Environ. Eng. Res., 20, 1-18.

15

Gupta, V. K., Srivastava, S. K., Moham, D. and Sharma, S. 1997. Design parameters for

16

fixed bed reactorsof activated carbon developed from fertilizer waste for the removal of some

17

heavy metal ions. Waste Management, 17, 517-522.

18

Gupta, V. K., Srivastava, S. K., and Tyagi, R. 2000. Design parameters for the treatment of

19

phenolic wastes by carbon columns (obtained From fertilizer waste material). Water Res. 34,

20

5, 1543-1550.

21

Gupta, V. K., Suhas, Tyagi, I., Agarwal, S., Singh, R., Chaudhary, M., Harit, A., Kushwaha,

AC C

EP

TE D

10

37

ACCEPTED MANUSCRIPT S., 2016. Column operation studies for the removal of dyes and phenols using a low cost

2

adsorbent. Global J. Environ. Sci. Manage., 2 (1), 1-10.

3

Haghseresht, F. and Lu, G., 1998. Adsorption characteristics of phenolic compounds onto

4

coal-reject-derived adsorbents. Energy and Fuels, 12, 1100-1107.

5

Hamdaoui, O. and Naffrechoux, E., 2007. Modeling of adsorption isotherms of phenol and

6

chlorophenols onto granular activated carbon. Part I. Two parameters models and equations

7

allowing determination of thermodynamic parameters. J. Hazard. Mater., 147, 381-394.

8

Henriksson, G., Lignin, in: M. Ek, G. Gellerstedt, G. Henriksonn (Eds.), Wood Chemistry

9

and Technology, Walter de Gruyter GmbH and Co., Berlin, 2009, pp. 121-146.

M AN U

SC

RI PT

1

Ho, Y.S,. 2006. Review of second-order models for adsorption systems, J. Hazard. Mater,

11

136, 681–689.

12

Ho, Y.S. and Wang C.C., 2008. Sorption equilibrium of mercury on ground-up tree fern. J.

13

Hazard. Mater., 156, 398–404.

14

Ho, Y.S., McKay, G., 1999., Pseudo-second order model for sorption processes, J. Process.

15

Biochem., 34, 51–465.

16

INIGEMM, (Instituto Nacional de Investigación Geológico Minero Metalúrgico), 2013,

17

Assessment of the susceptibility to contamination of underground water resources, mining

18

district of Zaruma – Portovelo, Ecuador, (Evaluación de la susceptibilidad, a la

19

contaminación del recurso hídrico subterráneo del distrito minero Zaruma – Portovelo,

20

Ecuador),, p. 239-241.

AC C

EP

TE D

10

38

ACCEPTED MANUSCRIPT Juang, R.S., Wu, F.C., Tseng, R.L., 2002. Characterization and use of activated carbons pre-

2

pared from bagasse for liquid-phase adsorption, Colloids Surf. A: Physicochem. Eng.

3

Aspects 201, 191–199.

4

Khachatourians, G. G., Arora, D.K., 2001. Applied Mycology and Biotechnology, Volume 1.

5

Agriculture and Food Production, first ed., Elsevier Science B.V., Amsterdan, 325.

6

Khalil, M. M., Radalla, A. M., 1998. Binary and ternary complexes of inosine. Talanta, 46,

7

53–61.

8

Khoramzadeh, E., Nasernejad, B., Halladj, R., 2013. Mercury biosorption from aqueous

9

solutions by Sugarcane Bagasse: J. Taiwan Inst. Chem. Eng., 44, 266–269.

M AN U

SC

RI PT

1

Kumar, K.V., 2006. Linear and non-linear regression analysis for the sorption kinetics of

11

methylene blue onto activated carbon: J. Hazard. Mater., B137, 1538–1544.

12

Kumar, P.R., Chaudhari, S., Khilar, K.C., Mahajan, S.P., 2004. Removal of arsenic from

13

water by electrocoagulation, Chemosphere, 55 (9), 1245–1252.

14

Lagergren, S., 1898. Zur theorie der sogenannten adsorption gelöster stoffe Kungliga

15

Svenska Vetenskapsakademiens. Handlingar, 24 (4), 1–39.

16

Li, B., Zhang Y., Ma, D., Z., Shi, and Ma, S., 2014. Mercury nano-trap for effective and

17

efficient removal of mercury(II) from aqueous solution. Nature Communication, 5:5537,

18

DOI: 10.1038/ncomms6537.

19

Limousin, G., Gaudet, J.-P., Charlet, L., Szenknect, S., Barthe`s, V., Krimissa, M., 2007.

20

Sorption isotherms: A review on physical bases, modeling and measurement, Appl.

21

Geochem., 22, 249–275.

AC C

EP

TE D

10

39

ACCEPTED MANUSCRIPT Lin, S.H., Peng, C.F., 1994. Treatment of textile wastewater by electrochemical method,

2

Water Res., 28, 277–282.

3

Liu, J. Feng, X., Fryxell, G. E., Wang, Li-Q., Kim, A. Y. and Gong, M. 1998. Hybrid

4

mesoporous materials with functionalized monolayers. Adv. Mater. 10, 161–165.

5

Liu, T., Wang, Z. L., Yan, X., Zhang, B., 2014. Removal of mercury (II) and chromium (VI)

6

from wastewater using a new and effective composite: Pumice-supported nanoscale zero-

7

valent iron, Chem. Eng. J., 245, 34–40.

8

Lv, J., Luo, L., Zhang, J., Christie, P., Zhang, S., 2012. Adsorption of mercury on lignin:

9

Combined surface complexation modeling and X-ray absorption spectroscopy studies,

M AN U

SC

RI PT

1

Environ. Pollut., 162, 255-261.

11

MacKay, G., Blair, H. S., Findon, A., 1989. Equilibrium studies for sorption of metal ions

12

onto chitosan. Indian J. Chem., 28A, 356-360.

13

Mahajan, G., Sud, D., 2013. Application of ligno-cellulosic waste material for heavy metal

14

ions removal from aqueous solution, J. Environ. Chem. Eng., 1, 1020–1027.

15

Manohar D.M., Krishnan K.A., Anirudhan T.S., 2002. Removal of mercury (II) from

16

aqueous solutions and chloralkali industry wastewater using mercaptobenzimidazole clay.

17

Water Res., (36) 1609-1619.

18

Medvidovic´ N.V., Peric´ J., and Trgo M., 2008 Testing of Breakthrough Curves for

19

Removal of Lead Ions from Aqueous Solutions by Natural Zeolite-Clinoptilolite According

20

to the Clark Kinetic Equation. Separation Science and Technology, (43) 944–959, 2008

AC C

EP

TE D

10

40

ACCEPTED MANUSCRIPT Mondal, D.K., Nandi, B.K., Purkait, M.K., 2013. Removal of mercury (II) from aqueous

2

solution using bamboo leaf powder: Equilibrium, thermodynamic and kinetic studies, J.

3

Environ. Chem. Eng., 1, 891–898.

4

Mrozowski, J., Zielinski, J., 1983. Studies of zinc and lead removal from industrial wastes by

5

electrocoagulation, Environ. Prot. Eng., 9. 77–85.

6

Murthy, Z.V.P., Parikh, P. A., Patel, N. B., 2013. Aplication of β – Zeolite, Zeolite Y, and

7

Modernitte as absorbent for remove Mercury from aqueos solutions, J. Dispers. Sci.

8

Technol., 34, 747–755.

9

Mustafa Y. A., Ebrahim S.E., 2010. Utilization Of Thomas Model To Predict The

M AN U

SC

RI PT

1

Breakthrough Curves For Adsorption And Ion Exchange, J. Eng. 4 (16), 6206 – 6223.

11

Naiya, T. K., Chowdhury, P., Bhattacharya, A.K., Das, S. K., 2009. Saw dust and neem bark

12

as low-cost natural biosorbent for adsorptive removal of Zn(II) and Cd(II) ions from aqueous

13

solutions, Chem. Eng. J., 148, 68–79.

14

Namasivayam, C., Kadirvelu, K., 1999. Uptake of mercury (II) from wastewater by activated

15

carbon from an unwanted agricultural solid by-product: coirpith, Carbon 37, 79–84

16

Neuman, S.P., 1990. Universal scaling of hydraulic conductivities and dispersivities in

17

geologic media. Water Resour. Res., 26, 1749–1758

18

Niu, Z., Fan, Q., Wang, W., Xu, J., Chen, L., Wu, W., 2009. Effect of pH, ionic strength and

19

humic acid on the sorption of uranium(VI) to attapulgite, Applied Radiation and Isotopes 67,

20

1582–1590.

21

Pearson, R. G., 1963. Hard and soft acids and bases, J. Am. Chem. Soc., 85, 3533-3539.

AC C

EP

TE D

10

41

ACCEPTED MANUSCRIPT Porto, R., De Tommaso, G. and Furia, E., 2005. The Second acidic constant of salicylic acid.

2

Ann. Chim., 95, 551-558.

3

Porwal, S. K. Furia, E., Harris, M. E., Viswanathan R., Devireddy L., 2015. Synthetic,

4

potentiometric and spectroscopic studies of chelation between Fe(III) and 2,5-DHBA

5

supports salicylate-mode of siderophore binding interactions. J. Inorg. Biochem. 145, 1–10.

6

Pugliese L., Poulsen T., Straface S., 2013. “Gas Solute Dispersivity Ratio in Granular Porous

7

Media as Related to Particle Size Distribution and Particle Shape”. Water Air Soil Pollut.,

8

224, (1691), 1-12.

9

Rao, M.M., Reddy, D.H.K.K., Venkateswarlu, P., Seshaiah, K., 2009. Removal of mercury

M AN U

SC

RI PT

1

from aqueous solutions using activated carbon prepared from agricultural by-product/waste, J.

11

Environ. Management., 90 (2009) 634-643.

12

Ratcliffe H. E., Swanson G. M., Fischer L.J., 1996, Human Exposure to Mercury: A Critical

13

Assessment of the Evidence of Adverse Health Effects, J. Toxicol Envirom Health, 49 (3),

14

221-270.

15

Rossotti F. J. C. and Rossotti H., 1965. Potentiometric titrations using Gran plots. A textbook

16

omission. J. Chem. Edu., 42 (7), 375-378.

17

Shin, Y., Fryxell, G. E., Um, W., Parker, K., Mattigod, S. V. and Skaggs, R., 2007. Sulfur-

18

functionalized mesoporous carbon. Adv. Funct. Mater. 17, 2897–2901.

19

Thielke, M. W., Bultema, L. A., Brauer, D. D., Richter, B., Fischer, M. and Theato, P., 2016.

20

Rapid Mercury(II) Removal by Electrospun Sulfur Copolymers. Polymers, 8, 266, 1-9.

AC C

EP

TE D

10

42

ACCEPTED MANUSCRIPT Tunali, S., Ozcan, A.S., Ozcan, A., Gedikbey, T., 2006. Kinetics and equilibrium studies for

2

the adsorption of acid red 57 from aqueous solutions onto calcined alunite, J. Hazard. Mater,

3

135, 141–148.

4

UNEP, 2013. Global Mercury Assessment 2013: Sources, Emissions, Releases and

5

Environmental Transport. UNEP Chemicals Branch, Geneva, Switzerland.

6

Visa, M., 2016., Synthesis and characterization of new zeolite materials obtained from fly ash

7

for heavy metals removal in advanced wastewater treatment, Powder Technol., 294, 338–

8

347.

9

Wang, S., Peng, Y., 2010. Natural zeolites as effective adsorbents in water and wastewater

M AN U

SC

RI PT

1

treatment: review, Chem. Eng. J., 156, 11–24.

11

Weber, W.J., Morris, C.J., 1962. Advances in water pollution research, in: Proceedings of the

12

First International Conference on Water Pollution Research, vol. 2, Pergamon Press, Oxford,

13

231.

14

Weisener C.G., Sale, K.S., Smyth, D.J.A., Blowes, D.W., 2005. Field Column Study Using

15

Zerovalent Iron for Mercury Removal from Contaminated Groundwater, Environ. Sci.

16

Technol., 39 (16), 6306–6312.

17

WHO, 1991, “Inorganic mercury: environmental health criteria 118,” in International

18

Programme on Chemical Safety.

19

Wilkin, R. T., Mc Neil, M. S., 2003. Laboratory evaluation of zero-valent iron to treat water

20

impacted by acid mine drainage, Chemosphere 53, 715–725.

AC C

EP

TE D

10

43

ACCEPTED MANUSCRIPT Wu F.C., Tseng, R.L., Juang, R.S., 2009. Initial behavior of intraparticle diffusion model

2

used in the description of adsorption kinetics, Chem. Eng. J., 153, 1–8.

3

Yee, K.-K. ; Reimer, N.; Liu, J.; Cheng, S.Y.; Yiu, S.M.; Weber, J.; Stock, N.; Xu, Z., 2013.

4

Effective mercury sorption by thiol-laced metal_organic frameworks: in strong acid and the

5

vapor phase. J. Am. Chem. Soc. 135, 7795–7798.

6

Zabihi, M., Haghighi Asl, A., Ahmadpour, A., 2010. Studies on adsorption of mercury from

7

aqueous solution on activated carbons prepared from walnut shell, J. Hazard. Mater., 174,

8

251–256.

9

Zhang, F.S., Nriagu, J.O., Itoh, H., 2005. Mercury removal from water using activated

SC M AN U

10

RI PT

1

carbons derived from organic sewage sludge, Water Res., 39, 389–395.

TE D

11

Figures captions

13

Figure 1. (a) Hg(II) adsorption as a function of the initial pH (C0 = 80 mg L-1; W = 0.5 g, T =

14

298 K). (b) Distribution diagram of Hg2+ in the presence of chloride (CM = 5.10-4 M and CL =

15

4.10-3 M).

16

Figure 2. Removal efficiency (RE%) (a) and Hg(II) concentration (Ct) as a function of the

17

contact time (b), at two different Hg(II) initial concentration (Hg 100 and 200 mg L-1). XRF

18

spectra of the Spanish broom before (black line) and after adsorption (red line) (c).

19

Figure 3. (a) Effect of the initial concentration on the adsorption process (W = 0.5 g, V= 50

20

mL); (b) effect of the adsorbent dosage on the efficiency of the adsorption process (C0 = 80

21

mg L-1, V= 50 mL).

AC C

EP

12

44

ACCEPTED MANUSCRIPT 1 Figure 4. Adsorption kinetics. (a) Intraparticle diffusion plot showing three regions of

3

linearity (C0 = 200 mg/L, V = 0.100 L, W = 5 g, T = 298 K). (b) Pseudo second order plots

4

showing the linear fitting of the experimental data points with eq. (7). The working conditions

5

for each set of data are reported in Table 1.

6

RI PT

2

Figure 5. Experimental adsorption isotherms for mercury at different temperatures and

8

nonlinear curve fitting with the Langmuir (a) and Freundlich model (b).

SC

7

9

Figure 6. Breakthrough plots showing the Ct /C0 ratio as a function of time with an inlet

11

Hg(II) concentration C0 = 100 mg L-1. Solid lines are the fitting curves modeled by a Single

12

Thomas (A) and a Double Thomas model.

13

Figure A1. Illustration of the graphical integration of the breakthrough curve for the

14

exhaustion capacities calculation.

TE D EP

16

AC C

15

M AN U

10

45

ACCEPTED MANUSCRIPT Highlights Static and dynamic Hg(II) adsorption tests were carried out on Spanish broom filters; Up to 86% removal of Hg(II) are obtained in typical groundwater conditions;

RI PT

Hg(II) adsorption occurs through the hydroxyls and carboxyls of the lignocellulosic material; A bimodal Thomas model is proposed to fit the double S-shape breakthrough curves;

AC C

EP

TE D

M AN U

SC

Aquifer polluted by Hg(II) can be remediated by the Spanish broom.