Journal of Colloid and Interface Science 512 (2018) 575–590

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

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

Regular Article

Synthesis and application of a new carboxylated cellulose derivative.

Part III: Removal of auramine-O and safranin-T from mono- and
bi-component spiked aqueous solutions
Filipe Simões Teodoro a, Megg Madonyk Cota Elias a, Gabriel Max Dias Ferreira a,
Oscar Fernando Herrera Adarme b, Ranylson Marcello Leal Savedra c, Melissa Fabíola Siqueira c,
Luis Henrique Mendes da Silva d, Laurent Frédéric Gil a, Leandro Vinícius Alves Gurgel a,⇑
a
Grupo de Físico-Química Orgânica, Departamento de Química, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto, Campus Universitário Morro do
Cruzeiro, s/n°, Bauxita, 35400-000 Ouro Preto, Minas Gerais, Brazil
b
Laboratório de Química Tecnológica e Ambiental, Departamento de Química, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto, Campus
Universitário Morro do Cruzeiro, s/n°, Bauxita, 35450-000 Ouro Preto, Minas Gerais, Brazil
c
Laboratório de Polímeros e Propriedades Eletrônicas de Materiais, Departamento de Física, Universidade Federal de Ouro Preto, Campus Universitário Morro do Cruzeiro,
s/n°, Bauxita, 35400-000 Ouro Preto, Minas Gerais, Brazil
d
Grupo de Química Verde Coloidal e Macromolecular, Departamento de Química, Centro de Ciências Exatas e Tecnológicas, Universidade Federal de Viçosa, Av. P. H. Rolfs, s/n°,
36570-000 Viçosa, Minas Gerais, Brazil

g r a p h i c a l a b s t r a c t

Adsorption Desorption Re-adsorption

HO O
OH O

Auramine-O qe-re= 2.67 mmol g-1

+ O
Py:DMA (1:1, v/v) Edes= 44.61 % Ere-ads= 77.90 %
OH
HO
OH
O
100 oC; 1 h; 300 rpm HO

n O Trimellitic O
Cellulose anhydride O
O O pH 4.5 qe = 2.628 mmol g-1
pH 7 qe = 5.172 mmol g-1
RO O
OR

n Binary System [HCl] = 0.01mol L-1

Cellulose modified 130 rpm
with Trimellitic Anhydride (CTA)
3 h, 25°C

AO qe = 1.051 mmol g-1

pH 7
ST qe = 3.258 mmol g-1

qe-re= 3.64mmol g-1

Edes= 78.87 % Ere-ads= 60.71 %
Celulose modified Safranin-T
Cellulose
with Trimellitic Anhydride (CTA)
pH 4.5 qe = 3.443 mmol g-1
pH 7 qe = 4.019 mmol g-1

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

Article history: In the third part of this series of studies, the adsorption of the basic textile dyes auramine-O (AO) and
Received 21 July 2017 safranin-T (ST) on a carboxylated cellulose derivative (CTA) were evaluated in mono- and bi-
Revised 21 October 2017 component spiked aqueous solutions. Adsorption studies were developed as a function of solution pH,
Accepted 23 October 2017
contact time, and initial dye concentration. Adsorption kinetic data were modeled by monocomponent
Available online 25 October 2017
kinetic models of pseudo-first- (PFO), pseudo-second-order (PSO), intraparticle diffusion, and Boyd, while
the competitive kinetic model of Corsel was used to model bicomponent kinetic data. Monocomponent
Keywords:
adsorption equilibrium data were modeled by the Langmuir, Sips, Fowler-Guggenhein, Hill de-Boer,
Adsorption
Auramine-O
and Konda models, while the IAST and RAST models were used to model bicomponent equilibrium data.
Safranin-T Monocomponent maximum adsorption capacities for AO and ST at pH 4.5 were 2.841 and 3.691 mmol
Desorption g
1, and at pH 7.0 were 5.443 and 4.074 mmol g
1, respectively. Bicomponent maximum adsorption
Isothermal titration calorimetry capacities for AO and ST at pH 7.0 were 1.230 and 3.728 mmol g
1. Adsorption enthalpy changes

⇑ Corresponding author.
E-mail addresses: legurgel@iceb.ufop.br, legurgel@yahoo.com.br (L.V.A. Gurgel).

https://doi.org/10.1016/j.jcis.2017.10.083
0021-9797/Ó 2017 Elsevier Inc. All rights reserved.
576 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
Thermodynamics (DadsH) were obtained using isothermal titration calorimetry. The values of DadsH ranged from
18.83 to
Molecular quantum mechanics
5.60 kJ mol
1, suggesting that physisorption controlled the adsorption process. Desorption and re-
adsorption of CTA was also evaluated.
1. Introduction
Contamination of ground and surface water by synthetic dyes
due to the discharge of untreated or partially treated industrial
wastewater is a serious environmental problem and is considered
to be a threat to humans and aquatic organisms [1,2]. Synthetic
dyes have various industrial uses including in cosmetics, food pro-
cessing, leather tanning, rubber, paper, plastics, printing and dye
manufacturing, and textiles [1,3].
The discharge of synthetic dyes into the hydrosphere is prob-
lematic primarily due to dyes’ potential to color water and reduce
sunlight penetration into water, as well as dyes’ resistance to pho-
tochemical and biological degradation [1,4]. In addition, synthetic
dyes and their degradation products include mutagenic com-
pounds, carcinogenic compounds, and compounds that may
increase the chemical oxygen demand (COD) and biochemical oxy-
gen demand (BOD) of water resources [1,3]. The complex aromatic
chemical structures of synthetic dyes make them very stable to
heat, oxidizing agents, photochemical degradation, and biodegra-
dation, which makes their removal from industrial wastewaters
difficult [1,3].
Auramine-O and safranin-T are classified as basic dyes [5],
which can be used to dye wool, silk, acrylic fibers, and leather in
textile industries [6]. Auramine-O can cause acute oral toxicity,
carcinogenicity, cytotoxicity, DNA damage, genotoxicity, and
mutagenicity [7]. Safranin-T can cause carcinogenicity, mitochon-
drial toxicity, mutagenicity, and nucleic acid damage [7]. According
to Terinte et al. [8], most conventional dyeing takes place in devel-
oping countries, where environmental legislation is often insuffi-
cient and/or not implemented properly. A survey describing the
percentages of unfixed dyes that may be discharged in the effluent
for different dyes, fabrics, and applications was recently reported
by Terinte et al. [8].
Three categories of treatment technologies have been applied to
effluent containing synthetic dyes: physical, chemical, and biolog-
ical [3,9]. Physical treatments include adsorption by activated car-
bon, membrane filtration and separation, ion-exchange,
irradiation, and electrokinetic coagulation. Chemical treatments
include oxidative processes, photochemical destruction, and elec-
trochemical destruction. Biological treatments include aerobic
and anaerobic treatment, among others [1,3,9]. Despite the devel-
opment and application of technologies for treatment of wastewa-
ters containing synthetic dyes, cost effective, efficient, and rapid
water treatment at the commercial level is still a challenging prob-
lem for the scientific community [1]. Although adsorption can treat
large flow rates, producing high quality effluents [1], if the adsor-
bent material cannot be cleaned by desorption and reused, the
applicability and usefulness of this treatment method is limited.
Thus, research to develop suitable adsorbents with high adsorption
capacity, good economic viability, versatility for removal of various
organic and inorganic pollutants, and better regeneration capacity
is still needed.
In the first and second parts of this series of studies [10,11], the
synthesis and application of a new adsorbent material (CTA) using
cellulose as a solid support was described in detail, as well as the
adsorption of Co2+, Cu2+, and Ni2+ from mono- and multi-
component aqueous solutions. The high adsorption capacity and
selectivity exhibited by the CTA adsorbent for removal of metal
Ó 2017 Elsevier Inc. All rights reserved.
ions from mono- and multi-component aqueous solutions, in addi-
tion to CTA’s high capacity for basic dye adsorption compared to
previously studied adsorbents (Supplementary Table 1), motivated
our research group to begin adsorption studies for the removal of
dyes from aqueous solutions. As pointed out by Gusmão et al.
[12], the use of the same adsorbent for various environmental
applications is desirable because it makes its production econom-
ically feasible.
In the third part of this series of studies, the removal of the basic
dyes auramine-O (AO) and safranin-T (ST) using the CTA adsorbent
was assessed as a function of solution pH, contact time (kinetics),
and initial dye concentration in spiked mono- and bi-component
aqueous solutions. Dye removal was also studied at two pH values
(i.e. 4.5 and 7.0). A suitable desorption process was developed to
evaluate the regeneration and reusability of the CTA adsorbent.
The enthalpy of the adsorption process was also evaluated and
assessed by isothermal titration calorimetry, and the removal
mechanism of both dyes was evaluated based on thermodynamic
data and molecular quantum mechanics calculations.
2. Material and methods
2.1. Material
Auramine-O (MW = 303.83 g mol
1; C17H21N3HCl; C.I. 41000;
kmax = 434 nm; purity  100%) and safranin-T (MW = 350.84 g
mol
1; C20H19N4Cl; C.I. 50240; kmax = 530 nm; purity  100%) were
purchased from Vetec (Brazil). Anhydrous citric acid (C6H8O7; 99%)
and trisodium citrate (Na3C6H5O72H2O; 99%) were purchased
from Synth (Brazil). Dye solutions used in the adsorption studies
were prepared in buffer solutions consisting of 0.05 mol L
1 citric
acid/sodium monocitrate (pH from 2.13 to 3.13), 0.05 mol L
1
sodium monocitrate/sodium dicitrate (pH from 4.13 to 5.15), and
0.05 mol L
1 sodium dicitrate/sodium tricitrate (pH from 6.13 to
7.13). Other materials and chemicals used in this investigation
are described in the first part of this series of studies [10].
2.2. Adsorption experiments
Monocomponent adsorption studies of AO and ST on the CTA
adsorbent were performed at two pH values (i.e. 4.5 and 7.0,) while
bicomponent adsorption studies of AO and ST on the CTA adsor-
bent were only carried out at pH 7.0, which was the pH of highest
adsorption capacity for both dyes. All adsorption studies were per-
formed in duplicate.
2.2.1. Adsorption of dyes onto CTA as a function of solution pH
For a typical monocomponent experiment, CTA samples (20.0
mg) weighed in cylindrical glasses (1.8 mm height  2.2 mm
diameter) were added to 250 mL Erlenmeyer flasks containing
100.0 mL of buffered dye solution of a known concentration
(0.932 mmol L
1) at pH values varying from 2.13 to 7.14. Flasks
were stirred at 130 rpm and 25 °C in a shaker incubator (Marconi,
model MA-830) for 24 h. Then, the suspensions were centrifuged
Ò
(Excelsa II centrifuge, model 206 BL) at 3600 rpm for 10 min to
separate the solid and liquid phases. The absorbance of the super-
natant phase was measure on a UV–Vis spectrophotometer
(Biospectro, model SP-220) at 434 nm for AO and 530 nm for ST.
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
Calibration curves were constructed to determine the dye concen-
tration in the supernatant phase.
For bicomponent experiments, the procedures described above
were used, except that buffered dye solutions contained both dyes
at known concentrations (0.93 mmol L
1). In a bicomponent dye
solution, the concentrations of A (AO) and B (ST), measured at k1
and k2, respectively, to give absorbances of A1 and A2, were deter-
mined using Eqs. (1) and (2) [13]:
kB2 A1
kB1 A2
CA ¼
kA1 kB2
kA2 kB1
kA1 A2
kA2 A1
CB ¼
kA1 kB2
kA2 kB1
where kA1 (0.12957), kB1 (0.01471), kA2 (0.0), and kB2 (0.10750) are
the calibration constants for components A and B at the wave-
lengths k1 and k2, respectively.
2.2.2. Adsorption of dyes onto CTA as a function of contact time
In a typical run for monocomponent experiments, Erlenmeyer
flasks of 250 mL containing 100.0 mL of buffered dye solution of
a known concentration (0.932 mmol L
1) at pH values of 4.5 and
7.0 were thermostated in a shaker incubator at 25 °C for 1 h. Then,
CTA samples of 20.0 mg weighed in cylindrical glasses were added
to each flask and stirred at 130 rpm and 25 °C for time intervals
from 5 to 1920 min. Adsorption capacity was calculated using Eq.
(3).

1 ðC 0
C e ÞV
qt =ðmmol g Þ ¼
wCTA
where qt (mmol g
1) is the adsorption capacity of AO or ST on the
CTA adsorbent at time t, V (L) is the volume of dye solution, C0
and Ct (mmol L
1) are the dye solution concentrations initially
and at time t, respectively, and wCTA (g) is the weight of the CTA
adsorbent.
For bicomponent experiments, the same experimental proce-
dures described above were used, except that buffered dye solu-
tions contained both dyes at known concentrations (0.932 mmol
L
1) at pH value 7.0 were added to the Erlenmeyer flasks. The flaks
were stirred for different time intervals from 5 to 1800 min. Other
procedures were the same as described in Section 2.2.1.
2.2.3. Adsorption of dyes onto CTA as a function of adsorbent dosage
CTA samples of 10.0, 20.0, 30.0, 40.0, 50.0, 60.0 and 70.0 mg
weighed in cylindrical glasses were added to 250 mL Erlenmeyer
flasks containing 100.0 mL of buffered dye solution of a known
concentration (0.932 mmol L
1) at pH 7.0. The flaks were stirred
at 130 rpm and 25 °C for 24 h. Other procedures were the same
as described in Section 2.2.1. The dye removal percentage was cal-
culated using Eq. (4).


C0
Ce
R=% ¼  100
C0
where C0 and Ce (mmol L
1) are the dye solution concentrations ini-
tially and at equilibrium, respectively.
2.2.4. Adsorption of dyes onto CTA as a function of initial dye
concentration
In typical monocomponent experiments, CTA samples of 20.0
mg weighed in cylindrical glasses were added to 250 mL Erlen-
meyer flasks containing 100.0 mL of buffered dye solutions at pH
values of 4.5 and 7.0 with dye concentrations varying from 0.21
to 2.2 mmol L
1 for AO and 0.25 to 1.38 mmol L
1 for ST. The flasks
were stirred at 130 rpm and 25 °C for 24 h. The adsorption capacity
was calculated using Eq. (5).
577

1 ðC i
C e ÞV
qe =ðmmol g Þ ¼ ð5Þ
wCTA
where qe (mmol g
1) is the adsorption capacity of dye on the CTA
adsorbent at equilibrium and Ci and Ce (mmol L
1) are the dye solu-
tion concentrations initially and at equilibrium, respectively.
For bicomponent experiments, 100.0 mL of buffered dye solu-
tions at pH 7.0 with dye concentrations varying from 0.17 to
1.06 mmol L
1 for AO and 0.19 to 1.16 mmol L
1 for ST were added
to the flasks, which were stirred at 130 rpm and 25 °C for 24 h.
ð1Þ
Other procedures were the same as described in Section 2.2.1.
2.3. Desorption experiments
ð2Þ
CTA adsorbent was loaded with AO or ST using the same exper-
imental procedures described in Section 2.2.1. Dye solutions of
0.87 mmol L
1 for AO and 0.99 mmol L
1 for ST at pH 7.0 were
used. After loading the CTA adsorbent with a dye, the CTA adsor-
bent was separated by a single filtration step and was rinsed with
an excess of distilled water to remove unadsorbed dye molecules.
Then, the CTA adsorbent loaded with dye was dried in an oven at
90 °C for 3 h.
2.3.1. Desorption of the CTA adsorbent as a function of desorption
agent type
Two desorption agents (HCl and HNO3) at two concentrations
(0.01 and 0.1 mol L
1) were evaluated for desorption of CTA adsor-
bent loaded with AO or ST. Samples of CTA loaded with a dye (20.0
mg) weighed into cylindrical glasses were added to 100 mL Erlen-
ð3Þ
meyer flasks containing 20.0 mL of desorption solution and stirred
at 130 rpm and 25 °C for 16 h. Solid and liquid phases were sepa-
rated by centrifugation. The pH of the supernatant was adjusted
to the pH at which the calibration curve was obtained by dropwise
addition of aqueous 0.1 mol L
1 NaOH. Absorbance was measured
on a UV–Vis spectrophotometer as described in Section 2.2.1. All
desorption experiments were performed in triplicate. Desorption
efficiency was calculated using Eq. (6) [10].


C e;des V
Edes =% ¼  100 ð6Þ
Q T;max w0CTA
where Edes (%) is the desorption efficiency, Ce,des (mg L
1) is the
equilibrium dye concentration in the desorption solution, QT,max
(mg g
1) is the maximum adsorption capacity determined by load-
ing a dye (AO or ST) on the CTA adsorbent, and w0 CTA (g) is the
weight of the CTA adsorbent in wCTA,dye, calculated using Eq. (7)
[10].
wCTA;dye
w0CTA =g ¼ Q  ð7Þ
T;max
þ1
1000
2.3.2. Desorption of the CTA adsorbent as a function of time
ð4Þ Dried samples of CTA loaded with a dye (20.0 mg) weighed into
cylindrical glasses were added to 250 mL Erlenmeyer flasks con-
taining 100.0 mL of aqueous 0.01 mol L
1 HCl solution and stirred
at 130 rpm and 25 °C for different time intervals (1, 3, 6, 12, and 24
h). Other procedures were the same described in Section 2.3.1.
2.4. Re-adsorption experiments
Dried samples of CTA adsorbent (20.0 mg) from the desorption
study weighed into cylindrical glasses were added to 250 mL
Erlenmeyer flasks containing 100.0 mL of a dye solution (0.932
mmol L
1). Other procedures were the same described in Sec-
tion 2.2.1. All re-adsorption experiments were performed in tripli-
cate. The re-adsorption efficiency was calculated using Eq. (8) [10].
578 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590


Q e;re
ads
Ere
ads =% ¼  100
Q T;max
where Qe,re-ads (mg g
1) is the adsorption capacity of the CTA adsor-
bent after the desorption process and QT,max (mg g
1) is the adsorp-
tion capacity of the CTA adsorbent before the desorption process.
Qe,re-ads is calculated using Eq. (9) [10].
w0dye þ w00dye
Q e;re
ads =ðmg g
1 Þ ¼
wCTA
0
where w dye (mg) is the weight of dye not desorbed from the CTA
adsorbent after desorption and w’0 dye (mg) is the weight of dye
adsorbed on the CTA adsorbent after re-adsorption.
2.5. Isothermal titration calorimetry
Calorimetry experiments to determine the changes in enthalpy
of dye adsorption on CTA were performed in a TAM III isothermal
titration nanocalorimeter (TA instruments, EUA) controlled by
TAM assistantTM dedicated software. A concentrated dye solution
(330 mg mL
1 for both ST and AO) prepared in citrate buffer was
titrated into the reaction cell that initially contained (i) 2.7 mL of
citrate buffer or (ii) 2.7 mL of citrate buffer plus 1.35 mg of CTA.
A volume of 2.7 mL of buffer filled the reference cell. Injections
of 40 lL of the dye solution were carried out stepwise using a
500 lL Hamilton syringe controlled by a piston pump. The time
interval between two consecutive injections was 40 min. The reac-
tion cell was stirred at 180 rpm with a helix stirrer during experi-
ments. Solutions were degassed for 10 min before experiments.
Measurements were performed at pH 4.5 and 7.0. The heat, in kJ,
resulting from each injection i was determined by integration of
the plots of potency against time provided by the equipment.
The adsorption enthalpy change (DadsH), in kJ mol
1, was obtained
from Eq. (10).
X
N (nCOOH: 6.81 ± 0.02 mmol g
1), the PZC (2.80 ± 0.09), and the textu-
ðqi;int
qi;dil Þ
i¼1
Dads H ¼
X
N
ni
i¼1
where ce is the activity coefficient at equilibrium and b (L mol
1) is
ð8Þ
the Langmuir constant. The activity coefficients of the present study
were calculated using the approximations of Guntelberg [15] and
Davies [16] (Supplementary Material), as suggested by Liu [14]
and Ghosal and Gupta [17]. For AO and ST isotherms at pH 4.5,
the values of the Langmuir constant used in the calculations were
those of the first plateaus (b1).
ð9Þ 2.7. Molecular quantum mechanics calculations
To provide insight into the structural and electronic behavior of
auramine-O and safranin-T, as well as the monomer model of the
carboxylated cellulose derivative (CTA) (Fig. 1), density functional
theory (DFT) calculations were carried out, implemented in the
electronic structure program package ORCA [18]. The effects of
increased acidity on both dyes were evaluated from calculations
of dyes’ different ionization states. Becke’s three-parameter hybrid
function was used with the non-local correlation of the Lee-Yang-
Parr (B3LYP) method in the gas phase [19–22], as this method has
been used successfully for a variety of organic molecules [23–26].
This functional was used with the Pople’s basis set 6-31G(2d,2p),
which has polarization functions to make it more flexible [27].
Frontier molecular orbitals (FMOs) and molecular electron density
(MED) maps were also calculated at the same level of theory. All
molecular graphics were built using the UCSF Chimera visualiza-
tion package [28].
3. Results and discussion
3.1. Synthesis and characterization of the CTA adsorbent
A detailed characterization of the CTA adsorbent was described
in the first part of this series of studies [10]. For this study, it is
helpful to summarize the amount of free carboxylic acid groups
ral properties of the CTA adsorbent (maximum pore diameter of
ð10Þ 570.95 ± 38.25 Å) [10].
3.2. Theoretical analysis

where qi,int and qi,dil are the energies absorbed or released in the
3.2.1. Model of the CTA adsorbent
reaction cell with and without CTA, respectively, at the ith injection
The CTA adsorbent’s electronic properties were estimated from
of the dye solution, and ni is the amount of dye, in mol, that was
the oligomer model shown in Fig. 1. Only the monomer was ana-
adsorbed on the adsorbent for the same injection. The adsorption
lyzed because the other bonds along the main chain are saturated,
isotherms were used to obtain values of ni. A control experiment
and therefore have minor effects on the electronic structure
adding only buffer to buffer containing 1.35 mg of CTA was per-
results.
formed, and the signal measured was of the same magnitude as
equipment noise. Experiments were carried out at 25.0000 ±
0.0001 °C in duplicate.

2.6. Calculation of free energy of adsorption

Changes in free energy of adsorption (DadsG°) were determined

in a similar way as described in the first part of this series of stud-
ies [10], with the exception of calculating the thermodynamic
equilibrium constant, Ka [14]. Thus, DadsG° values were calculated
using Eq. (11).

1
Dads G =ðkJ mol Þ ¼
RT ln K a ð11Þ

1
where R is the gas constant (8.314 kJ mol ), T (K) is the absolute
temperature, and Ka (dimensionless) is the thermodynamic equilib-
rium constant. Values of Ka for the adsorption system were calcu-
lated using Eq. (12).
Fig. 1. Fully optimized CTA oligomer model, calculated using the level of theory
b
1 B3LYP/6-31(2d,2p). Oxygen atoms are depicted by red, carbon by light brown, and
Ka ¼ ð1 mol L Þ ð12Þ hydrogen by white. (For interpretation of the references to colour in this figure
ce
legend, the reader is referred to the web version of this article.)
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
Fig. 2 shows that the inner atoms in both rings form a positive
site (blue-cyan in the CTA MED map), including the ester and car-
boxyl groups. In addition, the highest electron density (red areas) is
localized outside the rings on the oxygen atoms of ether and car-
boxyl groups. The frontier orbitals of CTA are presented in Supple-
mentary Fig. 1, which shows that electron donor character is due to
the non-aromatic ring of cellulose. However, the electron with-
drawing nature is attributable to the aromatic ring and carboxyl
groups of the trimellitate moiety.
3.2.2. Auramine-O (AO)
The neutral and monocation states of AO were considered (see
Fig. 3 for optimized structures), since dye removal was studied at
multiple pH values. As can be seen in Fig. 3, the equilibrated struc-
ture of the neutral AO is planar, while for the monocation state
(AO+1) an approximately 30-degree dihedral angle formed between
the aromatic rings.
MED maps exhibited similar profiles for both species, showing
higher electron densities localized on tertiary amines (red areas)
and lower densities in the central region (in blue), as shown in
Fig. 4. The frontier molecular orbitals HOMO and LUMO of AO
are shown in Supplementary Fig. 2.
3.2.3. Safranin-T (ST)
The +1 and +2 ionization states were considered for ST. ST+2 was
modeled by protonation of the primary amine. The electronic
structure calculations carried out for the fully optimized ST+1 and
ST+2 showed that in both cases the phenyl rings are displaced per-
pendicularly to the condensed rings, as shown in Fig. 5. The MED
maps have similar shapes for both cationic ST species, as shown
in Fig. 6. In these maps, it was observed that the amine groups in
the vicinity of the protonation sites have higher electron density
(red color). The frontier molecular orbitals HOMO and LUMO of
ST are shown in Supplementary Fig. 3.
3.3. Adsorption studies
Fig. 7a–c shows suggested adsorption mechanisms for AO and
ST removal from a single aqueous solution by the CTA adsorbent,
as well as a desorption mechanism for a cationic dye (AO or ST)
from the CTA adsorbent.
3.3.1. Effect of CTA dosage on dye removal
Fig. 8 shows the effect of CTA dosage on adsorption capacity and
removal percentage of AO. The removal percentage of AO increased
as the CTA dosage was increased due to the increase in available
adsorption sites. Contrarily, the amount of AO adsorbed per unit
weight of CTA (adsorption capacity) decreased with increasing
Fig. 2. Molecular electron density maps for the model of cellulose oligomer
calculated using the B3LYP/6-31(2d,2p) level of theory.
579
CTA dosage due to the decrease in the adsorbate-to-adsorbent ratio
[29]. Based on these results, subsequent adsorption experiments
were carried out with a CTA dosage of 0.2 g L
1, because greater
dosages led to significant decreases in the adsorption capacity of
CTA.
3.3.2. Effect of solution pH on dye removal
AO and ST removal is pH-dependent. Fig. 7d and e shows that
both AO and ST are positively charged in the pH range studied,
but ST may have one or two positive charges depending on the
solution pH. The PZC of the CTA adsorbent is 2.80. Thus, the
adsorption of the cationic dyes AO and ST should be favored at
pH > 2.8, where the CTA adsorbent has a net negative surface
charge. Fig. 9 shows the removal of the dyes AO and ST as a func-
tion of solution pH from 2.13 to 7.14.
The removal of both dyes increased as the solution pH increased
for mono- and bi-component systems, and reached a plateau at pH
values greater than 6 for AO and greater than 5 for ST, for the
monocomponent system. The adsorption of AO and ST increased
slightly until pH values greater than 7 for the bicomponent system.
At pH values greater than the PZC, the removal of both dyes is
expected to be controlled mainly by electrostatic interaction
between the negatively charged carboxylate groups on the CTA
adsorbent surface and the positively charged dye molecules. At
low pH values (pH < 3), the CTA carboxylate groups were proto-
nated, substantially reducing removal of both dyes. At pH values
closer to 2, the removal of AO was practically zero, while ST was
still removed by the CTA adsorbent in both mono- and bi-
component systems. Thus, at pH values lower than the PZC, the
removal of both dyes is expected to occur through hydrogen bond-
ing, dipole-dipole interactions, and p-p bonding (p stacking). In
addition, the adsorption capacities of AO and ST for the monocom-
ponent system were higher than for the bicomponent system,
showing that in the pH range studied the adsorption of one dye
is suppressed by the presence another dye.
Basic dyes such as AO and ST are primarily used to dye wool,
silk, acrylic fibers, and leather. In addition, exhaustion of dye baths
is sensitive to pH change and therefore dye baths containing basic
dyes may have pH values between 4 and 5 depending on the type
of fiber and its activation (e.g. presence of carboxylic or sulfonic
acid groups [6]). Thus, in the present study, removal of dyes AO
and ST was studied at two pH values (i.e. 4.5 and 7.0) as the CTA
adsorbent exhibits a good adsorption capacity at both pH values.
Removal of basic dyes at pH 4.5 is advantageous because the pH
of the dye effluent does not need to be changed, which decreases
the cost of the treatment. However, the removal of basic dyes at
pH 7.0 can decrease the amount of adsorbent required to treat
the effluent, as the adsorption capacity of the CTA adsorbent at
pH 7.0 is greater than at pH 4.5.
3.3.3. Mono- and multi-component adsorption kinetics
Two critical parameters for designing a wastewater treatment
plant (WWTP) operating in batch or semicontinuous mode are
equilibrium time and adsorption rate [30]. Thus, to study the kinet-
ics of adsorption of dyes on CTA adsorbent, pseudo-first-order
(PFO) [31], pseudo-second-order (PSO) [32], Boyd [33], and intra-
particle diffusion (IPD) [34] models were used to model monocom-
ponent adsorption, while PFO, PSO, and a competitive kinetic
model of Corsel et al. [35] were used to model bicomponent
adsorption. These models were described in the first [10] and sec-
ond [11] parts of this series of studies.
Experimental data were modeled by nonlinear regression anal-
ysis (NRL) using Microcal OriginProÒ 2015 and MATLAB 2014
(Mathworks, Inc.) software. Data fit quality for the mono- and bi-
component experimental data were evaluated as described in the
580 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590

Fig. 3. Fully optimized structures calculated for auramine-O in two states (a) neutral and (b) monocation, using the level of theory B3LYP/6-31(2d,2p). Nitrogen atoms are
blue, carbon are light brown, and hydrogen are white. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

Fig. 4. Molecular electron density maps calculated for the auramine-O in two states (a) neutral and (b) monocation AO+1, using the level of theory B3LYP/6-31(2d,2p). Red
color indicates high electron density, while blue indicates low electron density. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

Fig. 5. Fully optimized structure obtained for (a) ST+1 and (b) ST+2 using the level of theory B3LYP/6-31(2d,2p). Nitrogen atoms are blue, carbon light brown, and hydrogen are
white. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Molecular electron density maps calculated for ST+1 (a) and ST+2 (b) using the level of theory B3LYP/6–31(2d,2p). Red indicates high electron density, while blue
indicates low electron density. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

first [10] and second [11] parts of this series of studies,
respectively.
3.3.3.1. Monocomponent adsorption. Table 1 shows the parameters
estimated by modeling monocomponent experimental data with
the PFO, PSO, Boyd, and IPD models. Experimental data are avail-
able in Supplementary Tables 2 and 3. Fig. 10a–b shows plots of
qt over t for the monocomponent removal of AO and ST by the
CTA adsorbent, with the curves fit to experimental data using the
PFO and PSO kinetic models. As can be seen in Table 1, the PSO
model best described AO and ST removal rate at both pH values,
based on a higher coefficient of determination (R2), smaller
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590 581

Fig. 7. Suggested adsorption mechanism of auramine-O (a) and safranin-T (b) on the CTA adsorbent. (c) Desorption mechanism of a cationic dye (AO or ST) from the CTA
adsorbent surface. Dyes structures as a function of the solution pH for (d) auramine-O and (e) safranin-T.

reduced chi-square (v2red), and a smaller residual sum of squares estimated by the PSO model for AO and ST were closest to qe,exp val-
(RSS). Furthermore, experimental adsorption capacities (qe,exp) were ues. In addition, the initial adsorption rate (h) for the removal of both
compared to estimated adsorption capacities (qe,est), qe,est values dyes by the CTA adsorbent was strongly affected by the solution pH.
582 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
Fig. 8. Effect of CTA dosage on AO adsorption capacity (s) and removal percentage (d) (C0 = 0.932 mmol L
1, 130 rpm, 25 °C, and 24 h of agitation).
ðR
COOÞ2
dye þ 2 H3 Oþ
Fig. 9. Mono- and bi-component removal of auramine-O (AO) and safranin-T (ST)
by the CTA adsorbent as a function of the solution pH (25 °C, 130 rpm, 0.2 g L
1 CTA,
and 0.932 mmol L
1 AO and ST for mono- and bi-component systems).
The value of h for AO and ST increased 4 and 2.4 times, respectively,
as the solution pH increased from 4.5 to 7.0. Additionally, the equi-
librium adsorption capacities for AO and ST were higher at pH 7.0
than at pH 4.5, indicating that the CTA adsorbent surface has a
greater affinity towards AO and ST at higher pH values.
As discussed above, the PSO kinetic model best described the
removal of AO and ST by CTA adsorbent in the monocomponent
system. Eqs. (13) and (14) represent two possible forms of interac-
tion of AO and ST onto CTA adsorbent at an adsorption site. Eq. (14)
best describe ST adsorption as the pKa of ST is 6.4 and at pH 7.0 a
fraction of ST molecules still have two positive charges, while Eq.
(13) is most appropriate to describe AO adsorption. As suggested
by the molecular electron density maps calculated for AO+1, ST+1,
and ST+2, two forms of interaction, that is electrostatic attraction
and van der Waals (dispersive forces, hydrogen bonding, and/or
ion-dipole), may occur depending on the solution pH.
þ shows a graph of C overt t with the curves fit to the experimental
R
COOH þ dyeðaqÞ þ H2 O
R
COO
dye þ H3 Oþ ð13Þ
2 R
COOH þ dyeðaqÞ þ 2 H2 O
2þ
ð14Þ
Supplementary Fig. 4a–d present the intraparticle diffusion graphs
for monocomponent adsorption of AO and ST on the CTA adsorbent
at pH 4.5 and 7.0. These graphs demonstrate multilinear behavior
and the first straight lines for both dyes and pH values did not inter-
sect the origin, indicating that adsorption of dyes on the CTA adsor-
bent is initially controlled by film diffusion and then by intraparticle
diffusion until equilibrium is reached [36]. Considering that the
molecular sizes of AO and ST are 14.73 Å  4.92 Å and 11.36 Å 
8.10 Å, respectively, it is expected that adsorption of AO and ST
can occur in both micropores and mesopores, with adsorption in
micropores being slower than in mesopores due to micropores’
lower adsorbate-to-pore diameter ratio [37]. In addition, the calcu-
lated molecular volumes for neutral AO and AO+1 are 864.51 Å and
872.95 Å, while for ST+1 and ST+2 the values are 933.96 Å and
944.12 Å, respectively. The intraparticle diffusion graphs show
three stages for AO and ST adsorption at both pH values. The intra-
particle diffusion coefficients (ki) for the first and second stages of
adsorption are presented in Table 1.
Supplementary Fig. 5a–d present the Boyd graphs for AO and ST
adsorption at pH 4.5 and 7.0. The slopes of the Boyd graphs (B)
were obtained from the linear portions of the graphs of Bt over t
and were used for calculating effective diffusion coefficients (Di)
[36] (Table 1). The Boyd graphs were linear in the initial adsorption
period and did not intersect the origin. Values of B were close to
zero, suggesting that external mass transfer plays an important
role in the adsorption of AO and ST, but lower values of ki,2 com-
pared to ki,1 confirm that the rate-limiting step controlling the
adsorption of AO and ST is intraparticle diffusion.
3.3.3.2. Bicomponent adsorption. Table 2 shows the estimated
kinetic parameters obtained by modeling bicomponent
experimental kinetic data with PFO, PSO, and Corsel models.
Bicomponent experimental kinetic data are available in Supple-
mentary Table 4. Supplementary Fig. 6a shows a graph of qt over
t, with the curves fit to experimental data using the PFO and PSO
models for the AO-ST system at pH 7.0. Supplementary Fig. 6b
data using the Corsel model for bicomponent adsorption
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590 583

Table 1
Results of modeling monocomponent experimental data for the removal of auramine-O and safranin-T (C0 = 0.932 mmol L
1, 25 °C, 130 rpm and 0.2g L
1 CTA).

Monocomponent systems
Parameters pH 4.5 pH 7.0
Auramine-O Safranin-T Auramine-O Safranin-T
te (min) 720 720 300 450
qe,exp (mmol g
1) 2.721 ± 0.130 2.621 ± 0.190 3.111 ± 0.025 3.863 ± 0.045
Pseudo-first-order (PFO) kinetic model
qe,est (mmol g
1) 2.222 ± 0.073 2.424 ± 0.095 3.037 ± 0.042 3.705 ± 0.059
k1 (min
1) (1.48 ± 0.19)  10
2 (1.71 ± 0.00)  10
2 (4.03 ± 0.32)  10
2 (2.78 ± 0.21)  10
2
R2 0.9376 0.8548 0.9549 0.9673
v2red 0.0208 0.0373 0.0076 0.0110
RSS 0.2911 0.5589 0.1143 0.9673
Pseudo-second-order (PSO) kinetic model
qe,est (mmol g
1) 2.449 ± 0.053 2.602 ± 0.070 3.226 ± 0.024 3.986 ± 0.041
k2 (g mmol
1 min
1) (8.53 ± 0.95)  10
3 (1.01 ± 0.15)  10
2 (1.96 ± 0.10)  10
2 (1.04 ± 0.07)  10
2
h (mmol g
1 min
1) (5.11 ± 0.58)  10
2 (2.04 ± 0.11)  10
1 (6.87 ± 1.02)  10
2 (1.66 ± 0.11)  10
2
R2 0.9805 0.9474 0.9904 0.9901
v2red 0.0065 0.0134 0.0016 0.0033
RSS 0.0911 0.2005 0.0241 0.4999
Intraparticle diffusion (IPD)
Step 1
ki,1 (mmol g
1 min
1/2) 0.215 ± 0.045 0.129 ± 0.021 0.297 ± 0.029 0.281 ± 0.022
C (mmol g
1) -0.237 ± 0.259 0.460 ± 0.163 0.340 ± 0.181 0.522 ± 0.200
R2 0.9586 0.9065 0.9819 0.9751
Step 2
ki,2 (mmol g
1 min
1/2) 0.079 ± 0.009 0.037 ± 0.005 0.073 ± 0.009 0.018 ± 0.009
C (mmol g
1) 0.776 ± 0.128 1.607 ± 0.09 2.044 ± 0.100 3.343 ± 0.163
R2 0.9697 0.9426 0.9731 0.4450
Boyd plot
B (min
1) (6.06 ± 0.34)  10
3 (6.71 ± 0.74)  10
3 (1.15 ± 0.13)  10
2 (1.35 ± 0.05)  10
2
Di (m2 min
1) 9.59  10
12 9.77  10
12 1.82  10
11 2.14  10
11
R2 0.9495 0.9195 0.9915 0.9293

Fig. 10. Adsorption kinetics of auramine-O (AO) and safranin-T (ST) on the CTA adsorbent in equimolar concentration (CAO = CST = 0.932 mmol L
1) at 25 °C, 130 rpm, and 0.2
g L
1. Monocomponent systems of AO (a) and ST (b) at pH 4.5 and 7.0 were modeled by pseudo-first- (PFO) and pseudo-second-order (PSO) kinetic models.

(Supplementary Table 5) for the AO-ST system at pH 7.0. Adsorp-
tion of AO was suppressed by the presence of ST and likewise
adsorption of ST was suppressed by the presence of AO (Table 2).
However, in bicomponent systems the decrease in qe for AO
adsorption was greater than the decrease in qe for ST adsorption
compared to qe values calculated in monocomponent systems. In
addition, AO was adsorbed faster (4.5 times) on the CTA adsorbent
than ST was. The value of te for AO adsorption decreased from 300
to 240 min, while the value of te for ST remained constant (450
min) when compared to values for AO and ST adsorption in the
monocomponent system. The model that best described bicompo-
nent adsorption values for the systems AO-ST and ST-AO was the
PFO kinetic model, which yielded higher R2, smaller v2red, and smal-
ler RSS. It should be noted that the adsorption mechanism was differ-
ent in monocomponent and bicomponent systems.
As pointed out by Teodoro et al. [11], the PFO and PSO kinetic
models can only be used for comparing the adsorption rates and
qe values of dyes in binary systems. These models cannot predict
the influence of a dye i on the adsorption of a dye j, and vice versa.
Therefore, a competitive kinetic model developed by Corsel et al.
584 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590

Table 2 lower diffusion rates. Values of the unstirred layer (d) for bicompo-
Results of modeling bicomponent experimental data for the removal of auramine-O nent adsorption were small (Table 2).
and safranin-T (Ci = Cj = 0.932 mmol L
1, 25 °C, 130 rpm and 0.2g L
1 CTA).
Differences in adsorption kinetic parameters, that is rate con-
Bicomponent system (pH 7.0) stants, equilibrium adsorption capacities, and equilibrium times,
Parameters AO-ST ST-AO reveal that multicomponent adsorption of dyes requires further
te (min) 240 450
research, and the application of new adsorbent materials for treat-
qe.exp (mmol g
1) 1.051 ± 0.029 3.258 ± 0.021 ment of effluents containing dyes must be studied in detail to pre-
Pseudo-first-order (PFO) kinetic model
dict adsorbent behavior for target dye adsorption.
qe.est (mmol g
1) 1.057 ± 0.016 3.204 ± 0.226
k1 (min
1) (7.22 ± 0.59)  10
2 (1.59 ± 0.41)  10
2 3.3.4. Mono- and multi-component adsorption isotherms
R2 0.9737 0.8617 Adsorption isotherms can elucidate the physical phenomenon
v2red 0.0035 0.1492
governing the removal of an adsorbate by an adsorbent in aqueous
RSS 0.0556 2.3878
phase at a constant temperature and pH [29]. In addition, adsorp-
Pseudo-second-order (PSO) kinetic model
tion isotherms are useful for predicting the maximum adsorption
qe.est (mmol g
1) 1.108 ± 0.032 3.591 ± 0.318
k2 (g mmol
1 min
1) (8.89 ± 1.71)  10
2 (5.34 ± 2.25)  10
3 capacity of an adsorbent for one or more adsorbates and therefore
h (mmol g
1 min
1) (1.90 ± 0.21)  10
1 (6.88 ± 2.97)  10
2 are crucial to design of batch and semicontinuous adsorption sys-
R2 0.9213 0.8497 tems for WWTPs [39]. Five adsorption isotherm models were used
v2red 0.0104 0.1621
to describe the monocomponent removal of AO and ST by CTA
RSS 0.1664 2.5941
adsorbent: Langmuir [40], Sips [41], Fowler-Guggenheim (F-G)
Corsel kinetic model
[42], Hill-de Boer (H-B) [43,44], and Konda [45]. For a detailed
a 0.0129 0.0742
b 0.1322 0.7414
description of the first two isotherm models see the first part of
c 0.1451 0.8156 this series of studies [10]. In addition, IAST and RAST models were
ci - cj
0.6705 used to model bicomponent equilibrium data.
k+1 (L mmol
1 min
1) 64.35 181.89 The Fowler-Guggenheim (F-G) isotherm was developed by Fow-
k-1 (min
1) 0.8714 5.0342
ler and Guggenheim [42]. It assumes localized adsorption and lat-
k+1/k-1 (L mmol
1) 73.8467 36.1309
d (m) 1.829  10
9 eral interaction between adsorbed molecules on the surface of an
Di (m2 min
1)b 1.82  10
11 2.14  10
11 adsorbent. The F-G isotherm is given by Eq. (15).
Cmax (mmol m
2) 0.321 1.024
qt (mmol g
1)a 1.026 3.271 h
R2 K FG C e ¼ expð
chÞ ð15Þ
0.8364 0.8962 1
h
v2red 0.0061 0.0094
RSS 0.0216 0.1014 where KFG (L mmol
1) is the Fowler-Guggenheim constant, h is frac-
a
Calculated from a specific surface area of 3.196 ± 0.217 m2 g
1 (Teodoro et al. tional surface coverage (qe Q
1 max), and c (kJ mol

1
) is a constant
[10]). related to the energy involved in the interaction between adsorbed
b
Converted values from the data reported by Teodoro et al. [10]. molecules. If the interaction between adsorbed molecules is attrac-
tive, the c > 0 and the heat of adsorption will increase with increas-
ing surface coverage. On the contrary, if the interaction between
[35] for protein adsorption was adopted to evaluate the kinetics of
adsorbed molecules is repulsive, c < 0 and the heat of adsorption
the bicomponent systems AO-ST and ST-AO. The kinetic model of
will decrease with increasing the surface loading. When there is
Corsel et al. [35] is based on a generalized diffusion equation,
no lateral interaction between adsorbed molecules c = 0 and the
allowing it to include the effects of surrounding bulk medium
F-G isotherm reduces to the Langmuir isotherm [46].
and an unstirred layer. The association constant (K = k+1/k-1), which
The Hill-de Boer (H-B) isotherm was developed by Hill [43] and
is the ratio between the intrinsic adsorption (k+1) and desorption
de Boer [44]. It assumes mobile adsorption and lateral interaction
(k-1) rate constants, was higher for AO than for ST, suggesting that
between adsorbed molecules. The H-B isotherm is given by Eq.
the adsorption sites of the CTA adsorbent had a higher affinity for
(16).
AO than for ST (Table 2).


The interaction constants a and b can give relevant information h h
on possible lateral interactions between adsorbed solutes in a K HB C e ¼ exp
ch ð16Þ
1
h 1
h
bicomponent system. Smaller positive values of cooperative con-
stants (a) (Table 2) imply that dye molecules do not tend to adsorb where KHB (L mmol
1) is the H-B constant. When c is positive, there
near other adsorbed dye molecules. This inhibition is probably due is an attractive force between adsorbed molecules, while a negative
to a repulsive potential around adsorbed dye molecules or steric value means there is repulsion between adsorbed molecules. When
hindrance of the adsorbed dye molecules, which may hinder dye c = 0, the H-B isotherm is reduced to the Volmer isotherm that
molecules’ adsorption on adjacent sites [38]. Therefore, the higher assumes mobile adsorption but no lateral interaction between
the value of cooperative constants (a), the greater the decrease in adsorbed molecules [46].
the adsorption rate function as the CTA adsorbent surface becomes The Konda isotherm was developed by Konda et al. [45] to
saturated. On the other hand, the higher the value of cooperative describe single- and multi-step adsorption isotherms of organic
constants (b), the greater the increase in the desorption rate func- pesticides on soil. This isotherm is derived from the Langmuir iso-
tion as the CTA adsorbent surface becomes saturated. therm with the additional assumption that sterically or energeti-
If dkonCmax < < D, diffusion has no influence on adsorption cally heterogeneous adsorption sites are present on the
kinetics [35]. For AO and ST adsorption, values of dkonCmax were adsorbent surface. It also assumes potential interactions between
3.92  10
11 and 3.41  10
10, respectively. Thus, as pointed out adsorbed molecules, and that each step on the isotherm represents
for monocomponent adsorption of AO and ST, diffusion played an a different type of adsorption mechanism [45]. The Konda model is
important role in the bicomponent adsorption process of AO and given by Eq. (17).
ST on the CTA adsorbent. The thickness of the unstirred layer of
Q 1 b1 C e Q b2 ½ðC e
c2 Þ þ jC e
c2 j
water adjacent to CTA adsorbent particles can play an important qe ¼ þ 2 ð17Þ
1 þ b1 C e 2 þ b2 ½ðC e
c2 Þ þ jC e
c2 j
role in the adsorption process, because thicker layers result in
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
where Qi (mmol g
1) is the adsorption capacity for the ith adsorp-
tion step, bi (L mmol
1) is the adsorption equilibrium constant,
and ci (mmol L
1) is the limit concentration. The limit concentration
for the first adsorption step is zero. The maximum adsorption
P
capacity is defined by Qmax = Qi.
The ideal adsorbed solution theory (IAST) was proposed by
Myers and Prausnitz [47] for adsorption of a gas mixture and
was extended to the liquid phase by Radke and Prausnitz [48].
The IAST model provides a thermodynamically consistent and
practical method for predicting bicomponent adsorption isotherms
using only monocomponent isotherm data [13]. The RAST model
was proposed by Costa et al. [49] and is an extension of the IAST
model proposed by Myers [50] and Talu and Zwiebel [51] for mul-
ticomponent adsorption of non-ideal gas mixtures. The RAST
model is more appropriate than the IAST model to describe adsorp-
tion systems that exhibit synergistic or antagonistic behavior. The
RAST model is a derivation of the IAST model. However, it assumes
that the adsorbed phase is deviates from ideality [11].
Multicomponent adsorption equilibria using the IAST and RAST
models is predicted by two variables for IAST (i.e. the solid-phase
mole fractions and the reduced spreading pressure), and six vari-
ables for RAST (i.e. the solid-phase mole fractions, the reduced
spreading pressure, and adjustable parameters of the activity coef-
ficient model). The spreading pressure-dependent Wilson equation
was used for determination of the activity coefficients (c, Ʌ12, and
Ʌ21). The Sips isotherm was used to model monocomponent
adsorption equilibria to obtain the model parameters Qmax, b, and
n. The theoretical approach for the IAST and RAST models and
the solution methods used for both models is presented in detail
in the second part of this series of studies [11], in the Supplemen-
tary Material, and in the studies of Erto [52,53]. As reported by Erto
et al. [53], linear regression analysis using the modified Wilson
equation parameters ci and cj (see Eqs. (36) and (37) reported by
Teodoro et al. [11]) for the entire experimental data did not result
in a good linear fit. Thus, the approximation suggested by Erto et al.
[53] (see Eqs. (17)–(21) described by Erto et al. [53]) was applied in
the present study to calculate the value of cj from the value of ci,
which was obtained by regression analysis of the modified Wilson
equation (Eq. (36) described in the study of Teodoro et al. [11]).
3.3.4.1. Monocomponent adsorption. Table 3 shows the parameters
estimated by modeling the monocomponent experimental data
(Supplementary Tables 6 and 7) with the Langmuir, Sips, Fowler-
Guggenheim, Hill-de Boer, and Konda isotherm models through
NLR analysis using OriginProÒ 2015 and MATLAB 2014 (Math-
Works, Inc.) software. NLR analysis was as described in Sec-
tion 3.3.3. Fig. 11a and b shows plots of qe against Ce for the
monocomponent adsorption of AO and ST on the CTA adsorbent,
with isotherm curves fit to experimental data.
Notably, the monocomponent adsorption isotherms of AO and
ST exhibited different shapes. The adsorption isotherms of AO
and ST at pH 4.5 can be classified as type IV isotherms because they
exhibited more than one adsorption step, indicating the inhomo-
geneity of the CTA adsorbent surface. It seems that most of the
adsorption sites on the CTA adsorbent surface are saturated at a
relatively low equilibrium concentration (limit concentration),
and interactions between adsorbed molecules or conformational
changes of adsorbed molecules on the CTA adsorbent surface
may occur, allowing the adsorption of more AO and ST molecules.
The Konda model described the adsorption steps of AO and ST on
CTA very well and provided good estimates of Qmax (Fig. 11 and
Table 3). The binding energy (b1) of ST with adsorption sites of
the CTA adsorbent was greater than AO for the first adsorption
step, while the same behavior was noticed for the binding energy
(b2) of the second adsorption step, indicating that ST was bound
more strongly than AO on CTA adsorption sites. The coordination
585
numbers (CN) [54] of the CTA adsorbent for AO and ST adsorption
at pH 4.5 were greater than unity (i.e. 2.40 and 1.85 for AO and ST,
respectively), indicating that more than one carboxylic group is
adsorbing to a dye molecule.
The adsorption isotherm of ST on the CTA adsorbent at pH 7.0
(Fig. 11b) was type I, which was well described by the Langmuir
and Sips models. The value of the parameter n (0.67) of the Sips
model does not allow reducing the Sips model to the Langmuir
model, although the values of isotherm parameters Qmax and b
for the Sips model were similar to those obtained for the Langmuir
model. The parameter n of the Sips model indicates that ST adsorp-
tion on the CTA adsorbent is heterogeneous, which is more consis-
tent for this adsorption system than the assumptions of the
Langmuir model. On the other hand, the adsorption isotherm of
AO at pH 7.0 (Fig. 11a) was type V or S-shaped isotherm. This type
of isotherm occurs when there is lateral interaction between
adsorbed molecules. As the Langmuir model does not assume
adsorbate-adsorbate interaction, the Hill-de Boer and Fowler-
Guggenheim models were used to model the adsorption isotherm
of AO on the CTA at pH 7.0. As seen in Table 3, values of the con-
stant c for both isotherm models were positive, suggesting that
there are attractive interactions between adsorbates on the CTA
surface. Both models fit the experimental adsorption data well.
However, comparing the experimental adsorption capacity
(Qmax,exp) with the estimated adsorption capacity (Qmax,est), the F-
G model better represented the adsorption process than the H-B
model. On the other hand, the H-B model resulted in smaller values
of v2red and RSS compared to the F-G model. Therefore, analysis of fit
quality for the F-G and H-B models does not allow determination of
whether mobile adsorption occurred on the CTA surface. In addition,
CN values for AO and ST adsorption on the CTA adsorbent were 1.25
and 1.58, respectively. These values indicate that CN decreased as the
solution pH increased, suggesting a change in the mechanism by
which AO and ST molecules are adsorbed on the CTA surface. That
is, fewer adsorption sites are needed for adsorption of AO and ST,
and therefore, the adsorption capacity at pH 7.0 increased.
3.3.4.2. Bicomponent adsorption. Fig. 11c shows a plot of qe against
Ce for the bicomponent system AO-ST (Supplementary Table 8)
with curves fit to the experimental data using the IAST and RAST
models supplied with monocomponent Sips isotherm parameters
Qmax, b, and n (Table 3) obtained by fitting both monocomponent
experimental equilibrium data to this model. Table 4 shows the
values of the error functions used in the minimization of the resid-
ual sum of squares (RSS) between the experimental initial liquid
phase solute concentrations and those predicted by the IAST and
RAST models, as well as the coefficient of determination (R2) and
reduced chi-square (v2red). As can be seen, the quality of the predic-
tion of bicomponent adsorption equilibrium data for ST using the
IAST and RAST models was very good (R2 > 0.98) while for AO it
was satisfactory (R2 > 0.81). Supplementary Fig. 7 shows graphs of
qe,est against qe,exp, which are used to analyze if the experimental
data can be predicted by the model. Both models described the bin-
ary system AO-ST well.
The data presented in Table 3 and Fig. 11a–c show that the
adsorption capacity of the CTA adsorbent for both AO and ST was
greater in the monocomponent systems than the bicomponent sys-
tem. This indicates the occurrence of antagonistic interactions
between dyes, which decreased their adsorption capacity in the
bicomponent system. The adsorption capacity of AO was more
greatly affected than that of ST.
The plot of gamma of the component i against gamma of the
component j was nonlinear (Supplementary Fig. 8), so the activity
coefficient of the component j was calculated using Eqs. (17)–(21)
reported by Erto et al. [53]. The adjustable coefficients a and b
obtained by the approximation suggested by Erto et al. [53] are
586 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590

Table 3
Results of modeling the monocomponent equilibrium adsorption data of auramine-O and safranin-T with isotherm models (25 °C, 130 rpm and 0.2 g L
1 CTA).

Monocomponent adsorption systems

Isotherm model Parameters pH 4.5 pH 7.0
Auramine-O Safranin-T Auramine-O Safranin-T
Experimental data Qmax.exp (mmol g
1) 2.628 ± 0.052 3.443 ± 0.059 5.172 ± 0.006 4.019 ± 0.147
te (min) 720 720 300 450
l (mol L
1) 0.0775 0.0774 – 0.1507
ce 0.7744a 0.7745a – 0.7594b
Langmuir Qmax.est (mmol g
1) 2.531 ± 0.078 3.262 ± 0.164 – 4.314 ± 0.120
b (L mmol
1) 17.80 ± 2.02 23.57 ± 4.65 34.31 ± 3.69
R2 0.9417 0.8180 0.9277
v2red 0.0097 0.0487 0.0114
RSS 0.1352 0.8772 0.1482
Sips Qmax.est (mmol g
1) – – 5.055 ± 0.107 4.074 ± 0.143
b (L mmol
1) 3.03 ± 0.04 32.64 ± 2.37
n 0.17 ± 0.01 0.67 ± 0.14
1/n 5.768 ± 0.258 1.499 ± 0.307
R2 0.9942 0.9432
v2red 0.0114 0.0101
RSS 0.2279 0.1114
Hill-de Boer KHB (L mmol
1) – – 0.318 ± 0.050 –
Qmax,est (mmol g
1) 7.386 ± 0.701
c (kJ mol
1) 6.223 ± 0.533
R2 0.9856
v2red 0.1246
RSS 1.1491
Fowler-Guggenheim KFG (L mmol
1) – – 0.484 ± 0.077 –
Qmax,est (mmol g
1) 5.443 ± 0.504
c (kJ mol
1) 3.616 ± 0.361
R2 0.9860
v2red 0.1620
RSS 1.2406
Konda Q1 (mmol g
1) 2.200 2.600 – –
Q2 (mmol g
1) 0.641 1.091
Qmax.est (mmol g
1) 2.841 3.691
b1 (L mmol
1) 26.575 47.9774
b2 (L mmol
1) 7.983 17.5653
c2 (mmol L
1) 0.482 0.320
R2 0.9883 0.9710
v2red 0.0261 0.1902
RSS 0.0453 0.3286
a
Calculated using Guntelberg approximation.
b
Calculated using Davies approximation.

presented in Table 5. The plots of the experimental activity coeffi-
cients against experimental reduced spreading pressures are
shown in Supplementary Fig. 9. This plot presents an increasing
trend for the activity coefficient for one component when the
spreading pressure is increased and an opposite behavior for
another component.
Energy dispersive X-ray (EDX) spectroscopy (SEM-EDX) was
used to map the distribution of the adsorbed AO and ST along
the CTA surface for the mono- and bi-component systems. Results
are discussed and presented in the Supplementary Material.
3.4. Thermodynamics of adsorption
3.4.1. Isothermal titration calorimetry
Isothermal titration calorimetry (ITC) is a sensitive technique
used to measure the enthalpy change associated with a specific
thermodynamic process occurring within a calorimetric cell when
the system composition is varied by titration. In the first part of
this series of studies, ITC was successfully used to obtain the
enthalpy change associated with the adsorption of metal ions on
the CTA adsorbent [10]. In the same way, ITC was used in the pre-
sent study to determine the adsorption enthalpy change (DadsH) of
AO and ST on the CTA as a function of the adsorbent surface
coverage. Fig. 12 shows DadsH values as a function of the amount
of the dyes adsorbed on CTA at equilibrium at pH 4.5 and 7.0.
DadsH values were all negative in the qe range evaluated,
indicating that adsorption phenomenon is exothermic. DadsH val-
ues varied from
18.83 ± 0.64 kJ mol
1 to
5.60 ± 0.02 kJ mol
1,
suggesting that physisorption governs the adsorption process.
For the adsorption of ST, the profile of the curves of DadsH
against qe as well as the magnitude of DadsH depended strongly
on the solution pH, confirming the large contribution of electro-
static interactions to dye adsorption. For this dye, as the qe value
increased, the DadsH values became less exothermic at pH 4.5
and more exothermic at pH 7.0. DadsH values at pH 4.5 were always
more negative than at pH 7.0. To explain these results, DadsH was
broken into three terms associated with three independent
subprocesses occurring simultaneously during dye adsorption, as
presented in Eq. (18).
Dads H ¼ Dads Hdes þ Dads Hdye
CTA þ Dads Hdye
dye ð18Þ
where DadsHdes is the enthalpy change associated with the desolva-
tion of both cationic dyes and adsorption sites on the CTA surface,
DadsHdye-CTA is the enthalpy change for the formation of the interac-
tion between the dye and CTA, and DadsHdye-dye is the enthalpy
change related to the dye-dye interaction on the adsorbent surface.
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590 587

Fig. 11. Adsorption isotherms of auramine-O (AO) and safranin-T (ST) on the CTA adsorbent in equimolar concentration (Ci = Cj) at 25 °C, 130 rpm, and 0.2 g L
1 CTA.
Monocomponent adsorption systems of (a) AO and (b) ST at pH 4.5 and 7.0. Bicomponent adsorption system of (c) AO-ST at pH 7.0.

Table 4
Results of modeling the equilibrium adsorption data with the IAST and RAST model for bicomponent system AO-ST (25 °C, 130 rpm, pH 7.0 and 0.2 g L
1 CTA with equimolar dye
concentrations (Ci = Cj)).

Adsorption system Dye Model RSS R2 v2red

AO-ST AO IAST 0.3368 0.8526 2.410  10
2
RAST 0.3981 0.8171 2.345  10
2
ST IAST 2.4278 0.9943 5.754  10
2
RAST 1.8076 0.9827 4.559  10
2

Table 5
Values of adjustable parameters of the Wilson activity coefficient model for bicomponent system AO-ST (i = 1 and j = 2) and of the model proposed by Erto et al. [58].

Adsorption system (i-j) c K12 K21 a1 b2 R2i R2j RSS

AO-ST 53.8366 0.2660 2.0280 – – 0.9975 0.8895 1.005
– – – 1.5508 1.7462 0.9230 –
1,2
Values obtained by regression analysis of Eq. (17) according to the approximation suggested by Erto et al. [53].

The dye-dye interaction in solution was disregarded because the pH and low qe values (i.e. the condition under which DadsH was
dye equilibrium concentrations in the calorimetric experiments obtained), the solvation shells of both dyes and adsorption sites
were very low (in the range from 10
7 to 10
5 mol L
1). At constant should remain unchanged, and the term DadsHdes is constant as
588 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
Fig. 12. Adsorption enthalpy change as a function of the amount adsorbed at
equilibrium for AO and ST on the CTA surface at 25.0 °C: (s) AO-pH 4.5, (h) ST-
pH 4.5, and (j) ST-pH 7.0.
the qe value increases. Thus, the change in DadsH when the surface
coverage increases is mainly attributed to the subprocesses associ-
ated with the formation of the dye-CTA and dye-dye interactions.
Adsorption of the dyes on the CTA surface is mainly controlled
by electrostatic interactions between the negatively charged car-
boxylate groups on the CTA surface and the positively charged
dye molecules. Consequently, the greater the force of the
dye-CTA electrostatic attraction, the greater the magnitude of
DadsHdye-CTA (always negative). ST has two positive charges at pH
4.5, while ST mostly has one positive charge at pH 7.0 (pKa of ST
is 6.4). Thus, the dye-CTA electrostatic attraction is more intense
at pH 4.5, making the term DadsHdye-CTA more negative and there-
fore DadsH more exothermic at this pH.
When the surface coverage increases at pH 4.5, the distance
between positively charged ST molecules on the CTA surface
decreases, increasing the dye-dye electrostatic repulsion on the
adsorbent surface. Thus, the term DadsHdye-dye becomes more pos-
itive and the values of DadsH increase as qe increases at this pH.
When the pH is changed from 4.5 to 7.0, dye-dye electrostatic
repulsion on the CTA surface decreases due to deprotonation of
ST, making the term DadsHdye-dye less positive at pH 7.0 than at
4.5. Nonetheless, an increase in DadsH as qe increases would also
be expected at pH 7.0. However, DadsH decreased when the surface
coverage increased, suggesting that different conformations of car-
boxylic ligands and/or adsorbed ST molecules on the CTA may have
occurred.
For adsorption of AO on CTA at pH 4.5, the values of DadsH were
less negative than those for ST at the same pH, which was expected
due to the higher positive net charge of ST (i.e. DadsHdye-CTA is more
negative for ST than for AO). Moreover, values of DadsH for AO
adsorption at pH 4.5 were almost independent of qe, suggesting
that the energies involved in the dye-dye interaction and the con-
formational changes on the CTA surface may compensate for each
other. Values of DadsH were not obtained for AO at pH 7.0 because
in the AO concentration range evaluated (limited by the dye solu-
bility), dye adsorption was very low (Fig. 12) and the signal mea-
sured in the calorimetry experiment was of the same magnitude
as equipment noise.
3.4.2. Thermodynamic parameters of adsorption
To complete the thermodynamic analyses of the adsorption
process of AO and ST on CTA, standard adsorption thermodynamic
Table 6
Thermodynamic parameters of monocomponent adsorption of auramine-O (AO) and
safranin-T (ST) on the CTA adsorbent at 25 °C (298 K).
Thermodynamic AO ST ST
parameters
pH 4.5 pH 7.0
DadsG° (kJ mol
1)
25.88 ± 0.00
27.35 ± 0.00
26.45 ± 0.00
DadsH° (kJ mol
1)
7.24 ± 0.09
19.10 ± 0.79
1.66 ± 0.39
TDadsS° (kJ mol
1) 18.65 ± 0.22 8.26 ± 0.34 24.79 ± 5.84
parameters (DadsG°, DadsH°, and DadsS°) were determined (Table 6).
Values of DadsG° were determined using Eq. (11) while values of
DadsH° were obtained by extrapolating the curve of DadsH against
Ce to Ce equals zero. The entropic contribution, TDadsS°, was deter-
mined from the fundamental relationship given by Eq. (19).
Dads G ¼ Dads H
T Dads S ð19Þ
Values of DadsG° were negative for all conditions evaluated,
showing that the dyes preferentially concentrate on the CTA surface
under standard conditions. Moreover, values of DadsH° and
TDadsS° were always negative and positive, respectively, indicating
that the adsorption process under standard conditions is enthalpi-
cally and entropically favored. The entropy increase is attributed
to the water molecules released from the solvation shell of both
dyes and CTA, mainly due to the formation of dye-CTA hydrophobic
interactions.
3.5. Desorption
3.5.1. Type of desorption agent
Two types of desorption solutions were used in the experiments,
HNO3 and HCl, at concentrations of 0.01 and 0.1 mol L
1. Table 7
shows that aqueous 0.1 mol L–1 HCl was the best desorption agent
for ST, while aqueous 0.01 mol L–1 HNO3 was the best desorption
agent for AO. However, the Edes of aqueous 0.1 mol L–1 HCl was
not much better than the Edes of aqueous 0.01 mol L–1 HCl for ST
desorption. Similarly, the Edes of aqueous 0.01 mol L–1 HNO3 was
not much better to the Edes of aqueous 0.01 mol L–1 HCl for AO
desorption. Therefore, for practical purposes, the same desorption
solution can be used for both dyes. Aqueous 0.01 mol L–1 HCl solu-
tion was chosen for the continuation of desorption experiments as a
function of time.
3.5.2. Effect of contact time on desorption
After determining the best desorption agent and solution con-
centration for desorption of AO and ST from CTA adsorbent, the
effect of contact time on Edes was evaluated. Table 7 shows that Edes
increased as desorption time increased from 1 to 3 h for both dyes.
However, there was no significant increase in Edes after 3 h. As
desorption implies an additional process cost, shorter times are
preferable. Therefore 3 h was chosen as the optimal desorption
time for both dyes. Incomplete desorption of AO and ST from the
CTA surface suggests that ion-exchange is not the sole mechanism
controlling the desorption of AO and ST, and therefore, these dyes
also form strong interactions with CTA that cannot be broken by
the desorption solutions tested.
3.6. Re-adsorption
Re-adsorption efficiency (Ere-ads) is a very important process
variable for an adsorbent material and will define if the adsorbent
material has the possibility of being used in various adsorption
cycles, and therefore, determine if its use is economic feasible.
Table 7 shows that Ere-ads for AO and ST were not 100% because
both AO and ST were not fully desorbed from the CTA adsorbent.
 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590 589

Table 7
Desorption efficiency (Edes) as a function of desorption agent, desorption solution concentration and time and re-adsorption efficiency (Ere-ads) for auramine-O (AO) and safranin-T
(ST).

Type of desorption agent HNO3 HCl

Dye Concentration (mol L
1) 0.01 0.1 0.01 0.1
Auramine-O Edes (%)a 27.34 ± 0.03 4.51 ± 0.08 21.57 ± 0.13 3.95 ± 0.03
Safranin-T 46.26 ± 0.10 32.25 ± 0.05 74.83 ± 0.16 78.35 ± 0.16
Effect of contact time on desorption efficiency
Dye Time (h) 1 3 6 12 24
Auramine-O Edes (%)b 38.23 ± 0.59 44.61 ± 0.44 44.69 ± 0.53 45.54 ± 0.41 47.54 ± 0.33
Safranin-T 65.58 ± 0.46 78.87 ± 0.49 81.45 ± 0.51 82.64 ± 0.65 81.12 ± 0.18
Re-adsorption efficiency
Dye QT,max (mmol g
1) Qe,re-ads (mmol g
1) Ere-ads (%)
Auramine-O 2.67 ± 0.01 2.08 ± 0.06 77.90 ± 0.15
Safranin-T 3.64 ± 0.01 2.21 ± 0.01 60.71 ± 0.05
a
Volume of desorption solution was 20.0 mL.
b
Volume of desorption solution was 100.0 mL.

However, Ere-ads for AO was higher than Ere-des for ST, suggesting
that the surface of the CTA adsorbent had a higher affinity for AO
molecules after desorption. In addition, Ere-ads for ST was smaller
than for AO, indicating a loss of adsorption capacity of ST by CTA
after desorption. Similar observations were noticed by Ferreira
et al. [29] for desorption and re-adsorption of crystal violet (CV)
from sugarcane bagasse modified with Meldrum’s acid. Therefore,
desorption and re-adsorption experiments demonstrated that the
CTA adsorbent can be reused, but rigorous control is necessary to
ensure that dye effluent concentrations are within the limits of dis-
charge due to the partial desorption of dyes and the unpredictable
re-adsorption behavior of the CTA adsorbent. The mass balance for
desorption and re-adsorption experiments is presented in Supple-
mentary Table 9.
4. Conclusions
Cellulose modified with trimellitic anhydride (CTA) was very
efficient to remove AO and ST from aqueous solutions, demonstrat-
ing excellent monocomponent adsorption capacities at both pH 4.5
and 7.0. In bicomponent systems, the presence of ST affected the
adsorption of AO more than the presence of AO affected the
adsorption of ST on the CTA adsorbent. Despite the complexity of
these systems, both the IAST and RAST models satisfactory fit the
bicomponent equilibrium data. Desorption of dyes from the CTA
adsorbent was incomplete. However, CTA adsorbed more dye in a
second adsorption cycle, showing the reliability of reuse of CTA
adsorbent. The use of isothermal titration calorimetry allowed
evaluation of how the magnitude of the enthalpy of adsorption
depended on surface coverage, showing that the adsorption sites
on the CTA surface are distinct and that dye-dye interactions affect
adsorption energies. The possibility of using the CTA adsorbent to
remove dyes from liquid phase at different pH values shows how
versatile this material is, which could reduce costs and improve
efficiency of an effluent treatment process.
Acknowledgements
The authors are grateful to Universidade Federal de Ouro Preto
(UFOP grant number 23109.003209/2016-98) and Fundação de
Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG grant
numbers APQ-01945-13 and APQ-01287-15) for funding this
research. The authors are also grateful to Coordenação de Aper-
feiçoamento de Pessoal de Nível Superior (CAPES). The authors
would like to acknowledge the Nanolab Electronic Microscopy Lab-
oratory, at the Redemat, Escola de Minas, UFOP, MG, Brazil, for pro-
viding the equipment and technical support for experiments
involving electron microscopy and SEM-EDX analyses. This work
is a collaboration research project of members of the Rede Mineira
de Química (RQ-MG) supported by FAPEMIG (Project: CEX-RED-
00010-14).
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jcis.2017.10.083.
References
[1] V.K. Gupta, R. Kumar, A. Nayak, T.A. Saleh, M.A. Barakat, Adsorptive removal of
dyes from aqueous solution onto carbon nanotubes: a review, Adv. Colloid
Interface Sci. 193–194 (2013) 24–34.
[2] V.K. Gupta, R. Jain, S. Varshney, V.K. Saini, Removal of Reactofix Navy Blue 2
GFN from aqueous solutions using adsorption techniques, J. Colloid Interface
Sci. 307 (2007) 326–332.
[3] M.T. Yagub, T.K. Sen, S. Afroze, H.M. Ang, Dye and its removal from aqueous
solution by adsorption: a review, Adv. Colloid Interface Sci. 209 (2014) 172–
184.
[4] M. Vakili, M. Rafatullah, B. Salamatinia, A.Z. Abdullah, M.H. Ibrahim, K.B. Tan, Z.
Gholami, P. Amouzgar, Application of chitosan and its derivatives as
adsorbents for dye removal from water and wastewater: a review,
Carbohydr. Polym. 113 (2014) 115–130.
[5] V.K. Gupta, Suhas, Application of low-cost adsorbents for dye removal – a
review, J. Environ. Manage. 90 (2009) 2313–2342.
[6] K. Venkataraman, The Chemistry Of Synthetic Dyes, Academic Press, New York,
1971.
[7] R.W. Sabnis, Handbook of Biological Dyes and Stains: Synthesis and Industrial
Applications, John Wiley & Sons, New Jersey, 2010.
[8] N. Terinte, B.M.K. Manda, J. Taylor, K.C. Schuster, M.K. Patel, Environmental
assessment of coloured fabrics and opportunities for value creation: spin-
dyeing versus conventional dyeing of modal fabrics, J. Clean. Prod. 72 (2014)
127–138.
[9] M.A.M. Salleh, D.K. Mahmoud, W.A.W.A. Karim, A. Idris, Cationic and anionic
dye adsorption by agricultural solid wastes: a comprehensive review,
Desalination 280 (2011) 1–13.
[10] F.S. Teodoro, S.N.D.C. Ramos, M.M.C. Elias, A.B. Mageste, G.M.D. Ferreira, L.H.M.
da Silva, L.F. Gil, L.V.A. Gurgel, Synthesis and application of a new carboxylated
cellulose derivative. Part I: removal of Co2+, Cu2+ and Ni2+ from
monocomponent spiked aqueous solution, J. Colloid Interface Sci. 483 (2016)
185–200.
[11] F.S. Teodoro, O.F.H. Adarme, L.F. Gil, L.V.A. Gurgel, Synthesis and application of
a new carboxylated cellulose derivative. Part II: removal of Co2+, Cu2+ and Ni2+
from bicomponent spiked aqueous solution, J. Colloid Interface Sci. 487 (2017)
266–280.
[12] K.A.G. Gusmão, L.V.A. Gurgel, T.M.S. Melo, L.F. Gil, Adsorption studies of
methylene blue and gentian violet on sugarcane bagasse modified with EDTA
dianhydride (EDTAD) in aqueous solutions: kinetic and equilibrium aspects, J.
Environ. Manage. 118 (2013) 135–143.
[13] O.S. Chan, W.H. Cheung, G. McKay, Single and multicomponent acid dye
adsorption equilibrium studies on tyre demineralised activated carbon, Chem.
Eng. J. 191 (2012) 162–170.
590 F.S. Teodoro et al. / Journal of Colloid and Interface Science 512 (2018) 575–590
[14] Y. Liu, Is the free energy change of adsorption correctly calculated?, J Chem.
Eng. Data 54 (2009) 1981–1985.
[15] E. Guntelberg, Untersuchungen uber ioneninteraction, Z. Physik. Chem. 123
(1926) 199–247.
[16] C.W. Davies, Ion Association, Butterworths, London, UK, 1962, pp. 37–53.
[17] P.S. Ghosal, A.K. Gupta, Determination of thermodynamic parameters from
Langmuir isotherm constant-revisited, J. Mol. Liq. 225 (2017) 137–146.
[18] F. Neese, The ORCA program system, Wiley Interdisciplinary Rev.: Comput.
Mol. Sci. 2 (2012) 73–78.
[19] P. Hohenberg, W. Kohn, Inhomogeneous electron gas, Phys. Rev. 136 (1964)
B864–B871.
[20] W. Kohn, L.J. Sham, Self-consistent equations including exchange and
correlation effects, Phys. Rev. 140 (1965) A1133–A1138.
[21] A.D. Becke, Density-functional thermochemistry. III. The role of exact
exchange, J. Chem. Phys. 98 (1993) 5648–5652.
[22] C. Lee, W. Yang, R.G. Parr, Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density, Phys. Rev. B 37 (1988) 785–
789.
[23] A.J. Cohen, P. Mori-Sánchez, W. Yang, Challenges for density functional theory,
Chem. Rev. 112 (2012) 289–320.
[24] L.C. Cardoso, R.M.L. Savedra, M.M. Silva, G.R. Ferreira, R.F. Bianchi, M.F.
Siqueira, Effect of blue light on the electronic and structural properties of
Bilirubin isomers: insights into the photoisomerization and photooxidation
processes, J. Phys. Chem. A 119 (2015) 9037–9042.
[25] J.F. Gomes, K. Bergamaski, M.F.S. Pinto, P.B. Miranda, Reaction intermediates of
ethanol electro-oxidation on platinum investigated by SFG spectroscopy, J
Catal 302 (2013) 67–82.
[26] R.M.L. Savedra, M.F.S. Pinto, M. Trsic, E. T., K. Y., Quantum chemical study of
electronic and structural properties of retinal and some aromatic analogs, J.
Chem. Phys. 125 (2006) 144901.
[27] R. Ditchfield, W.J. Hehre, J.A. Pople, Self-consistent molecular-orbital methods.
IX. An extended gaussian-type basis for molecular-orbital studies of organic
molecules, J. Chem. Phys. 54 (1971) 724–728.
[28] E.F. Pettersen, T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C.
Meng, T.E. Ferrin, UCSF Chimera—a visualization system for exploratory
research and analysis, J. Comput. Chem. 25 (2004) 1605–1612.
[29] B.C.S. Ferreira, F.S. Teodoro, A.B. Mageste, L.F. Gil, R.P. de Freitas, L.V.A. Gurgel,
Application of a new carboxylate-functionalized sugarcane bagasse for
adsorptive removal of crystal violet from aqueous solution: kinetic,
equilibrium and thermodynamic studies, Ind. Crop. Prod. 65C (2015) 521–534.
[30] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process
Biochem. 34 (1999) 451–465.
[31] S.Y. Lagergren, Zur theorie der sogenannten adsorption gelöster stoffe,
kungliga svenska vetenskapsakademiens, Handlingar 24 (1898) 1–39.
[32] Y. Ho, G. McKay, Kinetic model for lead (II) sorption onto peat, Adsorpt. Sci.
Technol. 16 (1998) 243–255.
[33] G.E. Boyd, A.W. Adamson, L.S. Myers, The exchange adsorption of ions from
aqueous solutions by organic zeolites. II. Kinetics, J. Am. Chem. Soc. 69 (1947)
2836–2848.
[34] W.J. Weber, J.C. Morris, Kinetics of adsorption on carbon from solution, J. San.
Eng. Div. 89 (1963) 31–60.
[35] J.W. Corsel, G.M. Willems, J.M.M. Kop, P.A. Cuypers, W.T. Hermens, The role of
intrinsic binding rate and transport rate in the adsorption of prothrombin,
albumin, and fibrinogen to phospholipid bilayers, J. Colloid Interface Sci. 111
(1986) 544–554.
[36] B.H. Hameed, M.I. El-Khaiary, Malachite green adsorption by rattan sawdust:
isotherm, kinetic and mechanism modeling, J. Hazard. Mater. 159 (2008) 574–
579.
[37] X.-Y. Yang, B. Al-Duri, Application of branched pore diffusion model in the
adsorption of reactive dyes on activated carbon, Chem. Eng. J. 83 (2001) 15–23.
[38] V. Hlady, Adsorption kinetics of low density lipoprotein onto a hydrophobic-
hydrophilic gradient surface, in: G. Pifat-Mrzljak (Ed.), Supramolecular
Structure and Function 7, Springer, Rovinj, Croatia, 2001, pp. 45–61.
[39] K.Y. Foo, B.H. Hameed, Insights into the modeling of adsorption isotherm
systems, Chem. Eng. J. 156 (2010) 2–10.
[40] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and
platinum, J. Am. Chem. Soc. 40 (1918) 1361–1403.
[41] R. Sips, On the structure of a catalyst surface, J. Chem. Phys. 16 (1948) 490–
495.
[42] R.H. Fowler, E.A. Guggenheim, Statistical Thermodynamics, Cambridge
University Press, London, 1939.
[43] T.L. Hill, Statistical mechanics of multimolecular adsorption II. Localized and
mobile adsorption and absorption, J. Chem. Phys. 14 (1946) 441–453.
[44] J.H. de Boer, The Dynamical Character of Adsorption, Oxford University Press,
Oxford, 1953.
[45] L.N. Konda, I. Czinkota, G. Füleky, G. Morovján, Modeling of single-step and
multistep adsorption isotherms of organic pesticides on soil, J. Agr. Food Chem.
50 (2002) 7326–7331.
[46] D. Do, Adsorption analysis: equilibria and kinetics, Imperial College Press,
London, 1998.
[47] A.L. Myers, J.M. Prausnitz, Thermodynamics of mixed-gas adsorption, AIChE J.
11 (1965) 121–127.
[48] C.J. Radke, J.M. Prausnitz, Thermodynamics of multi-solute adsorption from
dilute liquid solutions, AIChE J. 18 (1972) 761–768.
[49] E. Costa, J.L. Sotelo, G. Calleja, C. Marrón, Adsorption of binary and ternary
hydrocarbon gas mixtures on activated carbon: experimental determination
and theoretical prediction of the ternary equilibrium data, AIChE J. 27 (1981)
5–12.
[50] A.L. Myers, Activity coefficients of mixtures adsorbed on heterogeneous
surfaces, AIChE J. 29 (1983) 691–693.
[51] O. Talu, I. Zwiebel, Multicomponent adsorption equilibria of nonideal
mixtures, AIChE J. 32 (1986) 1263–1276.
[52] A. Erto, A. Lancia, D. Musmarra, A modelling analysis of PCE/TCE mixture
adsorption based on Ideal Adsorbed Solution Theory, Sep. Purif. Technol. 80
(2011) 140–147.
[53] A. Erto, A. Lancia, D. Musmarra, A Real Adsorbed Solution Theory model for
competitive multicomponent liquid adsorption onto granular activated
carbon, Micropor. Mesopor. Mat. 154 (2012) 45–50.
[54] L.V.A. Gurgel, R.P. de Freitas, L.F. Gil, Adsorption of Cu(II), Cd(II), and Pb(II) from
aqueous single metal solutions by sugarcane bagasse and mercerized
sugarcane bagasse chemically modified with succinic anhydride, Carbohydr.
Polym. 74 (2008) 922–929.