You are on page 1of 16

Journal of Hazardous Materials 411 (2021) 125166

Contents lists available at ScienceDirect

Journal of Hazardous Materials


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

Research paper

Optimization of Sibipiruna activated carbon preparation by


simplex-centroid mixture design for simultaneous adsorption of rhodamine
B and metformin
Lucas Spessato a, *, Vitor A. Duarte b, Patrícia Viero a, Heloisa Zanella a, Jhessica M. Fonseca a,
Pedro A. Arroyo b, Vitor C. Almeida a, *
a
Department of Chemistry, State University of Maringá, Av. Colombo 5790, Maringá, Paraná, Brazil
b
Department of Chemical Engineering, State University of Maringá, Av. Colombo 5790, Maringá, Paraná, Brazil

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

Editor: Dr. H. Artuto The present paper reports the application of augmented simplex-centroid mixture design to obtain a high BET
surface area activated carbon using as reactants KOH, K2CO3 and K2C2O4. The optimum mixture composition was
Keywords: 2.51 g of KOH, 0.49 g of K2CO3 and absence of K2C2O4, generating an optimized AC (ACop) with SBET value
Sibipiruna pods equals to 1984 m2 g− 1. The results herein obtained show that low amounts of K2CO3 can catalyze the pore
Response surface methodology
development in the presence of KOH, increasing the surface area. Furthermore, the fractal dimensions of ACop are
Binary adsorption
greater than 2.72, indicating the material has a complex pore structure with irregularities self-similar upon
Adsorption mechanism
Adsorption island variations of resolution, as seen by SEM images. The TPD curves showed that the ACop has different oxygenated
molecular fragments, which agrees with the pHPZC value (5.05). The ACop was applied in the adsorption of
rhodamine B (RhB) and metformin (Met) in both binary and monocomponent systems. The simultaneous
adsorption at 30 ◦ C reveals that the adsorption capacity of RhB is 630.94 mg g− 1, while for Met the value is
103.83 mg g− 1.

1. Introduction detected in water bodies, causing endocrine disorders on aquatic eco­


systems (Niemuth and Klaper, 2015; Burns et al., 2018; Oosterhuis et al.,
Contamination of the environment with synthetic chemical com­ 2013). RhB is a highly water-soluble dye (15.0 g L–1 at 20 ◦ C) classified
pounds has been increasing exponentially after the industrialization of as a xanthene compound. It is mainly used as tracer to determine the rate
the society. There is a countless number of chemicals produced and used and direction of flow within water bodies (Singh et al., 2018).
by humankind which are found in soils, air, and waters. Recently, re­ Furthermore, RhB is widely used for paper printing, textile and leather
searchers have been paid attention to chemicals classified as emerging dyeing, besides as dye in food industries (Chen and Zhu, 2016; Huang
contaminants, which mostly appear in aquatic bodies and are not et al., 2020). The presence of RhB in environmental and food matrices
included in regulations of environmental institutions (e.g., Environ­ can cause serious carcinogenic, genotoxic, and neurotoxic effects on
mental Protection Agency). Emerging pollutants can be classified as public health, in addition to being irritating to the skin, eyes, and to the
pharmaceuticals, personal care products, flame retardants, dyes, and respiratory system (Sharma et al., 2017; Ji et al., 2019; Cheng and Tsai,
pesticides. 2017).
Among the studied emerging contaminants, it must be highlighted A considerable amount of works has been published in the literature
metformin (Met) and rhodamine B (RhB). Met is a highly water-soluble on advanced techniques to treat wastewater composed by Met or RhB.
antidiabetic drug (1.38 g L–1 at 20 ◦ C) and is within the list of essential For instance, photodegradation (Nezar and Laoufi, 2018; Adhikari and
medicines of the World Health Organization (WHO). According to WHO, Kim, 2020; Leroy et al., 2020), Fenton-like processes (Neamţu et al.,
about 422 million people worldwide have diabetes, and most of them 2014; Zhou et al., 2019), cavitation-based processes (Komarov et al.,
use Met to control the glycemic index. Met is one of the most consumed 2020), precipitation with ionic liquids (Kumar et al., 2020) and
drugs in the world and, due to the low metabolization rate, has been biodegradation (Poursat et al., 2019; Elizalde-Velázquez and

* Corresponding authors.
E-mail addresses: lucas.spessato@hotmail.com (L. Spessato), vcalmeida@uem.br (V.C. Almeida).

https://doi.org/10.1016/j.jhazmat.2021.125166
Received 21 October 2020; Received in revised form 31 December 2020; Accepted 14 January 2021
Available online 16 January 2021
0304-3894/© 2021 Elsevier B.V. All rights reserved.
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Gómez-Oliván, 2020). Such techniques have intrinsic disadvantages, 1.0 0.0


based on high energy demands, high environmental factors, complex 1
technical set-up and/or time-consuming, seldom becoming commer­ 0.9 0.1
cially attractive technologies. In front of these pitfalls, adsorption-based
0.8 0.2
processes using activated carbons (ACs) are considered adequate alter­
natives, once they tend to be environmentally friendly, low-energy 0.7 0.3
consuming and highly efficient. 4 10 9
Most ACs are produced from chemical activation, which is reported 0.6 0.4
to develop the highest values of specific surface areas. Among reagents
0.5 0.5
proposed for chemical activation, the most used are KOH, H3PO4,
K2C2O4, K2CO3 and ZnCl2 (Sevilla et al., 2017; Yagmur et al., 2020; 0.4 0.6
Wang et al., 2020; Oginni et al., 2019). Not only the chemical nature of 8
the activating agent is important for the pore development, but also the 0.3 5 13 0.7
operational parameters (e.g., temperature, heating rate, time of pyrol­
0.2 0.8
ysis and flow of an inert gas). In this sense, there is a great number of
11 12
studies focused on the optimization of the operational parameters using 0.1 0.9
one type of activating agent, and the most published papers are those
which make use of central composite designs based on response surface 0.0
2 6 7 3 1.0
methodologies (Yokoyama et al., 2019; Bedin et al., 2018; Yu et al., 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
2019; Sulaiman et al., 2018; Liang et al., 2020; Garba and Rahim, 2014). K2CO3
However, to the best of the authors knowledge, there is no reported
work in the literature based on the application of mixture designs to Fig. 1. Schematic demonstration for the augmented simplex centroid mixture
improve the surface area of ACs by combination of different chemicals. design in mass fraction.
From the mixture designs, it is possible to investigate if there exist
mixtures of two or more ingredients which produce more desirable
product properties or process improvement than are obtainable with the Table 1
single ingredients (Cornell, 2011). A few standard models can be used Simplex-centroid mixture design and response values.
when constraints are imposed on the mixture components, for example, Mixture Components (g) Response (m2 g− 1)
simplex-lattice (Scheffé, 1958) and simplex-centroid (Scheffé, 1963) KOH (X1) K2CO3 (X2) K2C2O4 (X3) SBET (Y1)
designs, which are powerful mathematical and statistical tools to eval­
AC1 3.0 0 0 2056
uate the influence of the components proportion over the response and
AC2 0 3.0 0 1057
to find out the optimum composition of the mixture. AC3 0 0 3.0 967
It must be emphasized that previous works of adsorption have been AC4 2.0 1.0 0 1678
limited to the preparation of ACs to application in adsorption systems of AC5 1.0 2.0 0 1079
AC6 0 2.0 1.0 979
a single component. Such studies often do not represent reality, once
AC7 0 1.0 2.0 1193
wastewaters are complex solutions in which competitive mechanisms AC8 1.0 0 2.0 1022
can govern the adsorption phenomenon. Nevertheless, competitive AC9 2.0 0 1.0 1260
adsorption of pollutants mixtures are barely presented in the literature. AC10 2.0 0.5 0.5 1493
Due to the scarcity of simultaneous adsorptions published in the litera­ AC11 0.5 2.0 0.5 1014
AC12 0.5 0.5 2.0 1276
ture, the present paper presents adsorptions of binary and mono­
AC13 1.0 1.0 1.0 1190
component solutions to investigate possible differences in the
adsorption mechanism.
In front of the aforementioned issues, the aim of the present paper is to a stainless-steel reactor, which was submitted to pyrolysis in a muffle
to optimize the preparation of an AC from Sibipiruna (Caesalpinia pel­ furnace (EDG-S equipment, EDG3P 7000) at 500 ◦ C for 2.0 h under N2
tophoroides) pods using an augmented simplex-centroid mixture design flow of 100 cm3 min–1 and heating rate of 20 ◦ C min–1. After cooling
and KOH, K2CO3 and K2C2O4 as activating agents. The efficiency of the down to room temperature, the material was stored and labeled pyro­
optimized AC (ACop) was investigated in the simultaneous removal of lyzed material (PM). Afterwards, the activating procedure was carried
Met and RhB, and the competitive adsorption mechanism was evaluated out from the impregnation ratio 3:1 (activating agent:PM, wt:wt). Thus,
by kinetic and equilibrium studies of adsorption and DFT-based quan­ 1.0 g of PM, 10 mL of deionized water and 3.0 g of a mixture of chemical
tum chemical descriptors. activation agents (KOH, K2CO3 and K2C2O4) were all transferred to a
vertical stainless-steel reactor, magnetically stirred for 2.0 h and dried in
2. Experimental section an oven at 150 ◦ C for 4.0 h. The chemical nature and amounts of each
activation agent were followed according to the augmented simplex-
2.1. Precursor material centroid mixture design (ASCMD), see next section. The reactor con­
taining the dried mixture was heated to 750 ◦ C in a muffle furnace for
Sibipiruna pods were collected over the campus of State University of 2.0 h, under N2 flow of 100 cm3 min–1 and heating rate of 20 ◦ C min–1.
Maringá and carried to the laboratory, where the pods were properly After cooling, the obtained materials were rinsed thrice with HCl 0.1
washed with deionized water and dried at 110 ◦ C for 24 h. The dried mol L–1 and several times with distilled water, until pH ≅ 6.0. After
precursor was milled in a knife mill, the particles between 425 and 250 purification, the materials were dried at 100 ◦ C for 24.0 h, stored and
µm were granulometrically separated, stored and labeled in natura. named ACZ, where the subscript “Z” is referred to the run of the ASCMD.

2.2. Preparation of the activated carbon (AC) 2.3. Augmented simplex-centroid mixture design (ASCMD)

The AC was prepared according to a two-step process (carbonization In the present study, a three-component augmented simplex-centroid
followed by chemical activation), commonly used in the author’s mixture design was proposed (Design-Expert® 7.0, Stat-Ease Inc., Min­
research group. Firstly, 20.0 g of the in natura material was transferred neapolis, MN) to get different proportions among the chemicals KOH

2
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

(X1), K2CO3 (X2) and K2C2O4 (X3), in order to investigate the influence measured by the pH at point of zero charges (pHPZC) according to Prahas
upon the response BET surface area (SBET)(Y1, m2 g–1) of the prepared et al. (2008). The morphology of the ACop was evaluated by scanning
ACs (Fig. 1). The mixture of chemicals were estimated considering its electronic microscopy (SEM) using a FEI Company (Quanta 250)
sum equal to 3.0 g, since high impregnation ratios (activating agent: equipment.
pyrolized material, wt:wt) tend to promote greater pore development,
and consequently high SBET values (Bedin et al., 2016; Spessato et al.,
2019; Pezoti et al., 2016; Martins et al., 2015). In this way, the sum of 2.5. Batch adsorption studies
the three different components must equals to 3.0 g in the form of X1 +
X2 + X3 = 3.0, or equivalently to X1 + X2 + X3 = 100% in mass fraction, The adsorption experiments were performed using mixtures of RhB
see Fig. 1 and Table 1. The overall equation for a three-component (C28H31ClN2O3, ≥ 98%, Sigma–Aldrich) and Met (C4H11N5, ≥ 97%,
system is represented by Eq. (1). Sigma–Aldrich) with 1:1 weight ratio. In a typical adsorption experi­
ment using a 50 mL polypropylene tube, 25 mg of the optimized AC was


3 3 ∑
∑ 3 3 ∑
∑ 3
( ) ∑3 ∑
3 ∑
3
Yi = bi Xi + bij Xi Xj + bij Xi Xj Xi − Xj + bijk Xi Xj Xk (1)
i=1 i≤j j i≤j j i≤j j≤k k

where Yi represents the predicted response, Xi, Xj and Xk stands for the placed in contact with 25 mL of the binary solution with desired values
components of the mixture and bi, bij, bijk are the coefficients of the of initial concentration of the adsorbates, initial pH and stirring time (at
models for the first, second and third order system, respectively. Fig. 1 fixed stirring speed of 200 rpm). For the pH influence study over the
and Table 1 show 13 different mixtures for the augmented adsorption capacity, an initial concentration of 500 mg L–1 from each
simplex-centroid design. The experiments were carried out in duplicate, adsorbate was set, under stirring time of 360 min, whilst the pH was
and only the average values of the response SBET were presented in this ranged from 3.0 to 13.0 (∆pH = 1.0); these solutions were prepared with
paper. 0.1 mol L–1 HCl and NaOH, the solution with pH values 12.0 and 13.0
were prepared from the standard solutions of NaOH. The pH value
which presented the highest value of adsorption capacity was set for the
2.4. Characterizations equilibrium and kinetic studies. The adsorption equilibrium study was
performed with initial concentrations ranged from 25 to 900 mg L–1 of
The textural properties were obtained by N2 physisorption at 77 K each component in the binary mixture, and the kinetic study was per­
using a QuantaChrome (Nova 1200e) surface area and pore size formed from 1.0 to 360 min with three different initial concentrations
analyzer. The specific surface area (SBET) was determined according to (300, 500 and 700 mg L–1). At the end of each adsorption experiment,
Brunauer-Emmett-Teller (BET) equation in the range of relative pressure the solutions were filtered using Millipore membranes (0.45 µm). The
(P/P0) from 0.07 to 0.16. The heat of adsorption for the formation of the equilibrium concentration values of RhB and Met were determined by
first N2 monolayer was calculated using the C constant of the BET derivative spectroscopy. The adsorption capacities were calculated ac­
equation (C = e(− ΔHads − ΔHvap )/RT) and assuming N2 ∆H◦ vap of cording to Eq. (3).
◦ ◦

5.56 kJ mol . The total pore volume (Vt) is the maximum amount of N2
–1 ( )
C0 − Ce,t V
adsorbed at P/P0 of 0.99, whereas the micropore volume (Vµ) was Qm = qe = qt = (3)
W
calculated using the t-plot and DeBoer thickness methods. The average
pore size (APS) and the pore size distribution were obtained using the where Qm (mg g–1) is the maximum adsorption capacity at a desired pH
ratio 4SBET/Vt and the non-local density functional theory (NLDFT), (fixed all other operational parameters), qe (mg g–1) is the amount
respectively. At last, the fractal dimension was determined by applying adsorbed at equilibrium, qt (mg g–1) is the amount adsorbed at time t, Ct
the linearized Frenkel-Halsey-Hill (FHH) equation on the adsorption and Ce are remaining concentrations (mg L–1) at time t and after equi­
branch of the physisorption results, as follows: librium, respectively, V (mL) is the used solution volume and W (mg) is
( )
P0 the amount of dried AC used. The adsorption experiments were per­
lnVads = lnK + hln2 (2) formed in triplicate and only the average values are presented in this
P
paper.
in which K is the FHH constant and h is the slope. In this way, a plot of ln To further understand the adsorption competitive mechanism,
Vads vs ln2(P0/P) is a straight line, the slope h is used to calculate the equilibrium isotherms were obtained using monocomponent solutions
fractal dimension (D). For a fractal surface, h is a function of surface of RhB and Met, under the same experimental conditions mentioned
coverage. For low surface coverage, where only attraction of adsorbent above.
and adsorbate is dominant, h = (D – 3)/3, whereas for high surface
coverage, where lateral interactions are dominant, h = D – 3 (Khalili 2.6. Derivative spectroscopy
et al., 2000). D values equal to 2.0 are attributed to perfectly smooth
surfaces, whereas D values equal to 3.0 are characteristic of highly rough The determination of the adsorbates in solution for the two-
and irregular surfaces (László et al., 1998). component system was carried out by derivative spectroscopy (DS)
The oxygenated surface groups of the ACop was evaluated by using an Agilent Cary 60 UV–Vis spectrophotometer and the mathe­
temperature-programmed desorption (TPD) using a Micromeritics matical software OriginPro, version 2018. DS is based on the use of first
(Autochem II 2920) equipment. In a typical procedure, 30 mg of the or higher-order derivatives of absorbance in relation to the wavelength
dried solid was submitted to a heat pretreatment at 300 ◦ C for 3.0 h with for qualitative or quantitative purposes; the overall equation for DS can
a heating rate of 10 ◦ C min–1 and He (99.999%) flow of 50 mL min–1. be expressed as:
After the system cooled to 100 ◦ C, the desorption of CO2 and CO was dn A dn ϵ
measured from 50 to 1000 ◦ C in heating rates from 5 to 20 K min–1 and = bC (4)
dλn dλn
He flow of 50 mL min–1. The buffering capacity of the ACop was

3
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 2. Optimized chemical structures and molecular eletronic potential surfaces for metformin and rhodamine B.

where A is the absorbance, λ is the wavelength, ϵ is the extinction co­ Qm Ks Ce K


efficient, b is the sample pathway, C is the sample concentration and n is q=( )( CS ) ;c = S (8)
Kl
the derivative order (e.g., for first-order derivative, n = 1; for second- Ce
1 − K l CS Ce Ce
1 − Kl CS + Ks CS
order derivative, n = 2; and so on). The mostly frequently uses of DS
are background elimination, matrix suppression and discrimination of Where qe (mg g–1) is the amount adsorbed at equilibrium, Ce (mg L–1) is
multicomponent in a mixture. the remnant concentration at equilibrium, Qm (mg g–1) is the maximum
Fig. S1a shows the UV-Vis molecular absorption spectra of RhB, Met adsorption capacity, KL is the Langmuir constant, KF and nF are
and the mixture (RhB-Met). As there are no interactions between RhB Freundlich constants, KDR is the Dubinin-Radushkevich constant, KS is
and Met that alters the ϵ, the absorbances are additive and the quanti­ the adsorption equilibrium constant for the first layer, Kl is the
fication at λmax of the individual components in a mixture is over­ adsorption equilibrium constant for upper layers, ε is the Polanyi po­
estimated. To overcome this issue, the absorption spectra can be derived tential, R is the universal gas constant (8.314 J K–1 mol–1), T (K) is the
to find out the zero crossing interferent points, allowing the quantifi­ absolute temperature, E (J mol–1) is the free energy of adsorption and CS
cation of each component in the binary system. In the present work, the (mg L–1) is the solubility of the adsorbate at the temperature of the
discrimination of RhB from Met was possible to be done from the first- system.
derivative spectra at the zero-crossing points (e.g., Met can be quanti­ Regarding binary equilibrium isotherms, competitive and non-
fied at 243 nm and RhB at 543 nm in the binary system), as can be seen dissociate models of non-modified Langmuir (Eq. (9)), modified Lang­
in Fig. S1b. It is important to highlight that the instrumental parameters muir (Eq. (10)), extended Freundlich (Eq. (11)) and Sheindorf-Rebuhn-
used to obtain the zero-order spectra were ∆λ = 0.5 nm, scanning speed Sheintuch (SRS) (Eq. (12)) were fitted to the experimental data.
of 300 nm min–1 and spectral range from 200 to 700 nm. Furthermore,
the Savitzky–Golay smooth technique (25 points) was used in the first- K iL Ce,i Qm,i
qe,i = (9)
derivative data points to improve the signal-noise ratio. ∑n
1 + K iL Ce,i
To obtain the calibration equations, a standard solution of 0.1 g L–1 i=1

Met and 0.1 g L–1 RhB were prepared by dissolving 0.01 g of the solids in ( / )
0.1 L of deionized water. The working solutions of the mixture were K iL Ce,i ηL,i Qm,i
qe,i = ∑n i ( / ) (10)
prepared by dilution to final concentrations from 1.0 to 8.0 mg L–1 (∆ = 1 + i=1 K L Ce,i ηL,i
1.0 mg L–1), a similar procedure was carried out for the monocomponent
( )
systems. Table S1 presents the calibration equations for the binary and 1
monocomponent solutions. KFi Ce,i
ni +xi

qe,i = ( )x ( )z (11)
Ce,i i + y2 Ce,i i
2.7. Fits of the equilibrium and kinetic models
[( )](1/ni )− 1

N
The monocomponent equilibrium adsorption isotherms were fitted qe,i = KF,i Ce,i aij Ce,i (12)
by the non-competitive and non-dissociative equations of Langmuir j=1
(Langmuir, 1916), Freundlich (Freundlich, 1907),
Dubinin-Radushkevich (Dubinin, 1960), and Brunauer-Emmett-Teller Where Qm,i (mg g–1) and KiL (L mg–1) are monolayer adsorption capacity
(BET) (Ebadi et al., 2009), which are presented below, respectively. and Langmuir constant for component i obtained from the respective
Qm KL Ce monocomponent model, KiF and ni are the Freundlich constants obtained
qe = (5) from the respective monocomponent model. The constant ηL,i is the
1 + KL Ce
Langmuir interaction factor for component i in the mixture, whereas xi,
qe = KF Ce1/nF (6) yi and zi are the Freundlich constants for component i during the
simultaneous adsorption. Also, aij is the coefficient of competition,
( )
1 1 frequently used to demonstrate how component j inhibits the adsorption
qe = Qm e− KDR ε2
; ε = RTln 1 + ; E = √̅̅̅̅̅̅̅̅̅̅̅ (7)
Ce 2KDR of component i.
The parameters of the monocomponent models were obtained using

4
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

the Origin software, whereas the parameters of the binary models were (Domingo et al., 2002; Ebenso et al., 2010). However, they have been
determined using the Solver function of non-linear regression in particularly useful to get new insights about adsorption phenomena
Microsoft Excel software, with the aim to minimize the error equations (Danaee et al., 2013; Pal and Chandrakumar, 2000; Baei and Moradi,
in non-linear regression analysis. Regarding the kinetic experimental 2012; de Souza et al., 2018). The descriptors mentioned above can be
data, the models of pseudo-first order (Eq. (13)), pseudo-second order calculated using the energy of HOMO (EH) and LUMO (EL) orbitals,
(Eq. (14)), Elovich (Eq. (15)) and Avrami (Eq. (16)) were fitted (Tan and obtained from DFT studies, the equations are presented below.
Hameed, 2017).
I = − EH (19a)
[ ]
qt = qe 1 − e− k1 t ; h0 = k1 qe (13)
A = − EL (19b)
k2 q2e t
qt = ; h0 = k2 q2e (14) Eg = |EH − EL | (19c)
1 + k2 qe t
( )
∂E I+A
1
qt = (lnαβ + lnt) (15) µ= ̃
̃− (19d)
β ∂N 2 v(r) 2

( ) ( )
qt = qe − qe e− kAv tn
(16) 1 ∂μ 1 ∂2 E I− A
η= = ̃
̃ (19e)
2 ∂N v(r) 2 ∂N 2 v(r) 2
Where qt is the amount adsorbed at time t, k1 and k2 are rate constants of
pseudo-first order and pseudo-second order models, respectively, h0 is ( )
∂N 1
the initial adsorption rate, α and β are the Elovich constant, kAv is the S=2 = (19f)
∂μ v(r) η
Avrami rate constant and n is the Avrami constant related to the
adsorption mechanism. μ2
The reliability of the models was evaluated by the determination ω= (19g)

coefficient values (R2), normalized standard deviations (∆qe) and Mar­
quadt’s percent standard deviation (MPSD). The values of ∆qe (%) and
3. Results and discussion
MPSD (%) were calculated according to Eqs. (17) and (18).
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑ [( )/ ]2 3.1. Model building and analysis of variance
qe,exp − qe,cal qe,exp
Δqe = 100 (17)
N− 1 Table 1 shows the details of the mixture design and the experimental
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ results of the response SBET. As can be seen from the results, the values of
√(
SBET ranged from 967 to 2056 m2 g–1, demonstrating that both compo­
√ )∑ N
1 ( )2
MPSD = 100√ qexp − qcal (18)
N − p i=1 sition and chemical nature of the mixture have great influence upon the
textural properties of the produced ACs. Several models were fitted to
where qe,exp and qe,cal are the experimental and calculated adsorption the experimental data, the fit summary is presented in Table S2. As
capacities (mg g–1), respectively, N is the number of adsorption assays evidenced, the best fitted model was the cubic one, once it provides the
and p is the total number of parameters from each model. highest values of R2 (0.9925), and adjusted R2 (0.9699), despite the
lowest value of SD (55.47). Thus, the cubic model allows to perform
valid predictions of a new set of mixtures, within the boundary condi­
2.8. DFT-based quantum chemical descriptors
tions of the design.
To further investigate the significance of the cubic model, the anal­
Fig. 2 presents the dimensions of the optimized geometries of the
ysis of variance (ANOVA) was performed, with the aim to verify if the
molecules in the ground state and the respective molecular electrostatic
observed response differences are real or due to variability, the results
potential surfaces. The optimization of the geometries and the energy of
are shown in Table S3. The significance of the model and factors was
the frontier molecular orbitals were obtained by quantum chemistry
determined by the Fisher distribution (F-value) and by the probability
calculations using GAMESS software (Schmidt et al., 1993). All com­
value (p-value) with 95% of confidence. A significant model is the one
putations were performed using density functional theory under
which presents Fcalculated > Ftabulated for the respective degrees of
generalized gradient approximation with B3LYP hybrid functional at the
freedom and p-value lower than 0.05. As shown in Table S3, Fcalculated
basis set 6-311++G(2d,p) for Met and at the basis set 6-31+G(d,p) for
attributed to the regression of the model is equal to 43.98, which is
RhB, such differences of the basis sets are attributed to the computa­
higher than the tabulated (F9,3 = 8.81). Furthermore, the p-value is
tional demand of the calculations. All quantum calculation output files
0.0050, ten times lower than the maximum value of 0.05 for a statisti­
are also attached in this manuscript as supplementary materials. The
cally significant model. In front of such results, it is possible to ensure
plots of the calculations were obtained using Gabedit graphical
that the cubic model is highly significant to predict the experimental
interface.
responses.
When a molecule approaches the surface of the adsorbent, its ability
A similar interpretation was performed for the factors and their in­
to stick itself to the surface is closely related to the frontier molecular
teractions. According to Table S3, only linear mixtures are significant,
orbitals, e.g., the highest occupied molecular orbital, HOMO, and the
showing p-value of 0.0010 and Fcalculated of 145.07 (higher than the
lowest unoccupied molecular orbital, LUMO. The former is attributed to
tabulated F2,3 = 9.55). The most important interactions in the mixture
the electron donating ability, while the latter is characteristic of the
are X1X2, X1X3, X1X2(X1 − X2), X2X3(X2 − X3), because the p-values are
electron acceptance ability of a molecule. A few other quantum de­
0.0442, 0.0067, 0.0483, and 0.0329, respectively, whereas the F-values
scriptors within the DFT concept of Parr, Pearson and Yang (Parr and
are higher than the tabulated one (F1,3 = 10.13). Controversially, if the
Pearson, 1983; Parr and Yang, 1989; Parr et al., 1999) also play
p-value is higher than 0.05 the terms of the model are not significant, e.
important contributions to describe the adsorption system, e.g., ioni­
g. 0.2204, 0.1965 and 0.0787, for the interactions X2X3, X1X2X3 and
zation potential (I), electron affinity (A), energy gap (Eg), global chem­
X1X3(X1 − X3), respectively, and the F-values are lower than F1,3.
ical hardness (η), global chemical softness (S), chemical potential (µ) and
According to the ANOVA results, an equation to predict the response
electrophilicity index (ω). Such quantum descriptors are often used to
SBET (Y1) is given by Eq. (20), only as a function of the significant terms.
explain Diels–Alder reactions and corrosion inhibition mechanisms

5
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Y1 = 687X1 + 353X2 + 322X3 − 92X1 X2 − 185X1 X3 + 57X1 X2 (X1 − X2 ) − 66X2 X3 (X2 − X3 ) (20)

demonstrates the model is useful to make good predictions of the results


The positive and negative signals associated with the interaction (which is in accordance with the high value of adjusted R2, 0.9699).
terms are attributed to synergic and antagonistic effects, respectively. A Furthermore, the measurement of the signal to noise ratio was calcu­
synergic effect is the one that produces greater responses from two or lated to be 22.48, which is greater than the desirable value (4.0).
more interacting factors than the sum of the individual responses, while Because of this, the cubic model is robust and can be used to navigate the
the antagonistic effect is the one that causes the final response to be design space within the composition range studied.
lower than the individual responses that would be obtained. In this Fig. 3 shows the projection of the response SBET (m2 g–1) of the
sense, it is possible to observe that the interaction factor X1X2(X1 − X2) experimental design as a contour plot and as a 3D surface. The diagram
is the only one that can cause an increase of the response by synergic reveals variations on the response from the highest value of 2056 m2 g–1
effect. However, such observation is restricted under the values of X1 (darker red region) to the lowest value of 967 m2 g–1 (darker blue re­
> X2; otherwise, this interaction factor becomes antagonistic. gion). The overall result is a curvilinear increase of SBET, in direction to
It is worth mentioning that the coefficient of variation (Table S3) is the mass fraction increase of component X1 (KOH). It is possible to
equals to 4.43%, setting forth that experimental error associated with observe that ACs with high SBET values (>1346 m2 g–1) can be obtained
the assays is low. The analyses of the residues are shown in Fig. S2. As with ternary mixtures of the components. The advantage to produce ACs
can be seen from Fig. S2(a–c), the residues have a normal distribution, from ternary mixtures of KOH, K2CO3 and K2C2O4 as activating agents
not demonstrating any kind of structure or tendency. The graphical relies on the possibility to produce efficient materials using less quantity
analysis of the residues shows the results are homoscedastic and so there of KOH, which is reported to be a hazardous and an expensive substance.
is no need to perform corrections in the model. Fig. S2d shows the According to EPA, KOH is a potentially toxic substance that might cause
correlation between the predicted values and the actual ones, that adverse effects to humans and to the environment.

Fig. 3. Response surface of the experimental design.

6
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 4. (a) N2 physisorption isotherms and pore size distribution (insert), (b) FHH plot as a relationship between ln Vads and ln2(P0/P).

In spite of KOH be an activating agent capable to produce ACs with 2.51 g of KOH and 0.49 g of K2CO3 with an expected response (SBET) of
high surface areas (which might be superior to 2000 m2 g–1), its use on 1925 m2 g–1, this optimization was selected by the highest value of the
large scale can be economically and environmentally unfavorable for the desirability function (0.64). The activated carbon produced in the
whole production process, once this alkali is highly corrosive and can optimized mixture (ACop) presented SBET value of 1984 m2 g–1, that is
cause liquefaction necrosis (Luttrell, 2010). Additionally, despite not only 2.97% different from the theoretical value; this variance is within
being combustible, a huge amount of KOH in contact with water can the coefficient of variation calculated from ANOVA (4.43%).
release enough heat to ignite combustibles, once the temperature in­ To confirm the synergic effect of the interaction between KOH and
crease may be superior to 86.5 ◦ C (Bespalko et al., 2019). Thus, there is a K2CO3, an AC was produced using only 2.51 g of KOH, whose mean SBET
need to reduce the amount of KOH used without drastically losing sur­ value was 1208 m2 g–1 (vide Fig. S3 in the Supplementary Material),
face area. In front of this situation, the optimization of the response was suggesting that, in fact, the presence of K2CO3 causes a catalytic influ­
performed. The optimum mixture composition suggested by the model is ence to develop a high SBET value. It must be emphasized that the SBET

Fig. 5. SEM images of the ACop.

7
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 6. (a) pHPZC, and (b) TPD patterns at different heating rates.

value of the ACop (1984 m2 g–1) is only 3.50% different from the highest (with APS of 1.89 nm). In this pore size range, there is a principal
one of the experimental design (2056 m2 g− 1), while the amount of KOH maximum at 1.33 nm and secondary maxima at 2.04, 2.36 and 2.59 nm,
is significantly decreased by 16.33%. confirming that the ACop shows a hierarchically structured pore distri­
The activation by KOH and K2CO3 might be attributed to (i) the bution, composed mainly by narrow and wider micropores.
direct reactions between the chemicals and the carbon matrix and (ii) The degree of roughness of the ACop surface was analyzed by the
indirect reactions, where gaseous products previously generated can FHH fractal dimension (D), the linear fits are presented in Fig. 4b. It is
damage the carbon structure under high temperature (Spessato et al., possible to observe that there are three distinct straight lines at the
2020). The main reactions for the chemical activation using KOH and entire relative pressure range. All correlation coefficient (R2) values are
K2CO3 at temperatures above 600 ◦ C are presented in reactions R1 and greater than 0.92, and indeed the ACop shows fractal characteristics. The
R2 (Spessato et al., 2020; Kim et al., 2020; Yue et al., 2018). In these values of D1, D2 and D3 are 2.97, 2.77 and 2.72, respectively. The D1
redox reactions, the KOH activator is consumed by the carbon matrix to value directly reflects the ACop structure, once it is attributed to the
produce K2CO3, reduced potassium and molecular hydrogen. Both relative pressure range of 0.07 < P/P0 < 0.16, which is characteristic of
produced and added K2CO3 are also consumed by the carbon matrix to N2 monolayer formation where van der Waals attractive forces between
generate reduced potassium and carbon monoxide. The gaseous prod­ adsorbent and adsorbate are dominant (Thommes et al., 2015). Due to
ucts are small molecules that aid the micropores development via the the high D1 value, it is possible to affirm that the material has a complex
damage to the carbon framework. Another pathway to the micropores pore structure whose irregularities are self-similar upon variations of
development is the intercalation of reduced potassium throughout the resolution. Furthermore, such high D value is attributed to disorganized
carbon matrix. carbon cross-linked between small graphite crystallites (Hayashi et al.,
2002). If the material would be composed of packed graphite structures
6KOH + 2C→2K + 3H2 +2K2 CO3 (R1)
domains, the fractal dimension would approach to 2.0, and the surface
would be flatter (Hayashi et al., 2002). Additionally, the calculated D2
K2 CO3 + 2C→2K + 3CO (R2)
and D3 values are 2.77 and 2.72, respectively, at the relative pressure of
0.16 < P/P0 < 0.99. In this range, there is a high number of lateral in­
3.2. Characterizations teractions between adsorbed N2 molecules and between successive N2
monolayers, in a way that the interface is controlled by the N2 surface
Fig. 4a presents the N2 physisorption isotherms of ACop and its pore tension, resulting in a decrease of the fractal dimension.
size distribution (insert). According to the IUPAC classification, the The results of the FHH fractal dimension can be corroborated by the
isotherm resembles type I(b), which is attributed to microporous solids SEM images (Fig. 5). As can be seen from Fig. 5, the ACop shows a certain
with small external surface areas. In fact, such statement was observed degree of roughness, e.g., the presence of hills, valleys, channels, and
from the micropores analysis, whose results demonstrated the ACop cracks, which can all be related to the high SBET values. The micrographs
presents an external surface area of 95 m2 g–1 and a micropore area of also reveal that the ACop presents a stepped surface, which is also an
1889 m2 g–1, which is directly attributed to the high micropore volume indicative that the adsorption sites are energetic inhomogeneous.
(0.85 cm3 g–1) in front of the total pore volume (0.96 cm3 g–1). The surface ability to equally adsorb negative and positive charges
Furthermore, the SBET value is 1984 m2 g–1, coinciding to the sum of was evaluated by the pHPZC, which is presented in Fig. 6a. As can be
external and micropore areas, suggesting the absence of wider meso­ seen, the pHPZC of ACop is equals to 5.05, which is the pH value where
pores. It must be pointed out that the C constant of BET equation is equal the material surface tends to be electrically neutral. This means the
to 2012.76, confirming high affinity between N2 molecules to the ACop material is negatively charged in aqueous medium at pHsolution > pHPZC,
surface. Also, the calculated ∆H◦ ads is equals to − 10.43 kJ mol–1 over whilst it is positively charged when pHsolution < pHPZC. The acid surface
the range 0.07 < P/P0 < 0.16, which illustrates the adsorption of the of the ACop is also observed by the TPD measurements.
first N2 monolayer is more exothermic than the N2 liquefaction. TPD has been mainly used in carbon materials field for the analysis of
The type I(b) isotherm is also attributed to materials that present a oxygen-containing functional groups. When a TPD analysis is being
broad pore size distribution, as confirmed by the NLDFT for slit and performed, oxygen-rich surfaces tend to thermally decompose in CO2
cylindrical pores (Fig. 4a insert). The ACop presents pores with diameters and CO which are released in different temperatures, depending on the
from a lower confidence limit of 1.33 nm to an upper limit of 3.54 nm thermal stability of the oxygenated surface groups. Strictly speaking,

8
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 7. Influence of initial pH over the adsorption capacity of (a) RhB and (b) Met (C0 = 500 mg L− 1, T = 30 ◦ C, pH = 3.0–13.0, 200 rpm for 360 min).

TPD measurements of carbon materials cannot be considered as a equilibrium conditions and their species distribution, both as function of
desorption technique, once the oxygen surface groups are decomposed the initial solution pH. From the results, it can be observed that ACop has
and not desorbed. As a result, the interpretation of TPD curves is rather high Qm values in the whole studied pH range. Specifically, Qm values
different from the case of non-carbon based materials, which are mainly for RhB and Met are greater than 395.49 and 58.82 mg g–1, respectively.
focused on the desorption of NH3 or CO chemically adsorbed. The TPD The Qm values for RhB tend to be constant (486.73 ± 4.95 mg g–1) in the
patterns of the ACop in different heating rates are shown in Fig. 6b. It is pH range 3.0–10.0, while the Qm values for Met are within the interval
straightforward to observe that the total amount of decomposition gases 95.89 ± 3.34 mg g–1 at the same pH range. From these analyses, the
is a function of the heating rate, in a way that the greater is the heating initial solution pH of the subsequent adsorption experiments was fixed
rate, the more is the released amount. It is notable the presence of at at 6.0, which is the pH value that provided great adsorption capacities of
least 4 overlapped peaks. In front of this, the TPD profiles were decon­ both adsorbates, besides being within the standards and effluent release
voluted using a similar procedure reported by Zhou et al. (2007). It was conditions established by the Environmental Protection Agency.
found that the TPD curves contained 5 peaks (shown in Fig. S4(a-c)) Some trends can be observed from the results presented in Fig. 7. On
centered at c.a. 431, 570, 702, 884 and 928 ◦ C. The TPD patterns of ACop the one hand, despite RhB adsorption tends to be constant at pH range
at different heating rates were used to estimate the energy for desorption from 3.0 to 10.0, it decreases at pH range 10.0–13.0. This observation
(Ed), according to the Kissinger method via linear dependence of ln can be attributed to the high hydroxide ions (OH-) concentration. At pH
(RT2p/β) vs T–1p (Vyazovkin et al., 2011). The linear fits are shown in > pHPZC, ACop surface is negatively charged and RhB is in a zwitterionic
Fig. S4d, the maximum temperature of each fitted peak (Tp) and the structure (RhB±). The positive charges of RhB± molecules are strongly
activation energy for desorption (Ed) are presented in Table S4. shielded by OH- and the greater is the OH- concentration, the greater is
Researches have tried to assign TPD peaks to determine surface the shielding effect. As a result, the formed complex RhB±(OH-)n tend to
groups on carbon-based materials (Figueiredo et al., 1999; Liu et al., be repelled by the ACop surface, decreasing the Qm values.
2019; Krzyżyńska et al., 2020; Ishii and Ozaki, 2020). Thus, analyzing On the other hand, Met adsorption increases with pH increasing,
Fig. S4(a-c) and comparing the Tp values with those of the aforemen­ such behavior is influenced by the electrostatic interactions between
tioned authors, it is appropriate to presume that peak 1 is attributed to ACop and Met species, which can be separated in three regions. Firstly, at
carboxylic and anhydride groups, peak 2 and 3 are probably due to the pH < pKa1 Met is mainly present as biprotonated species (over 50%) and
presence of lactones, ether and phenolic groups on different sites of the the ACop surface is positively charged (pHPZC = 5.05). In this pH inter­
graphitic edges (e.g., zig-zag or armchair), peak 4 is likely to be attrib­ val, long-range repulsive interactions between positive charges occur,
uted to carbonyl and semiquinone groups, whereas peak 5 is attributed hampering the molecules to access the adsorbent micropores, which is
to pyrone type structures (e.g., chromenes and phenantrenes). However, responsible for the lowest Qm value (58.82 mg g–1) observed at pH
qualitative analyses of TPD curves can be ambiguous, because the = 3.0. Secondly, at 5.0 < pH < 9.0 Met is mainly present as monop­
thermal decomposition is a function of the porous structure of the ma­ rotonated species and the ACop surface is negatively charged, resulting
terial (Krzyżyńska et al., 2020). Thus, due to the presence of micropores in long-range attractive forces which can be observed by the increase of
in the ACop, thermal secondary reactions can strongly affect the TPD Qm values to 95.89 ± 3.34 mg g–1. It is worth mentioning that the ACop
patterns, and qualitative assignments of deconvoluted peaks can lead to surface is positively charged at pH 4.0, whereas Met molecules are
important controversies among different papers. mainly present as monoprotonated species (> 94%), which results in
long-range repulsive interactions. However, the intensity of these re­
pulsions is weaker than those at pH 3.0, allowing the adsorbate mole­
3.3. Adsorption study
cules to approach the adsorbent surface more easily. Finally, at pH
> pKa2 Met molecules are mainly present as neutral species. The highest
3.3.1. Initial solution pH effect and DFT quantum descriptors
Qm values in this pH region (248.48 mg g–1 at pH 13.0) are due to the
The presence of electrical charges on surfaces is probably one of the
fact that Met molecules are less solvated, which facilitates mobility and
most important variables that encompass the adsorption phenomena.
therefore diffusivity within micropores.
The Coulombic interactions between adsorbent and adsorbate are
Since Met molecules are smaller than RhB molecules, Met could be
responsible for a myriad of responses, from high values of adsorption
expected to have the highest Qm values compared to RhB, once the
capacity (Qm) to the formation of adsorbate agglomerates that might
number of adsorption sites occupied for each Met molecule is lower than
cause a significant decrease of Qm (Spessato et al., 2019, 2020; Saleh,
RhB. However, what really happens is the opposite, RhB presents the
2015).
highest adsorption capacity, regardless of the experimental condition.
Fig. 7 presents the adsorption capacities results of Met and RhB at

9
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Table 2
Calculated values of the DFT-based quantum descriptors for RhB and Met.
Adsorbate Quantum descriptors (eV)

EH EL I A Eg µ η S ω
RhB − 5.78 − 3.01 5.78 3.01 2.77 − 4.40 1.38 0.72 13.39
Met − 6.51 − 0.33 6.51 0.33 6.17 − 3.42 3.09 0.32 18.07

Fig. 8. Equilibrium isotherms at different temperatures for simultaneous adsorption of (a) RhB and (b) Met (C0 = 25–900 mg L− 1, T = 30–50 ◦ C, pH = 6.0, 200 rpm
for 360 min).

Such observation can be attributed to three main reasons. Firstly, the important descriptors to the adsorption mechanism related to the
presence of aromatic rings in RhB structure and the absence in Met one. resistance of changes upon charge transference. The higher η value for
It is known that π-π electron-donor acceptor interactions between aro­ Met (3.09 eV) than RhB (1.38 eV) demonstrates that Met may bounce on
matic rings of the adsorbent and adsorbate play an important role to different points of the ACop surface before being adsorbed, whereas RhB
increase Qm (Georgakilas et al., 2012; Saleh, 2018). Secondly, the minor may have a lower collision rate to the surface before being adsorbed. The
aggregation index of RhB compared to Met. It is reported that aqueous electrophilicity index (ω) demonstrates that Met has a powerful elec­
solutions of RhB, up to concentrations of about 1.91 g L–1, are mainly trophilic center (ω = 18.07 eV) in comparison to RhB (ω = 13.39 eV). It
present as monocomponents systems (McHedlov-Petrosyan and Kholin, implicitly refers that RhB has a greater nucleophilic center than Met
2004; Ilich et al., 1996), whereas aqueous solutions of Met can form (Domingo et al., 2002) and both molecules can effectively interact with
dimers, trimers or higher-order dimeric structures (Nanubolu et al., each other (as later proved by the equilibrium studies of adsorption).
2013). At last, the intrinsic characteristics of each adsorbate, e.g., the
quantum chemical descriptors. One can be seen that the high SBET value 3.3.2. Simultaneous adsorption equilibrium isotherms
of an AC is not always the main factor that leads to great adsorption The equilibrium isotherms can be used to evaluate chemical affinities
capacities in front of different adsorbates. of the adsorbate for the adsorbent and correlate these affinities with
To better understand the interactions between ACop surface and the structural characteristics of the adsorbate. Fig. 8 shows the equilibrium
adsorbates, RhB and Met, the quantum chemical descriptors were isotherms at different temperatures (30–50 ◦ C) for simultaneous
calculated, the results are presented in Table 2. The energy values of the adsorption of RhB and Met on ACop. As can be seen, the equilibrium
frontier molecular orbitals are − 5.78 eV (EH) and − 3.01 eV (EL) for RhB, isotherms shapes of both adsorbates present remarkable differences.
whereas they are − 6.51 eV (EH) and − 0.33 eV (EL) for Met. The EH, EL, I According to the Giles classification (Giles et al., 1960), Met presents the
and A values are related to the chemical reactivity of the molecule. The L2 isotherm profile while RhB has the H2 type. L2 isotherm occurs when
higher is the EH value (thus, the lower I value), the greater is the the adsorption becomes progressively difficult to happen as a function of
chemical reactivity, favoring the electron-donating ability of HOMO the surface coverage degree, and the orientation of the adsorbate can be
orbitals. On the other hand, the higher is the EL value (thus, the lower A flat or end-on on adsorbent surface. On the other hand, H2 isotherm is
value), the lower is the electron-acceptance ability of LUMO orbitals. characteristic of high affinity between adsorbent and adsorbate, once
Therefore, RhB has a greater tendency to donate and to accept electrons adsorbate molecules are completely adsorbed at initial concentrations,
than Met. the adsorbate orientation tends to be flat to the surface in form of
It has been published that adsorption mechanisms can be governed J-aggregates species.
by electron sharing between adsorbate and adsorbent (Georgakilas It must also be pointed out that, changing the temperature of the
et al., 2012). Because of this, it is strictly important to determine the ease system have not caused any difference in the type of the adsorption
of electron flow from HOMO to LUMO inside a molecule by the bandgap isotherms. However, it is possible to observe that the positions of the
energy (Eg). The Eg value of RhB and Met is 2.77 (conductive) and curves are altered, demonstrating that the qe values are directly
6.17 eV (insulator), respectively. It is almost twice energetically easier dependent of the temperature of the system (for further details regarding
to excite HOMO electrons of RhB than Met ones. In this way, it is more the influence of temperature, the reader may check the thermodynamic
likely that RhB shares electron density with the ACop structure than Met. study section). As can be seen in Fig. 8, the experimental qm values for
The chemical hardness (η), and its reciprocal softness (S), are also RhB are 630.94 mg g− 1 at 30 ◦ C, 750.99 mg g− 1 at 40 ◦ C, and

10
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 9. Equilibrium isotherms at different temperatures for monocomponent adsorption of RhB (a–c) and Met (d–f). (C0 = 25–900 mg L− 1, T = 30–50 ◦ C, pH = 6.0,
200 rpm for 360 min).

808.49 mg g− 1 at 50 ◦ C, whereas the qm values for Met are 5.30%, respectively, over the investigated temperature range. Besides,
103.83 mg g− 1 at 30 ◦ C, 114.23 mg g− 1 at 40 ◦ C, and 113.88 mg g− 1 at the data in Fig. S5 are close to the 45◦ line, revealing that the calculate
50 ◦ C. adsorption capacities for both adsorbates are close to the experimental
To better understand the interactions between adsorbate and ones. The extended Freundlich model is an empirical model used to
adsorbent surface, besides the possible competition mechanisms, the describe the adsorption of multicomponent systems on energetically
non-linear models of multicomponent system (non-modified Langmuir, heterogeneous surfaces and when interactions among adsorbed mole­
modified Langmuir, extended Freundlich, and SRS) were fitted to the cules play an important role to develop multilayers.
experimental data. The parameters of the models and MPSD values are The Sheindorf-Rebuhn-Sheintuch (SRS) model is used to describe
presented in Table S5. Furthermore, the calculated qe vs experimental qe multi-component adsorption systems in which each component of the
plots are shown in Fig. S5. mixture follows the Freundlich assumptions, the model also assumes
As can be seen from Table S5 and Fig. S5, the models show high that the adsorption energy distribution takes an exponential form
values of MPSD and the data points are far from the 45◦ line, demon­ (Sheindorf et al., 1981). The SRS model presented satisfactory fit over
strating that simultaneous adsorption of RhB and Met might not be the experimental values of Met, whereas a poor fit was observed over the
strongly affected by competition mechanisms over the entire studied experimental data of RhB. The same conclusions are obtained from
concentration range. Firstly, non-modified competitive Langmuir model Fig. S5, in which the data are close to the 45◦ line only for Met results,
shows a poor fit over the experimental data with MPSD values ranging regardless the temperature of the system. The MPSD values for RhB
from 23.90% to 70.10% for RhB and from 44.80% to 81.60% for Met in ranged from 55.90% to 64.80%, whereas for Met it ranged from 8.60%
the temperature range 30–50 ◦ C, despite great distances of data from the to 14.80%. The aij is the competition coefficient, it demonstrates how
45◦ line in Fig. S5. However, it is seen a decrease of MPSD values upon component j inhibits the adsorption of component i. Thus, the aij values
increasing the temperature of the system, such observation is also for RhB range from 11.5 to 7.4 and the aij values of Met range from 1.6 to
proved by Fig. S5, in which the data points for non-modified Langmuir 1.0. Thus, it is more likely RhB inhibits the adsorption of Met than the
model approach the 45◦ line at low initial concentration of RhB and Met. opposite, specially at 30 ◦ C.
Thus, a competitive mechanism described by the non-modified Lang­ It must be highlighted that, increasing the temperature of the system,
muir model is present between RhB and Met for the available adsorption the competition rate between RhB and Met for an available adsorption
sites only at initial concentrations (up to 300 mg L–1) at 50 ◦ C. This site increases, as evidenced by the fit improvement of non-modified
model implies that different adsorbates compete for the same adsorption Langmuir and extended Freundlich models. Such trends can also be
site. Also, all the adsorption sites are energetically homogeneous and observed in Table S5, in which there is a decrease of aij coefficient for
there is no lateral interaction between adsorbed species. RhB from 11.5 to 7.4 in SRS model, the RhB tendency to inhibit the
Modified competitive Langmuir model also demonstrates a poor fit adsorption of Met molecules decreases at higher temperatures, which is
regardless the temperature of the system, with MPSD values ranging observed by the increase of Met experimental qe values from 103.83 to
from 7.70% to 42.10% for RhB and from 50.0% to 72.50% for Met. Also, 114.23 mg g–1. To better understand the influence of competition
the data points in Fig. S5 are far distance from the 45◦ line. Thus, the mechanism on simultaneous RhB and Met adsorption onto ACop, the
competitive Langmuir constant ηL,i worsened the fit of the model to monocomponent isotherms of RhB and Met at temperatures from 30 to
describe the experimental data. The extended Freundlich model 50◦ C were carried out.
demonstrated the best fit over the experimental data for both RhB and
Met, with MPSD values ranging from 3.50% to 9.40% and from 3.40% to

11
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Table 3 system temperature.


Thermodynamic parameters for monocomponent adsorption of RhB and Met On the other hand, the BET model presented the best fit over the
molecules. experimental data of Met adsorption from 30 to 50 ◦ C, as evidenced by
Temperature Linear equation Thermodynamic parameters the highest values of R2 (0.9845–0.9889) and the lowest values of Δqe
(K)
∆G◦ ∆H◦ ∆S◦ (J K-1
(5.2–9.3%) and MPSD (15.6–21.8%). In this model, the authors used
(kJ mol-1) (kJ mol-1) mol-1) Langmuir’s approach to explain multilayer adsorption. The Qm values
for BET model are 43.20, 46.26 and 38.41 mg g–1 at 30, 40 and 50 ◦ C,
Met
303 -36.14 2.13 126.29 respectively, which are close to the experimental qm values in the first
In Ke = 15.19 − 255.86 × T− 1 plateau of the isotherms, attributed to the first monolayer formation.
Experimentally, a second plateau and an asymptotic rise at higher initial
313 -37.39 concentrations are observed. According to the BET model, the second
323 R2 = 0.9780 -38.67 plateau is characteristic of the complete second monolayer formation,
RhB while the asymptotic rise is attached to the thickness of the adsorbed
303 -8.01 76.21 276.94
multilayer. One can observe that the asymptotic rise is also attributed to
the approximation of the equilibrium concentration (Ceq) to the water
1
In Ke = 33.31 − 9166.34 × T−

solubility concentration (CS) of Met, in a way that lim qe = ∞, simi­


313 -9.32 Ceq →CS
323 R2 = 0.9577 -12.84 larly to the type II physisorption isotherm presented by IUPAC classifi­
cation (Thommes et al., 2015). Furthermore, as the shape of Met
isotherms from the binary system are different from those obtained in
3.3.3. Monocomponent equilibrium isotherms the monocomponent system, it is evident that RhB interacts with Met
Fig. 9 shows the isotherms of RhB and Met in monocomponent sys­ molecules in a way that the differentiation of the first and second
tems, respectively. As can be seen from Fig. 9(a-c), the RhB mono­ plateau are no more evident, suggesting that interactions between RhB
component isotherms are H2 type according to the Giles classification and Met are more favorable to the system than those amongst Met
(Giles et al., 1960), similar of those obtained for simultaneous adsorp­ molecules.
tion, suggesting that the same interactions governs the adsorption To evaluate the ACop efficiency, the experimental qm values of RhB
mechanism. On the other hand, Fig. 9(d-f) demonstrates that Met and Met were compared with those obtained for other materials using
monocomponent isotherms are remarkable different from those ob­ the same adsorbates. As shown in Table S7, the Qm values obtained in
tained in binary mixtures. As stated by Giles, Met monocomponent this paper for RhB and Met are higher than other papers, either in
isotherms could be classified as L5, showing two plateaus and a further monocomponent adsorption or binary simultaneous adsorption systems.
rise at the end of the curves. The first plateau is indicative of a complete
monolayer formation over the ACop surface, whereas the second plateau 3.3.4. Interaction coefficient
represents the development of a new surface, which can be energetic The experimental interaction effect (EIEi) coefficient was calculated
regions of the previously adsorbed Met layer or exposed regions of the for each adsorbate. The EIEi is the ratio between the maximum
ACop surface. The further increase of the curves is likely to be related to adsorption capacity of adsorbate i in the binary adsorption system and
cooperative adsorption, as a result of hydrogen bonding amongst the the maximum adsorption capacity of the same adsorbate i in mono­
nitrogenated groups of Met structures, the phenomenon is also called the component system, the calculated EIEi values are presented in Fig. S6.
growth of adsorption islands (Dubinin et al., 1960). Overall, the EIEi is directly dependent of the system temperature, the
In order to better understand the interactions between adsorbates values are smaller than the unity at temperature range from 30 to 40 ◦ C,
and ACop surface in the monocomponent adsorption system, and whereas it is greater than the unity when the system temperature in­
consequently to get new insights regarding the binary system, the crease to 50 ◦ C. Such results mean that the adsorption capacities of Met
Langmuir, Freundlich, Dubinin-Radushkevich (DR) and Brunauer- and RhB is greater in the binary system than those obtained for the
Emmett-Teller (BET) models were adjusted to the experimental data. monocomponent system only at 50 ◦ C. In front of such results, it is
The models parameters, R2, ∆qe, MPSD and experimental qm values are possible to reaffirm that, increasing the temperature of the system, the
presented in Table S6. tendency of RhB molecules to inhibit the adsorption of Met is decreased.
The DR model showed the best fit over the experimental data for RhB
adsorption at the temperature range from 30 to 50◦ C, as confirmed by 3.3.5. Thermodynamics of adsorption
the highest values of R2 (0.9374–0.9855) and the lowest values of Δqe Thermodynamic parameters such as enthalpy change (∆H◦ ), entropy
(3.1–8.9%) and MPSD (46.9–315.8%). This model was originally change (∆S◦ ) and Gibbs free energy change (∆G◦ ), were calculated ac­
developed as a semi-empirical equation to describe adsorption equilib­ cording to the thermodynamic laws and the van’t Hoff equation, using a
rium isotherms of microporous materials considering Polanyi’s treat­ method carefully described by Lima et al., (2019). The thermodynamic
ment for adsorption potential theory (Dubinin, 1960). The Qm values of parameters were only calculated to the monocomponent system, once
DR model were 632.87, 764.53 and 771.26 mg g–1 at 30, 40 and 50 ◦ C, interactions between RhB and Met in simultaneous adsorption can un­
respectively, which are remarkably close to the experimental qm values. derestimate the energy values from the adsorption systems ACop-RhB
Furthermore, the magnitude of the E values (38.23–98.06 kJ mol–1) are and ACop-Met.
characteristic of strong interactions between adsorbent and adsorbate As previously shown from Fig. 9, the shape of the equilibrium iso­
(chemisorption), such as Coulombic, dipole-dipole, and hydrogen therms does not change upon temperature variations. However, the
bonding interactions, despite the Lennard-Jones 12/6 potential inter­ higher is the temperature, the higher is the qe values for RhB, whereas qe
action (Jensen et al., 2011). It is also worth mentioning that RhB values for Met subtly increases. Such results demonstrated that the
adsorption is limited to the monolayer formation (c values of the BET adsorption of both adsorbates is endothermic. The thermodynamic pa­
model are high (Adamson, 1994)), perhaps because of the opposing rameters are presented in Table 3. As can be seen, the ∆H◦ value for Met
walls of the involved pores, in which there would be a steric limitation to adsorption is 2.13 kJ mol–1, while for RhB adsorption is 76.21 kJ mol–1.
the thickness of the multilayer, this can be seen by the satisfactory fit of Such values are in accordance with the adsorption energy foreseen by
Langmuir model over the experimental data points. These mentioned the Dubinin-Radushkevich model. Specifically, for RhB adsorption, in
conclusions can be extrapolated to the adsorption of RhB in the presence which the mean adsorption energy in the temperature range 30–50 ◦ C is
of Met, once the shapes of the isotherms are similar, regardless of the 71.09 kJ mol–1, that is remarkably close to the experimental value.

12
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Fig. 10. Kinetic isotherms for simultaneous adsorption of RhB (a–c) and Met (d–f). (C0 = 300, 500 and 700 mg L− 1, T = 30 ◦ C, pH = 6.0, 200 rpm, t = 1.0–360 min).

Adsorption energy values can establish if the adsorption mechanism is the adsorbent tends to zero, whereas if Kn > 1.0 the diffusion is
governed by physisorption or chemisorption. Usually, ∆H◦ controlled by molecular flow, caused by the bouncing of molecules from
< 40 kJ mol–1 suggests the adsorption mechanism has physical nature, one wall region to another. The calculated Kn values of RhB and Met are
whereas if ∆H◦ > 40 kJ mol–1 suggests the adsorption has chemical 1.89 and 8.45, respectively. In this case, the diffusion of such molecules
nature. Thus, it is possible to conclude that RhB is chemisorbed (thus, its within the pores are followed by frequent collisions on the walls of ACop.
electronic density is significantly perturbed upon adsorption) while Met Kn for Met is greater than RhB, suggesting that Met experiences more
is physisorbed (there is no significant changes in eletronic structure). frequent collisions with the surface before to be adsorbed (which cor­
The adsorption process for both adsorbates is spontaneous (∆G◦ < 0) and roborates with the DFT quantum-based descriptors). Once Kn values are
evolve towards a state with higher entropy (∆S◦ > 0). Because both ∆H◦ greater than the unity, the Knudsen diffusivity (DK) can be calculated
and ∆S◦ are positives, the spontaneity of the process is entropy-driven. according to Eq. (22).
√̅̅̅̅̅̅̅̅̅
3.3.6. Kinetic isotherms DK =
d 8RT
(22)
Because the ACop is a nanoporous material, the mass transfer of the 3 πM
adsorbates through a porous material is challenging and a tough task to
where d is the APS (nm) of the ACop, R is the universal gas constant
be foreseen. As a result, do not exist a thoroughly satisfactory theory to
(8.314 J K–1 mol–1), T is the absolute temperature (K), and M is the
describe diffusion in porous materials. To date, the transport of mole­
adsorbate molecular weight (kg mol–1). The calculated DK values for
cules inside porous adsorbents can be driven by different mechanisms (e.
RhB and Met at 30 ◦ C are 7.30 × 10–8 m2 s–1 and 1.40 × 10–7 m2 s–1,
g., surface diffusion, Knudsen diffusion, viscous diffusion and effusion),
respectively. Such results demonstrate that Met approaches equilibrium
once molecular movements and the accessibility of active adsorption
faster than RhB, which is verified in the kinetic studies at 30 ◦ C of the
sites are governed by several variables, for example, the fractal dimen­
binary mixtures presented in Fig. 10.
sion of the adsorbent and the morphology of the nanopores (Malek and
Fig. 10 shows that the experimental equilibrium times (teq) of Met
Coppens, 2003).
are lower than the RhB ones. For instance, at initial concentration of
Knudsen diffusion is likely to be the dominant mechanism of
300 mg L–1, teq for Met is 5.0 min, whilst it is 30 min for RhB. At
adsorption in nanoporous materials. This type of mechanism assumes
500 mg L–1, teq for Met is 60 min, whereas it is 120 min for RhB. At
that, once within the pores, RhB and Met collide more frequently with
700 mg L–1, teq for Met is 180 min and it is 210 min for RhB. Overall,
the ACop walls than with themselves. The Knudsen number (Kn) is the
Met reaches equilibrium faster because it has a smaller molecular
parameter that defines what type of collision is dominant. Kn can be
structure. One can also be observed that teq values are directly propor­
calculated by the ratio between the mean free path (λ) of the molecule
tional to the initial concentration of the adsorbates, which can be due to
and the diameter of the pore (Do, 1998), in this case, the APS value
the high frequency of collisions and attractive interactions amongst the
obtained from the N2 physisorption results (1.89 nm). The λ (nm) values
molecules in the bulk phase, hampering the entrance of them to the
can be calculated according to Eq. (21).
porous structure of the material.
kT To get more details about the kinetics of adsorption, the models of
λ= √̅̅̅ (21)
πd2 P 2 pseudo-first order, pseudo-second order, Elovich and Avrami were
adjusted to the experimental data, the results are reported in Tables S8
in which k is the Boltzmann constant (1.38 × 10–23 m2 kg s–2 K–1),
and S9, along with the MPSD, R2, and ∆qe values. All models demon­
d is the molecule diameter (nm), T is the absolute temperature (K), and P
strated satisfactory fit over the experimental data, as can be seen from
is the total pressure (N m–2). If Kn < 1.0, then the diffusion is controlled
the high R2 values and low MPSD and ∆qe ones.
by viscous diffusion, in which the velocity of the solution on the walls of

13
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

other hand, at C0 > 200 mg L− 1, k1/k2 is smaller than θe, and the
Langmuir kinetics is reduced to pseudo-second-order (PSO) rate.
Fig. 11b shows the results for RhB, where k1/k2 is greater than θe at C0
< 300 mg L− 1, region which Langmuir kinetics can be reduced to PFO
rate, whereas at C0 > 200 mg L− 1, k1/k2 is smaller than θe, and the
Langmuir kinetics can be reduced to PSO rate. It is worth mentioning
that the critical turning point from PFO to PSO is the C0 value where the
ACop surface is half occupied by the adsorbates, that is, when θe = 0.5.

4. Conclusion

Sibipiruna pods were firstly used to produce activated carbons by


chemical activation method under augmented simplex-centroid mixture
design using the reactants KOH, K2CO3 and K2C2O4. It was found that
the optimum composition mixture is 2.51 g of KOH, 0.49 g of K2CO3 and
absence of K2C2O4, the experimental SBET value generated by this opti­
mization was 1984 m2 g− 1. The amount of KOH used in the production is
decreased by 16.33% in the presence of K2CO3 without drastically losing
surface area. The pore size distribution confirmed the ACop presents
narrow and wider micropores within the range of 1.33–3.54 nm. The
high values of FHH fractal dimension confirmed that the ACop has ir­
regularities self-similar upon variations of resolution, which were also
observed by SEM images. TPD curves demonstrated that the ACop pre­
sents different oxygenated molecular fragments, which is related to
pHPZC equals to 5.05. The adsorption equilibrium and the thermody­
namic studies evidenced that RhB adsorption is governed by chemi­
sorption, while Met is controlled by physisorption followed by the
development of adsorption islands. In the binary system, the qe values
for RhB and Met at 30 ◦ C were 630.94 and 103.83 mg g− 1, whereas for
the monocomponent system the qe values for RhB and Met at 30 ◦ C were
646.26 and 112.54, respectively. The adsorption kinetics proved that
the adsorption is fast and controlled by molecular flow (Knudsen
diffusion).

Fig. 11. Plots of k1/k2 and θe vs C0 for (a) Met and (b) RhB. CRediT authorship contribution statement

The best-fitted models for both RhB and Met, were PSO (for Lucas Spessato: Investigation, writing, and molecular modeling –
300 mg L–1) and Avrami (for 500 and 700 mg L–1). PSO model assumes original draft preparation. Vitor A. Duarte: Formal analysis and English
that the adsorption rate is second order with respect to the surface active review. Patrícia Viero: Formal analysis. Heloisa Zanella: Formal
sites (Tan and Hameed, 2017). As can be seen from Tables S8 and S9, the analysis. Jhessica M. Fonseca: Formal analysis. Pedro A. Arroyo:
qe values for RhB and Met (at C0 = 300 mg L–1) are 291.45 and Formal analysis. Vitor C. Almeida: Review and editing.
66.91 mg g–1, respectively, which are close to the experimental qe
values of 292.42 mg g–1 for RhB and 68.16 mg g–1 for Met. On the other Declaration of Competing Interest
hand, the Avrami kinetic model considers that the adsorption rate co­
efficient could have a temporal dependency during the adsorption pro­ The authors declare that they have no known financial interests or
cess. Kinetic isotherms which presents a time-dependent rate coefficient personal relationships that could have appeared to influence the work
is said to exhibit a “fractal-like kinetics” (Tan and Hameed, 2017). Ac­ reported in this paper.
cording to Tables S8 and S9, the fitted Avrami model describes kinetics
of fractional orders, with n values equals to 0.449 for RhB and 0.3811 for Acknowledgements
Met, both at C0 = 500 mg L− 1, besides 0.3964 for RhB and 0.3875 for
Met, both at C0 = 700 mg L− 1. Despite the Elovich model did not present LS acknowledge Coordenação de Aperfeiçoamento de Pessoal de
the best parameters over the experimental data, it presented a sufficient
Nível Superior (CAPES–Brazil) for his Ph.D. fellowship. VCS acknowl­
adjustment so that we can observe that all α values are superior to β edge Conselho Nacional de Desenvolvimento Científico e Tecnológico
values, for both RhB and Met, regardless the concentration range of the
(CNPq–Brazil) for his research project (Grant number: 426890/2016-7).
adsorbates, suggesting that the adsorption is preferable over the
desorption of such adsorbates on the ACop.
Appendix A. Supporting information
For the better understanding the adsorption mechanism under
nonequilibrium conditions at initial concentrations not reported by
Supplementary data associated with this article can be found in the
Fig. 10, it was carried out approximations of the modified Langmuir
online version at doi:10.1016/j.jhazmat.2021.125166.
kinetics, as reported by Liu and Shen (2008). Fig. 11 shows the relative
plots between the ratio of the first- and second-order kinetic constants
References
(k1/k2) and the degree of coverage of the surface (θe) versus the initial
concentration of the adsorbates (C0). As can be seen from Fig. 11a, for Adamson, A.W., 1994. Phys. Chem. Surf.
Met, k1/k2 is greater than θe at C0 < 200 mg L− 1, in this condition the Adhikari, S., Kim, D.H., 2020. Heterojunction C3N4/MoO3 microcomposite for highly
Langmuir kinetics is reduced to pseudo-first-order (PFO) rate. On the efficient photocatalytic oxidation of Rhodamine B,. Appl. Surf. Sci. 511 https://doi.
org/10.1016/j.apsusc.2020.145595.

14
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

Baei, M.T., Moradi, A.V., 2012. Quantum molecular descriptors and adsorption Ilich, P., Mishra, P.K., Macura, S., Burghardt, T.P., 1996. Direct observation of
properties of SCN2 on (6,0), (7,0), (8,0), and Ga-doped (6,0) zigzag single-walled rhodamine dimer structures in water. Spectrochim. Acta - Part A Mol. Spectrosc. 52,
boron nitride nanotubes: a computational study. Monatshefte Chem. 143, 1323–1330. https://doi.org/10.1016/0584-8539(96)01719-9.
1115–1121. https://doi.org/10.1007/s00706-011-0703-3. Ishii, T., Ozaki, J. ichi, 2020. Understanding the chemical structure of carbon edge sites
Bedin, K.C., Cazetta, A.L., Souza, I.P.A.F., Pezoti, O., Souza, L.S.C., Souza, P.S.C., by using deuterium-labeled temperature-programmed desorption technique. Carbon
Yokoyama, J.T.C., Almeida, V.C., 2018. Porosity enhancement of spherical activated N. Y. 161, 343–349. https://doi.org/10.1016/j.carbon.2020.01.079.
carbon: influence and optimization of hydrothermal synthesis conditions using Jensen, B., Kuznetsova, T., Kvamme, B., Oterhals, Å., 2011. Molecular dynamics study of
response surface methodology. J. Environ. Chem. Eng. 6, 991–999. https://doi.org/ selective adsorption of PCB on activated carbon. Fluid Phase Equilib. 307, 58–65.
10.1016/j.jece.2017.12.069. https://doi.org/10.1016/j.fluid.2011.05.001.
Bedin, K.C., Martins, A.C., Cazetta, A.L., Pezoti, O., Almeida, V.C., 2016. KOH-activated Ji, R., Zhao, Z., Yu, X., Chen, M., 2019. Determination of rhodamine B in capsicol using
carbon prepared from sucrose spherical carbon: adsorption equilibrium, kinetic and the first derivative absorption spectrum. Optik 181, 796–801. https://doi.org/
thermodynamic studies for Methylene Blue removal. Chem. Eng. J. 286, 476–484. 10.1016/j.ijleo.2018.12.141.
https://doi.org/10.1016/j.cej.2015.10.099. Khalili, N.R., Pan, M., Sandí, G., 2000. Determination of fractal dimensions of solid
. Bespalko, O. Halychyi, M. Roha, S. Poliakov, G. Kaleinikov, T. Naumenko, Experimental carbons from gas and liquid phase adsorption isotherms. Carbon N. Y. 38, 573–588.
Study of the Thermal Effect of the Dissolution Reaction for Some Alkalis and Salts https://doi.org/10.1016/S0008-6223(99)00143-8.
with Natural Mixing and Forced Stirring, in: E3S Web Conf., 2019. doi: https://doi. Kim, M.J., Choi, S.W., Kim, H., Mun, S., Lee, K.B., 2020. Simple synthesis of spent coffee
org/10.1051/e3sconf/201911801026. ground-based microporous carbons using K2CO3 as an activation agent and their
Burns, E.E., Carter, L.J., Kolpin, D.W., Thomas-Oates, J., Boxall, A.B.A., 2018. Temporal application to CO2 capture. Chem. Eng. J. 397 https://doi.org/10.1016/j.
and spatial variation in pharmaceutical concentrations in an urban river system. cej.2020.125404.
Water Res. 137, 72–85. https://doi.org/10.1016/j.watres.2018.02.066. Komarov, S., Yamamoto, T., Fang, Y., Hariu, D., 2020. Combined effect of acoustic
Cheng, Y.Y., Tsai, T.H., 2017. Pharmacokinetics and biodistribution of the illegal food cavitation and pulsed discharge plasma on wastewater treatment efficiency in a
colorant rhodamine B in rats. J. Agric. Food Chem. 65, 1078–1085. https://doi.org/ circulating reactor: a case study of Rhodamine B. Ultrason. Sonochem. 68 https://
10.1021/acs.jafc.6b04975. doi.org/10.1016/j.ultsonch.2020.105236.
Chen, J., Zhu, X., 2016. Magnetic solid phase extraction using ionic liquid-coated core- Krzyżyńska, B., Malaika, A., Ptaszyńska, K., Tolińska, A., Kirszensztejn, P.,
shell magnetic nanoparticles followed by high-performance liquid chromatography Kozłowski, M., 2020. Modified activated carbons for esterification of acetic acid with
for determination of Rhodamine B in food samples. Food Chem. 200, 10–15. https:// ethanol. Diam. Relat. Mater. 101 https://doi.org/10.1016/j.diamond.2019.107608.
doi.org/10.1016/j.foodchem.2016.01.002. Kumar, V., Singh, M., Behera, K., Pandey, S., 2020. Ionic liquid induced removal of
Cornell, J.A., 2011. Experiments with Mixtures: Designs, Models, and the Analysis of Rhodamine B from water. J. Mol. Liq. 319 https://doi.org/10.1016/j.
Mixture Data. John Wiley & Sons. molliq.2020.114195.
Danaee, I., Ghasemi, O., Rashed, G.R., Rashvand Avei, M., Maddahy, M.H., 2013. Effect Langmuir, I., 1916. The constitution and fundamental properties of solids and liquids.
of hydroxyl group position on adsorption behavior and corrosion inhibition of Part I. Solids. J. Am. Chem. Soc. 38, 2221–2295. https://doi.org/10.1021/
hydroxybenzaldehyde Schiff bases: electrochemical and quantum calculations. ja02268a002.
J. Mol. Struct. 1035, 247–259. https://doi.org/10.1016/j.molstruc.2012.11.013. László, K., Bóta, A., Nagy, L.G., Subklew, G., Schwuger, M.J., 1998. Fractal approach of
Domingo, L.R., Aurell, M.J., Pérez, P., Contreras, R., 2002. Quantitative characterization activated carbons from solid waste materials. Colloids Surf. A Physicochem. Eng.
of the global electrophilicity power of common diene/dienophile pairs in Diels-Alder Asp. 138, 29–37. https://doi.org/10.1016/S0927-7757(98)00227-1.
reactions. Tetrahedron 58, 4417–4423. https://doi.org/10.1016/S0040-4020(02) Leroy, S., Blach, J.F., Huvé, M., Léger, B., Kania, N., Henninot, J.F., Ponchel, A.,
00410-6. Saitzek, S., 2020. Photocatalytic and sonophotocatalytic degradation of rhodamine B
Do, D.D., 1998. Adsorption Analysis: Equilibria and Kinetics. https://doi.org/10.1142/ by nano-sized La2Ti2O7 oxides synthesized with sol-gel method. J. Photochem.
9781860943829. Photobiol. A Chem. 401 https://doi.org/10.1016/j.jphotochem.2020.112767.
Dubinin, M.M., 1960. The potential theory of adsorption of gases and vapors for Liang, Q., Liu, Y., Chen, M., Ma, L., Yang, B., Li, L., Liu, Q., 2020. Optimized preparation
adsorbents with energetically nonuniform surfaces. Chem. Rev. 60, 235–241. of activated carbon from coconut shell and municipal sludge. Mater. Chem. Phys.
https://doi.org/10.1021/cr60204a006. 241, 122327 https://doi.org/10.1016/j.matchemphys.2019.122327.
Dubinin, M.M., Zhukovskaya, E.G., Zaverina, E.D., 1960. Adsorption properties of carbon Lima, E.C., Hosseini-Bandegharaei, A., Moreno-Piraján, J.C., Anastopoulos, I., 2019.
adsorbents.communication 5. Peculiarities in the sorption of water vapor by active A critical review of the estimation of the thermodynamic parameters on adsorption
carbons at high relative pressures. Bull. Acad. Sci. USSR Div. Chem. Sci. 9, 903–911. equilibria. Wrong use of equilibrium constant in the Van’t Hoof equation for
https://doi.org/10.1007/BF00903960. calculation of thermodynamic parameters of adsorption. J. Mol. Liq. 273, 425–434.
Dubinin, M.M., 1960. Theory of the physical adsorption of gases and vapors and https://doi.org/10.1016/j.molliq.2018.10.048.
adsorption properties of adsorbents of various natures and porous structures. Bull. Liu, Y., Réty, B., Matei Ghimbeu, C., Soucaze-Guillous, B., Taberna, P.L., Simon, P., 2019.
Acad. Sci. USSR Div. Chem. Sci. 9, 1072–1078. https://doi.org/10.1007/ Understanding ageing mechanisms of porous carbons in non-aqueous electrolytes for
BF01161525. supercapacitors applications. J. Power Sources 434. https://doi.org/10.1016/j.
Ebadi, A., Soltan Mohammadzadeh, J.S., Khudiev, A., 2009. What is the correct form of jpowsour.2019.226734.
BET isotherm for modeling liquid phase adsorption? Adsorption 15, 65–73. https:// Liu, Y., Shen, L., 2008. From Langmuir kinetics to first- and second-order rate equations
doi.org/10.1007/s10450-009-9151-3. for adsorption. Langmuir 24, 11625–11630. https://doi.org/10.1021/la801839b.
Ebenso, E.E., Arslan, T., Kandemirli, F., Love, I., Öǧretir, C., Saracoǧlu, M., Umoren, S.A., Luttrell, W.E., 2010. Toxic tips: potassium hydroxide. J. Chem. Heal. Saf. 17, 54–55.
2010. Theoretical studies of some sulphonamides as corrosion inhibitors for mild https://doi.org/10.1016/j.jchas.2009.11.005.
steel in acidic medium. Int. J. Quantum Chem. 110, 2614–2636. https://doi.org/ Malek, K., Coppens, M.O., 2003. Knudsen self- and Fickian diffusion in rough nanoporous
10.1002/qua.22430. media. J. Chem. Phys. 119, 2801–2811. https://doi.org/10.1063/1.1584652.
Elizalde-Velázquez, G.A., Gómez-Oliván, L.M., 2020. Occurrence, toxic effects and Martins, A.C., Pezoti, O., Cazetta, A.L., Bedin, K.C., Yamazaki, D.A.S., Bandoch, G.F.G.,
removal of metformin in the aquatic environments in the world: recent trends and Asefa, T., Visentainer, J.V., Almeida, V.C., 2015. Removal of tetracycline by NaOH-
perspectives. Sci. Total Environ. 702 https://doi.org/10.1016/j. activated carbon produced from macadamia nut shells: kinetic and equilibrium
scitotenv.2019.134924. studies. Chem. Eng. J. 260, 291–299. https://doi.org/10.1016/j.cej.2014.09.017.
Figueiredo, J.L., Pereira, M.F.R., Freitas, M.M.A., Órfão, J.J.M., 1999. Modification of the McHedlov-Petrosyan, N.O., Kholin, Y.V., 2004. Aggregation of rhodamine B in water.
surface chemistry of activated carbons. Carbon N. Y. 37, 1379–1389. https://doi. Russ. J. Appl. Chem. 77, 414–422. https://doi.org/10.1023/B:
org/10.1016/S0008-6223(98)00333-9. RJAC.0000031281.69081.d0.
Freundlich, H., 1907. Über die Adsorption in Lösungen. Z. Phys. Chem. 57U https://doi. Nanubolu, J.B., Sridhar, B., Ravikumar, K., Sawant, K.D., Naik, T.A., Patkar, L.N.,
org/10.1515/zpch-1907-5723. Cherukuvada, S., Sreedhar, B., 2013. Polymorphism in metformin embonate salt-
Garba, Z.N., Rahim, A.A., 2014. Process optimization of K2C2O4-activated carbon from recurrence of dimeric and tetrameric guanidinium-carboxylate synthons.
Prosopis africana seed hulls using response surface methodology. J. Anal. Appl. CrystEngComm 15, 4448–4464. https://doi.org/10.1039/c3ce26986h.
Pyrolysis 107, 306–312. https://doi.org/10.1016/j.jaap.2014.03.016. Neamţu, M., Grandjean, D., Sienkiewicz, A., Le Faucheur, S., Slaveykova, V.,
Georgakilas, V., Otyepka, M., Bourlinos, A.B., Chandra, V., Kim, N., Kemp, K.C., Colmenares, J.J.V., Pulgarín, C., De Alencastro, L.F., 2014. Degradation of eight
Hobza, P., Zboril, R., Kim, K.S., 2012. Functionalization of graphene: covalent and relevant micropollutants in different water matrices by neutral photo-Fenton process
non-covalent approaches, derivatives and applications. Chem. Rev. 112, 6156–6214. under UV254 and simulated solar light irradiation - A comparative study. Appl.
https://doi.org/10.1021/cr3000412. Catal. B Environ. 158–159, 30–37. https://doi.org/10.1016/j.apcatb.2014.04.001.
Giles, C., MacEwanT, Nakhwa, S., Smith, D., 1960. Studies in adsorption part 16. Nezar, S., Laoufi, N.A., 2018. Electron acceptors effect on photocatalytic degradation of
J. Chem. Soc. 3, 3973–3993. 〈http://dns2.asia.edu.tw/~ysho/YSHO-En metformin under sunlight irradiation. Sol. Energy 164, 267–275. https://doi.org/
glish/1000%20WC/PDF/J%20Che%20Soc60,%203973.pdf〉 (accessed 20 April 10.1016/j.solener.2018.02.065.
2019). Niemuth, N.J., Klaper, R.D., 2015. Emerging wastewater contaminant metformin causes
Hayashi, J., Horikawa, T., Muroyama, K., Gomes, V.G., 2002. Activated carbon from intersex and reduced fecundity in fish. Chemosphere 135, 38–45. https://doi.org/
chickpea husk by chemical activation with K2CO3: preparation and characterization. 10.1016/j.chemosphere.2015.03.060.
Microporous Mesoporous Mater. 55, 63–68. https://doi.org/10.1016/S1387-1811 Oginni, O., Singh, K., Oporto, G., Dawson-Andoh, B., McDonald, L., Sabolsky, E., 2019.
(02)00406-7. Effect of one-step and two-step H3PO4 activation on activated carbon
Huang, Y., Wang, D., Liu, W., Zheng, L., Wang, Y., Liu, X., Fan, M., Gong, Z., 2020. Rapid characteristics. Bioresour. Technol. Rep. 8, 100307 https://doi.org/10.1016/j.
screening of rhodamine B in food by hydrogel solid-phase extraction coupled with biteb.2019.100307.
direct fluorescence detection. Food Chem. 316 https://doi.org/10.1016/j. Oosterhuis, M., Sacher, F., ter Laak, T.L., 2013. Prediction of concentration levels of
foodchem.2020.126378. metformin and other high consumption pharmaceuticals in wastewater and regional

15
L. Spessato et al. Journal of Hazardous Materials 411 (2021) 125166

surface water based on sales data. Sci. Total Environ. 442, 380–388. https://doi.org/ theoretical investigation of chemical reactivity of basic dyes using DFT-based
10.1016/j.scitotenv.2012.10.046. descriptors. Appl. Surf. Sci. 448, 662–670. https://doi.org/10.1016/j.
Pal, S., Chandrakumar, K.R.S., 2000. Critical study of local reactivity descriptors for apsusc.2018.04.087.
weak interactions: qualitative and quantitative analysis of adsorption of molecules in Spessato, L., Bedin, K.C., Cazetta, A.L., Souza, I.P.A.F., Duarte, V.A., Crespo, L.H.S.,
the zeolite lattice. J. Am. Chem. Soc. 122, 4145–4153. https://doi.org/10.1021/ Silva, M.C., Pontes, R.M., Almeida, V.C., 2019. KOH-super activated carbon from
ja993617t. biomass waste: Insights into the paracetamol adsorption mechanism and thermal
Parr, R.G., Pearson, R.G., 1983. Absolute hardness: companion parameter to absolute regeneration cycles. J. Hazard. Mater. 371, 499–505. https://doi.org/10.1016/j.
electronegativity. J. Am. Chem. Soc. 105, 7512–7516. https://doi.org/10.1021/ jhazmat.2019.02.102.
ja00364a005. Spessato, L., Cazetta, A.L., Melo, S., Pezoti, O., Tami, J., Ronix, A., Fonseca, J.M.,
Parr, R.G., Szentpály, L.V., Liu, S., 1999. Electrophilicity index. J. Am. Chem. Soc. 121, Martins, A.F., Silva, T.L., Almeida, V.C., 2020. Synthesis of superparamagnetic
1922–1924. https://doi.org/10.1021/ja983494x. activated carbon for paracetamol removal from aqueous solution. J. Mol. Liq. 300,
Parr, W., Yang, R.G., 1989. Density Functional Theory of Atoms and Molecules. 112282 https://doi.org/10.1016/j.molliq.2019.112282.
Pezoti, O., Cazetta, A.L., Bedin, K.C., Souza, L.S., Martins, A.C., Silva, T.L., Santos Sulaiman, N.S., Hashim, R., Mohamad Amini, M.H., Danish, M., Sulaiman, O., 2018.
Júnior, O.O., Visentainer, J.V., Almeida, V.C., 2016. NaOH-activated carbon of high Optimization of activated carbon preparation from cassava stem using response
surface area produced from guava seeds as a high-efficiency adsorbent for surface methodology on surface area and yield. J. Clean. Prod. 198, 1422–1430.
amoxicillin removal: kinetic, isotherm and thermodynamic studies. Chem. Eng. J. https://doi.org/10.1016/j.jclepro.2018.07.061.
288, 778–788. https://doi.org/10.1016/j.cej.2015.12.042. Tan, K.L., Hameed, B.H., 2017. Insight into the adsorption kinetics models for the
Poursat, B.A.J., van Spanning, R.J.M., Braster, M., Helmus, R., de Voogt, P., Parsons, J.R., removal of contaminants from aqueous solutions. J. Taiwan Inst. Chem. Eng. 74,
2019. Biodegradation of metformin and its transformation product, guanylurea, by 25–48. https://doi.org/10.1016/j.jtice.2017.01.024.
natural and exposed microbial communities. Ecotoxicol. Environ. Saf. 182 https:// Thommes, M., Kaneko, K., Neimark, A.V., Olivier, J.P., Rodriguez-Reinoso, F.,
doi.org/10.1016/j.ecoenv.2019.109414. Rouquerol, J., Sing, K.S.W., 2015. Physisorption of gases, with special reference to
Prahas, D., Kartika, Y., Indraswati, N., Ismadji, S., 2008. Activated carbon from jackfruit the evaluation of surface area and pore size distribution (IUPAC Technical Report).
peel waste by H3PO4 chemical activation: pore structure and surface chemistry Pure Appl. Chem. 87, 1051–1069. https://doi.org/10.1515/pac-2014-1117.
characterization. Chem. Eng. J. 140, 32–42. https://doi.org/10.1016/J. Vyazovkin, S., Burnham, A.K., Criado, J.M., Pérez-Maqueda, L.A., Popescu, C.,
CEJ.2007.08.032. Sbirrazzuoli, N., 2011. ICTAC Kinetics Committee recommendations for performing
Saleh, T.A., 2015. Isotherm, kinetic, and thermodynamic studies on Hg(II) adsorption kinetic computations on thermal analysis data. Thermochim. Acta 520, 1–19.
from aqueous solution by silica- multiwall carbon nanotubes. Environ. Sci. Pollut. https://doi.org/10.1016/j.tca.2011.03.034.
Res. 22, 16721–16731. https://doi.org/10.1007/s11356-015-4866-z. Wang, L., Sun, F., Hao, F., Qu, Z., Gao, J., Liu, M., Wang, K., Zhao, G., Qin, Y., 2020.
Saleh, T.A., 2018. Simultaneous adsorptive desulfurization of diesel fuel over bimetallic A green trace K2CO3 induced catalytic activation strategy for developing coal-
nanoparticles loaded on activated carbon. J. Clean. Prod. 172, 2123–2132. https:// converted activated carbon as advanced candidate for CO2 adsorption and
doi.org/10.1016/j.jclepro.2017.11.208. supercapacitors. Chem. Eng. J. 383 https://doi.org/10.1016/j.cej.2019.123205.
Scheffé, H., 1958. Experiments with mixtures. J. R. Stat. Soc. Ser. B 20, 344–360. https:// Yagmur, E., Gokce, Y., Tekin, S., Semerci, N.I., Aktas, Z., 2020. Characteristics and
doi.org/10.1111/j.2517-6161.1958.tb00299.x. comparison of activated carbons prepared from oleaster (Elaeagnus angustifolia L.)
Scheffé, H., 1963. The simplex-centroid design for experiments with mixtures. J. R. Stat. fruit using KOH and ZnCl2. Fuel 267. https://doi.org/10.1016/j.fuel.2020.117232.
Soc. Ser. B 25, 235–251. https://doi.org/10.1111/j.2517-6161.1963.tb00506.x. Yokoyama, J.T.C., Cazetta, A.L., Bedin, K.C., Spessato, L., Fonseca, J.M., Carraro, P.S.,
Schmidt, M.W., Baldridge, K.K., Boatz, J.A., Elbert, S.T., Gordon, M.S., Jensen, J.H., Ronix, A., Silva, M.C., Silva, T.L., Almeida, V.C., 2019. Stevia residue as new
Koseki, S., Matsunaga, N., Nguyen, K.A., Su, S., Windus, T.L., Dupuis, M., precursor of CO2-activated carbon: optimization of preparation condition and
Montgomery, J.A., 1993. General atomic and molecular electronic structure system. adsorption study of triclosan. Ecotoxicol. Environ. Saf. 172, 403–410. https://doi.
J. Comput. Chem. 14, 1347–1363. https://doi.org/10.1002/jcc.540141112. org/10.1016/j.ecoenv.2019.01.096.
Sevilla, M., Ferrero, G.A., Fuertes, A.B., 2017. Beyond KOH activation for the synthesis of Yue, L., Xia, Q., Wang, L., Wang, L., DaCosta, H., Yang, J., Hu, X., 2018. CO2 adsorption
superactivated carbons from hydrochar. Carbon N. Y. 114, 50–58. https://doi.org/ at nitrogen-doped carbons prepared by K2CO3 activation of urea-modified coconut
10.1016/j.carbon.2016.12.010. shell. J. Colloid Interface Sci. 511, 259–267. https://doi.org/10.1016/j.
Sharma, J., Sukriti, P. Anand, Pruthi, V., Chaddha, A.S., Bhatia, J., Kaith, B.S., 2017. jcis.2017.09.040.
RSM-CCD optimized adsorbent for the sequestration of carcinogenic rhodamine-B: Yu, Q., Zhao, H., Zhao, H., Sun, S., Ji, X., Li, M., Wang, Y., 2019. Preparation of tobacco-
kinetics and equilibrium studies. Mater. Chem. Phys. 196, 270–283. https://doi.org/ stem activated carbon from using response surface methodology and its application
10.1016/j.matchemphys.2017.04.042. for water vapor adsorption in solar drying system. Sol. Energy 177, 324–336.
Sheindorf, C., Rebhun, M., Sheintuch, M., 1981. A Freundlich-type multicomponent https://doi.org/10.1016/j.solener.2018.11.029.
isotherm. J. Colloid Interface Sci. 79, 136–142. https://doi.org/10.1016/0021-9797 Zhou, P., Li, W., Zhang, J., Zhang, G., Cheng, X., Liu, Y., Huo, X., Zhang, Y., 2019.
(81)90056-4. Removal of Rhodamine B during the corrosion of zero valent tungsten via a tungsten
Singh, S., Parveen, N., Gupta, H., 2018. Adsorptive decontamination of rhodamine-B species-catalyzed Fenton-like system. J. Taiwan Inst. Chem. Eng. 100, 202–209.
from water using banana peel powder: a biosorbent. Environ. Technol. Innov. 12, https://doi.org/10.1016/j.jtice.2019.04.023.
189–195. https://doi.org/10.1016/j.eti.2018.09.001. Zhou, J.H., Sui, Z.J., Zhu, J., Li, P., Chen, D., Dai, Y.C., Yuan, W.K., 2007.
de Souza, T.N.V., de Carvalho, S.M.L., Vieira, M.G.A., da Silva, M.G.C., do, D., Brasil, S. Characterization of surface oxygen complexes on carbon nanofibers by TPD, XPS and
B., 2018. Adsorption of basic dyes onto activated carbon: experimental and FT-IR. Carbon N. Y. 45, 785–796. https://doi.org/10.1016/j.carbon.2006.11.019.

16

You might also like