You are on page 1of 14

Chemical Papers

https://doi.org/10.1007/s11696-021-01637-4

ORIGINAL PAPER

Experimental and modeling study on adsorption of emerging


contaminants onto hyper‑crosslinked cellulose
Sarra Benosmane1 · Meriem Bendjelloul1,2 · El Hadj Elandaloussi2   · Moufida Touhami3 ·
Louis‑Charles de Ménorval4

Received: 3 March 2021 / Accepted: 31 March 2021


© Institute of Chemistry, Slovak Academy of Sciences 2021

Abstract
The primary objective of this study was to synthesize a novel hyper-crosslinked cellulosic adsorbent (CLC) by a straight-
forward one-pot esterification reaction using cellulose and sebacoyl chloride as crosslinker agent. Efficient crosslinking of
cellulose was confirmed by FTIR, CPMAS 13C NMR, elemental analysis, TGA, SEM and BET surface area analysis. The
CLC material with high content of ester groups was employed to remove paracetamol (PCT) and niflumic acid (NFA) from
aqueous solutions in batch adsorption experiments. The mesoporous structure of CLC created upon crosslinking was found
to be a determinant in the adsorption behavior of the drugs. Indeed, PCT and NFA adsorption isotherms onto CLC were
S-shaped and were adjusted by the Gu–Zhu, Frumkin–Fowler–Guggenheim and Hill–de Boer models. From the GZ isotherm
model, the results indicate a cooperative adsorption mechanism leading to the formation of aggregates containing 2.53 and
4.25 molecules of PCT and NFA, respectively. Furthermore, FFG and HdB models reflect that the lateral interactions are
attractive in nature for both drugs. The experimental kinetic data were fitted to the pseudo-second-order and intraparticle
diffusion models, and the obtained parameters were linked to the aggregates arrangement of the drugs.

Keywords  Crosslinked cellulose · Niflumic acid · S-type isotherm · Cooperative adsorption · Adsorption kinetics

Introduction effluents of wastewater treatment plants (WWTPs) and sur-


face water worldwide (Larsson et al. 2007; Miege et al.
Over the past decades, the scientific community has become 2009; Joss et al. 2006; Sim et al. 2011).
increasingly concerned about the potential problems related Current research emphasizes on the design and develop-
to water pollution by emerging pollutants (EPs). Among ment of materials able to remove a large variety of chemicals
them, pharmaceuticals and personal care products (PPCPs) from wastewater effluents and natural water compartments.
are detected at concentrations of mg/L or below in secondary In this context, adsorption is considered a promising alter-
native as it removes a wide variety of organic and inorganic
compounds and generates less toxic products than many
* El Hadj Elandaloussi other remediation methods (Rivera-Utrilla et  al. 2013).
elhadj.elandaloussi@univ-relizane.dz; Thus, biological adsorbents, activated carbons, lignocellu-
andalou777@yahoo.com losic wastes, polymer resins, and other solids have been suc-
1
Laboratoire de Valorisation des Matériaux, Université cessfully applied for pharmaceutical removal from aqueous
Abdelhamid Ibn Badis, B.P. 227, 27000 Mostaganem, solutions (Escapa et al. 2017; Streit et al. 2021; Nourmo-
Algeria radi et al. 2018; Bahamon and Vega 2017; Boudrahem et al.
2
Department of Chemistry, Faculty of Sciences 2017; Ferreira et al. 2015; Thakur et al. 2020; Villaescusa
and Technology, Ahmed Zabana University-Relizane, et al. 2011; Robberson et al. 2006; Menk et al. 2019).
48000 Relizane, Algeria In recent years, biopolymers such as cellulose have been
3
Modeling and Calculation Methods Laboratory, Moulay thoroughly investigated in the removal of pollutants from
Tahar University, 20000 Saida, Algeria water owing to their wide availability, biodegradability, low
4
ICG–AIME–UMR 5253, Université Montpellier 2, Place cost, low toxicity, regenerability, and lower sludge genera-
Eugène Bataillon CC 1502, 34095 Montpellier Cedex 05, tion compared to many other processes (Bhatnagar et al.
France

13
Vol.:(0123456789)
Chemical Papers

2015; Hokkanen et al. 2016). Moreover, chemical modi- Herein, we developed an efficient and simple one-step
fication of cellulose, especially esterification by means of procedure to prepare a highly crosslinked cellulose CLC for
anhydrides or acid chlorides, is an appealing approach that the removal of targeted pharmaceuticals. The CLC material
allows the synthesis of materials with desired functionalities can be easily synthesized by a simple chemical crosslinking
designed for a significant enhancement of the adsorption method since cellulose contains abundant hydroxyl groups
capacity (Li et al. 2019a, b; Wang 2019; Ji et al. 2019; Zhou which can be crosslinked with each other by means of seba-
et al. 2015). coyl chloride (ClOC–(CH2)8–COCl) as crosslinker agent.
To date and to the best of our knowledge, literature relat- The as-prepared adsorbent material was fully characterized
ing functionalized cellulose as adsorbent materials for phar- and evaluated for its sorption capacity of PCT and NFA
maceuticals removal is very scarce. Nonetheless, Selkälä drugs. In addition, the surface functional group coverage
et al. (2018) have recently reported the use of anionic cel- along with the morphology and porosity of CLC material
lulose nanofibrils for the removal of pharmaceutical salbu- were investigated to identify the sorption mechanism of the
tamol from aqueous solutions. pharmaceuticals removal.
Taking into account the abundant surface hydroxyl groups
on cellulose, polyesterification by means of a diacid chloride
could be an effective method to introduce plenty of adsorp-
Experimental
tion sites on the resulting crosslinked material. We hypothe-
sized that the incorporation of ester sites within the polymer
Chemicals
framework should provide an excellent sorption property
involving weak van der Waals interactions and hydrogen
Cellulose was kindly provided by a local sugar factory.
bonding, especially for nonionic pharmaceutical pollut-
The material is used in sugar refining process as filter aid.
ants. For this study, paracetamol (PCT) and niflumic acid
Paracetamol (PCT) and niflumic acid (NFA) were kindly
(NFA) were chosen as model nonsteroidal anti-inflammatory
provided by Saidal pharmaceutical group (Oran, Algeria).
drugs (NSAIDs) due to their wide use (Coelho et al. 2010).
All others chemicals were analytical reagent grade and were
Because of its relative high solubility and hydrophilicity,
used without further purification.
paracetamol (PCT) among other drugs is of recent concern
due to its persistence in the environment which indicates its
incomplete removal by the conventional processes applied Synthesis of CLC
in WWTPs (Petrie et al. 2015; Franca et al. 2016; Xiong and
Hu 2017). Although, niflumic acid (NFA), a widely used The crosslinking reaction was carried out using 8.1  g
NSAID, is also enlisted among the EPs of possible concern (50 mmol) of cellulose pre-swollen for 8 h in 150 mL of
(Mila et al. 2019; Ibanez et al. 2016; Alygizakis et al. 2016), dry pyridine. The crosslinking agent (5.98 g of sebacoyl
it has not been extensively studied so far. Besides, to the chloride, 25 mmol) in 10 mL of pyridine was then slowly
best of our knowledge, there is lack of relevant report on the added to the slurry cooled in an ice bath. After allowing the
removal of niflumic acid through adsorption in the literature, mixture to come to room temperature, it was then heated at
which limits the comparison of our results to the work of 80 °C for 8 h. On cooling, the hypercrosslinked polymer was
others. The molecular structures of these pharmaceuticals collected by filtration and washed with methanol, 0.5 mol/L
along with some of their physicochemical properties are HCl in acetone, and again with methanol. The material was
depicted in Table 1. then dried in an electric drying oven at 100 °C for 24 h and

Table 1  Molecular structures and properties of selected pharmaceuticals


Compound Molecular structure Molecular weight λmax (nm) pKa Molecular Size (nm)
(length × height × width)

Paracetamol (PCT) H 151.16 243 (pH 1.2–10.0) 9.52 (Box and Comer. 1.16 × 0.7 × 0.41
N
2008)

O
HO

Niflumic acid (NFA) CO2 H 282.22 254 (pH 1.2) 2.26, 4.44 (Box et al. 1.32 × 0.94 × 0.5
H 286 (pH 6.8) 2006)
F3 C N 288 (Ethanol)

13
Chemical Papers

passed through a 250-μm sieve to afford 11.27 g of CLC as The removal of PCT and NFA by CLC sorbent was
an off-white solid. investigated in batch experiments by stirring 50 mg of CLC
material with 50 mL of pharmaceuticals solutions in 200-
Characterization techniques mL stoppered glass bottles equipped with magnetic stirring
bars. All studies were conducted at room temperature on a
The following characterization methods were used for the stirring plate adjusted to a 100 rpm stirring rate for 60 min.
chemical analysis of the CLC sorbent. The Fourier trans- The drugs concentrations were measured with a HACH
form infrared (FTIR) spectra were collected using a Nico- DR4000U UV–visible spectrophotometer at 243 and 286 nm
let Avatar 330 Fourier transform IR spectrometer. The 13C for PCT and NFA, respectively. The sorbed amounts of phar-
cross-polarization magic angle spinning nuclear magnetic maceuticals were determined from the difference between
resonance (CP/MAS NMR) spectra were recorded on a the initial and final concentrations by the following mass
Bruker 300 (Digital NMR Avance) spectrometer. The N ­ 2 balance equation:
adsorption–desorption experiments at 77 K were carried out ( )
using a Micromeritics Tristar 3000. Prior to the analysis, Ci − Ce × V
qe = (1)
CLC sample was first treated to desorption at reduced pres- w
sure (< ­10−2 Torr) at 150 °C for 5 h. The specific surface
where qe is the amount (mg g­ −1) of drug sorbed, Ci and Ce
area (SSA) was calculated using the BET method based on
are the initial and equilibrium drug concentrations (mg L ­ −1)
the adsorption isotherm and pore size, and pore volume val-
in solution, respectively, V is the adsorbate volume (L) and
ues were calculated using the BJH method on desorption
w is the sorbent weight (g).
isotherm branch. The morphology of the crosslinked poly-
To optimize the variables affecting the drugs adsorption,
mer was examined at high magnification using a HITACHI
the effects of pH were investigated during 60 min at 20 ± 2 °C.
S-4800 SEM. High-resolution transmission electron micros-
Experiments with each drug were performed to determine the
copy (HR-TEM) image of CLC was collected using a JEM-
effect of pH on drug removal. The pH range studied was from
3010 (JEOL, Japan). The thermal stability of the samples
1.2 to 10. The sample mass was 50 mg, and drug concentration
was performed using TGA on a NETZSCHSTA 409 PC/PG
was 30 mg ­L−1 (50 mL) in this series of tests. The maximum
simultaneous thermal analyzer at a heating rate of 10 °C/min
absorption wavelength of PCT was determined as equal to
under nitrogen atmosphere.
243 nm. At this wavelength, spectra of PCT were found not to
vary in the pH range (1.2–10.0). For NFA, λmax shifted from
Adsorption experiments
254 nm at pH 1.2–286 nm at pH 6.8. The kinetic measure-
ments were carried out with similar equipment and conditions.
A series of adsorption experiments were conducted by pre-
The sorbent mass was 50 mg, and the volume of the drug
paring solutions of pharmaceutical compounds in distilled
solution was 50 mL (30 mg L ­ −1) in this series of tests. The
water or buffer solutions at different initial concentrations
mixtures were stirred at predetermined intervals of time and
ranging from 0 to 100 mg/L. Stock solution of PCT (100 mg
were drawn for drug concentration analysis. Each isotherm
­L−1) was prepared by dissolving the drug in distilled water
consisted of 10 drugs concentrations varied from 10 to 100 mg
and subsequently diluted to the required concentrations. For
­L−1. The equilibrium concentrations of different combinations
the sparingly water-soluble NFA drug, aqueous solutions of
were measured by the spectrophotometer and referenced with
10–100 mg ­L−1 were prepared by using either 0.1 M HCl
the calibration curves. Drugs saturated CLC that remained
(pH 1.2), buffer solution (pH 1.2) or phosphate buffer (pH
after centrifugation in the adsorption assays (for 100 mg ­L−1
6.8), under vigorous stirring for 4–6 h to ensure complete
initial drugs concentrations) were used for further desorption
dissolution of the drug. When required, PCT solutions of
studies. The recovered materials were washed with distilled
desired pH values were made by adjustment using 0.1 M
water, air-dried, and then suspended in 20 mL of aqueous HCl
HCl or NaOH solutions. Buffer solution at pH 1.2 was
solutions (pH 1.2 for NFA and pH 4 for PCT), and thoroughly
prepared by mixing 25 mL of 0.2 M sodium chloride with
stirred during 30 min. Afterward, they were centrifuged for
42.5 mL of 0.2 M of hydrochloric acid. Prior to make up the
adsorbate determination in the supernatant. The influence of
volume to 200 mL, the final pH value was adjusted. Buffer at
temperature on the removal process was studied at three dif-
pH 6.8 was prepared mixing 125 mL of a potassium dibasic
ferent temperatures (25, 35 and 45 °C) with CLC suspensions
phosphate (0.1 M) with 14 mL sodium hydroxide (0.1 M);
in PCT and NFA solutions (30 mg L ­ −1). The suspensions were
the final pH value was adjusted either with HCl or sodium
stirred during 60 min, and then the drugs concentrations were
hydroxide as required.
analyzed.

13
Chemical Papers

Modeling of PCT and NFA adsorption OH C-H CH2 C=O C-H C-O C-H

}
----
2848 ----

----
----

----
3456 ----

3040 ----
2930 ----

1710 ----
3348 ----

----
1320

1063
1164

920
1112
Adsorption kinetic and isotherm models

Transmittance (a.u.)
The control mechanism of the adsorption process of PCT
and NFA onto the CLC was evaluated using nonlinear
and linear kinetic equations. Therefore, three widely used
kinetic models including pseudo-first order, pseudo-second
order, and intraparticle diffusion were applied to analyze
the obtained data of adsorption experiments. Additionally,
three adsorption isotherm models, viz. Gu–Zhu, Frum-
kin–Fowler–Guggenheim and Hill–de Boer, have been used
to describe the equilibrium adsorption of pharmaceuticals Cellulose
in the present study. CLC

3500 3000 2500 2000 1500 1000 500


Fitness of kinetic and isotherm models Wavenumbers (cm ) -1

To select the most suitable kinetic and isotherm models, it Fig. 1  FTIR spectra of cellulose and CLC
is necessary to evaluate their validity. Therefore, in this
study, apart from the correlation coefficient (R 2), the
adjusted determination factor ( R2adj ) and the root-mean- Table 2  Elemental analyses of cellulose and crosslinked cellulose

square error (RMSE) were also used as criteria for the good- Samples Elemental analysis (wt %)
ness of fit between the experimental and predicted data. O/C C O H
These functions have been used previously (Babaei et al.
2016; Ciğeroğlu et al. 2021) and are given as: Cellulose 1.16 43.24 50.41 6.35
CLC 0.58 59.28 34.75 5.97

∑n � �2
qi,exp − qi,cal
RMSE =
i=1 (2)
n to the stretching vibration of C=O bonds of ester groups.
The increasing intensities of the symmetric (2848 ­cm−1) and
asymmetric (2930 ­cm−1) ­CH2 stretching vibrations as well
( )
) n−1
R2adj = 1 − 1 − R2 (3)
(
n−p as the increase in the out-of-plane bending vibrations of the
­CH2 (719 ­cm−1) and the in-plane bending vibrations of the
where R2 is the correlation coefficient, qi,calc is each value of C−H bonds (1469 ­cm−1) also reflect the occurrence of the
q predicted by the fitted model, qi,exp is each value of q meas- esterification reaction between cellulose hydroxyl groups
ured experimentally, n is the number of experiments per- and sebacoyl chloride. These observations are in agree-
formed, and p is the number of parameters of the fitted ment with previous work dealing with cellulose esters (Li
model. The smaller RMSE value and the R2adj value close to et al. 2019a; Jandura et al 2000). Furthermore, the extinc-
one suggest the best and the most valid model. tion of the low intensity band at 1640 ­cm−1 is indicative of
a decrease in water content from cellulose precursor after
esterification. On the other hand, the decrease in intensities
Results and discussion of bands at 1112, 1063 and 1032 ­cm−1 originated from the
vibrations of the C−O bonds of alcohols bound to carbon
Characterization of CLC sorbent 2, 3 and 6 of cellulose suggested that the esterification is
not selective and that sebacoyl chloride reacts equally easily
The CLC material was analyzed using a series of charac- with primary and secondary alcohols.
terization techniques to confirm the structure of chemical In addition, elemental analyses revealed that the carbon
functional groups resulting from the crosslinking reaction. content of the cellulose precursor was lower than that of
Figure 1 shows FTIR spectra of precursor cellulose and the theoretical values for cellulose, ­C6H10O5, which is indu-
the synthesized crosslinked cellulose material. Compared bitably attributed to the water content of the microfibrils.
to the precursor sample, CLC spectrum showed distinctive Moreover, as can be seen in Table 2, the smaller O/C ratio
intense absorption band at 1710 ­cm−1, which corresponds found for the crosslinked cellulose is attributed to the 8
methylene groups of the coupling agent, resulting in higher

13
Chemical Papers

carbon content. These results give a clear evidence for the (a) TGA Cellulose DTG Cellulose
occurrence of esterification reaction. 100
TGA CLC DTG CLC
0
The solid-state 13C-NMR spectra of cellulose sample
and synthesized crosslinked cellulose polyester are shown 80

in Fig. 2. The spectrum of unmodified cellulose shows sig-

dW/dt (%/min)
Weight (%)
nals of the six carbon atoms of the glucose unit which were 60

assigned according to the literature (Foston et al. 2011;


40 -5
Larsson et al. 1995), while CLC spectrum shows additional
peaks. One peak at 176 ppm is due to the resonance of
20
carbonyl groups and other peaks at 26–35 ppm, which are
attributed to the 8 methylene carbons of sebacoyl linkers. 0
The thermal stability of the synthesized CLC and its pre- -10
cursor were studied by TGA, the TG curves and their first 0 100 200 300 400 500
derivatives are shown in Fig. 3a. Upon initial heating, the Temperature (°C)
gradual weight loss of both materials (8–10%) proceeded
around 50–60 °C, which is probably due to the loss of physi- (b) 14 Cellulose 320 °C
448 °C

---------------------------
cally adsorbed water from cellulose and CLC. For the lat- CLC
ter material, this weight loss could also be attributed to the 12

----------------------------------------
495 °C

Heat flow(mW mg-1)


evaporation of residual solvent entrapped in the polymer 10
network from the reaction workup. Moreover, the com-
bustion profiles of the samples indicated a higher thermal 8

stability for cellulose than the crosslinked cellulose. After 6


100 °C, thermal analysis showed that cellulose and CLC
exhibit two major weight-loss stages. Cellulose showed its 4
330 °C
maximum rate of weight loss at 312 °C, whereas the effect 2
of crosslinking was apparent for CLC whose maximum rate
of weight loss was displayed at 290 °C. During this stage, 0

the cellulose and CLC were almost completely burnt out 0 100 200 300 400 500

because of the destruction of the crystalline structure and Temperature (°C)


oxidative thermal degradation. The decreased temperature
Fig. 3  a TGA responses of cellulose and CLC at a rate of 10 °C/min,
in nitrogen and their derivatives, b differential scanning calorimetry
analyses of cellulose and CLC

at which the maximum rate of weight loss was observed


indicates decreased thermal stability, which may be attrib-
uted to the degradation nature of the chemically modified
cellulose. Thus, increasing the carbon content brought about
a decrease in the decomposition temperature of crosslinked
cellulose material. Therefore, during the thermal degrada-
tion process, sebacoyl linkers could be released at much
lower temperatures. TGA data showed that some residual
materials remained for both samples after heating to the
highest temperature (600 °C). The higher weight residue
of crosslinked cellulose was probably due to the grafted
sebacoyl linkers. Similarly, on the differential scanning
calorimetry analysis (Fig. 3b), a lighter exothermic release
at about 50 °C was observed for CLC, which corresponds
to the elimination of either adsorbed water and or residual
solvent entrapped in the polymer network. A second exo-
thermic peak was observed at 320 °C for CLC. Thus, it is
very likely that the polymeric network of CLC undergoes
Fig. 2  Solid state CPMAS 13C NMR spectra of cellulose and CLC a partial fusion during the thermal decomposition, unlike

13
Chemical Papers

pure cellulose. Likewise, the exothermic peak of cellulose


observed at 330 °C was found after chemical functionali-
zation. The last exothermic peak observed at 495 °C cor-
responds to the onset of decomposition of the crosslinked
cellulose and the formation of residual char.
The morphology of the crosslinked cellulose was
assessed by scanning electron microscopy (SEM). A highly
porous structure consisting of aggregated fibrils assembled
into an interconnected structure forming a continuous three-
dimensional network (Fig. 4a, b). The images also show
pores of various sizes and many larger open cavities. The
images show irregularly distributed pores of different sizes
and shapes bound together in a three-dimensional network.
The random orientation of the pores within the polymer net-
work suggests that growth defects occur due to steric crowd-
ing within the sebacoyl linkers (Bendjelloul et al. 2017).
As shown in Fig. 4c, the TEM micrograph confirms the
Fig. 5  N2 adsorption–desorption isotherms curves measured at 77 K
mesoporous structure of CLC polymer. for CLC and Barret, Joyner, and Halenda (BJH) pore size distribution
To further confirming the inner architecture of the (inset)
crosslinked cellulose, nitrogen adsorption–desorption iso-
therms were carried out to quantify the surface area (SBET).
As shown in Fig. 5, a classical type IV isotherm with type CLC surface can be either positively or negatively charged.
H2 hysteresis is observed, suggesting the presence of PCT as a weak electrolyte (­ pKa ~ 9.52) is not dissociated
mesopores within the synthesized CLC material. below pH 9.52; therefore, the steady adsorption rate cannot
A Brunauer–Emmett–Teller (BET) surface area (SBET) of be related to electrostatic interactions between the CLC sur-
148 ­m2 ­g−1 and a pore volume of 0.70 c­ m3·g−1 are obtained face and PCT molecules. Besides, the crosslinked polymer is
for CLC. The Barret, Joyner, and Halenda (BJH) pore size nonionic and mainly composed of ester and hydroxyl polar
distribution is shown in Fig. 5 (inset). A sharp peak around groups which could play a dominant role in adsorption by
27 nm is observed, strongly suggesting the presence of abun- weak van der Waals interactions and/or hydrogen bonding
dant mesopores in the CLC structure. between these groups’ sites on surface of the adsorbent and
PCT molecules.
Effect of pH On the other hand, the sorption rate of PCT and NFA
onto CLC was investigated at pH 1.2 and 6.8 in buffer
The influence of the solution pH of PCT on adsorp- solutions. As can be seen in Fig. 6b, the results clearly
tion capacity of CLC was investigated within a wide pH indicate that the sorption efficiency for PCT is neither
range (2–10). As shown in Fig. 6a, the experimental data altered nor influenced by the presence of other cations
show clearly that the solution pH has nearly no effect on (e.g., ­Na +, ­K +). The same trend was observed for the
the amount of PCT sorbed onto CLC. The p­ Hpzc of CLC, ampholyte NFA drug where at pH 1.2, NFA is protonated
measured by the pH drift method according to the proce- (the ­pKa of the basic form is ~ 2.26) and the CLC surface
dure described by Bakatula et al. (2018), was found to be is also positively charged. Likewise, at pH 6.8, both NFA
3.12 (Fig. 6c). Under these conditions (pH range 2–10), the molecules ­( pK a ~ 4.44 for the acid form) and the CLC

(a) (b) (c)

100 nm
3.00 um 600 nm

Fig. 4  SEM and TEM images of CLC

13
Chemical Papers

80 80 Adsorption isotherms
(a) PCT (b) NFA
PCT
The sorption capacity of CLC material bearing weakly
60 60 acidic ester functional groups was investigated using par-
acetamol and niflumic acid as model pharmaceutical
% Removal

% Removal
compounds. Figure 7a shows the experimental adsorption
40 40 isotherms of PCT and NFA drugs from aqueous solutions
onto CLC polymer. Typical S-shaped curves, according to
Giles’ classification, are obtained. The initial parts of the
20 20 S curves show a slight increase in adsorption capacities at
low concentrations. However, as the concentration of drugs
in the liquid phase increased (up to 40 mg L ­ −1), adsorption
0 0
0 2 4 6 8 10 1,2 6,8
occurred more readily and the isotherms are characterized by
pH pH a very favorable shape, indicating a greater affinity of CLC
for the drugs molecules. This behavior is characteristic of
cooperative adsorption resulting in a favorable association
(c) between adsorbed molecules at the surface of the adsorbent
0,8
(Torrellas et al. 2015; Gómez et al. 2007).
First, it is noteworthy to state that the Langmuir and Fre-
undlich models did not fit adequately the experimental data
∆pH = pHi -pHf

0,4
since these models do not take into account of the adsorb-
ate–adsorbate lateral interaction. The experimental data
were analyzed with the three parameters Gu–Zhu isotherm
0,0
2 4 6 8 10
model which is a combination of Langmuir and Freundlich
pHi isotherms used for various types of S-shaped adsorption
pHPZC= 3.12
isotherms. The GZ model is expressed by Eqs. (4) and (5):
-0,4
n
Ce g
qe = q∞ KGZ n (non - linear form) (4)
Fig. 6  a Effect of pH on the sorption of PCT. b Effect of pH on sorp-
1 + KGZ Ce g
tion of PCT and NFA in buffer solutions at pH 1.2 and 6.8. Condi-
tions: initial drug concentration of 30 mg/L, adsorbent dose of 1 g/L, [ ]
qe
and reaction time of 60 min. c Determination of pHpzc of CLC by the ln = ng ln Ce + ln KGZ (linear form) (5)
pH drift method q∞ − qe

where qe (mg ­g−1) is the amount of drug adsorbed per gram


surface are negatively charged. These results confirm that of adsorbent, q∞ (mg ­g−1) is the maximum sorption capac-
electrostatic interactions are unlikely to prevail between ity per gram of adsorbent, Ce (mg L ­ −1) is the equilibrium
the CLC surface and drugs molecules under these sorption concentration of drug in solution, KGZ is the model constant,
conditions. Therefore, we hypothesize that hydrogen bond- and ng is the adsorption order.
ing between oxygen sites onto the adsorbent and adsorbate The adsorption data for PCT and NFA drugs onto CLC
may possibly play a dominant role in the adsorption pro- were analyzed by a regression analysis to fit the GZ isotherm
cess. In fact, PCT and NFA molecules possess amide and model. Since q∞ can be obtained from the adsorption iso-
amine functional groups (N–H) that are potentially capable therms plateaus (experimental qmax), the correlation of data
of interacting with oxygen atoms at the CLC surface via with the linear expression of GZ equation gave a straight line
hydrogen bonding. over the whole concentration range as shown in Fig. 7b. The
The above results suggest that the solution pH and the values of the constants ng and KGZ obtained from the slopes
ionic strength have nearly no effect on the drugs removal. and intercepts of the plots are given in Table 3. The model
Thus, for the subsequent adsorption experiments, PCT perfectly describes the adsorption isotherms with very high
solutions (having pH values around 6.8 units) were pre- R2adj and relatively low values of RMSE. Consequently, the
pared without pH adjustment, whereas NFA solutions
adsorption isotherms of PCT and NFA onto CLC can be ade-
were prepared in phosphate buffer (pH 6.8). This could be
quately described by the GZ isotherm model. On the other
advantageous for wastewater treatment applications since
hand, the average aggregation number ng was found to be 2.53
the pH of typical municipal wastewaters is in the range of
and 4.25 for PCT and NFA, respectively. The standard free
6 to 8 units.

13
Chemical Papers

(a) (b)
40
3,0

30 1,5

ln[qe/(q8-qe)]
qe (mg g-1)

20 0,0

-1,5
10
NFA
NFA PCT
PCT -3,0 Linear Gu-Zhu
0
0 10 20 30 40 50 60 70 -10,5 -10,0 -9,5 -9,0 -8,5 -8,0
-1
Ce (mg L ) ln Ce
(c) (d) -7

10 -8

θ)
θ]-[θ/(1-θ -9
θ)]
ln[θ/Ce(1-θ

9 -10
θ)/θ
ln[Ce(1-θ

-11

8 NFA -12 NFA


PCT PCT
Linear Frumkin-Fowler-Guggenheim Linear Hill-de Boer
-13
0,1 0,2 0,3 0,4 0,5 0,6 0,1 0,2 0,3 0,4 0,5 0,6
θ θ

Fig. 7  a Adsorption isotherms of PCT and NFA for CLC; Conditions: pH 6.8, initial sorbate concentration of 0–100 mg/L, adsorbent dose of
1 g/L, and contact time of 60 min. b–d Linear Gu and Zhu, FFG and HdB plots for PCT and NFA drugs removal by CLC

energy of aggregation (ΔG0) for 1 mol of pharmaceuticals was where KF is the model constant, a is the interaction param-
calculated( from)KGZ values (Table 3) by using the equation, eter, 𝜃 = qe ∕qm is the surface coverage, K1 is the model con-
ΔG0 = − 1∕ng RT ln KGZ , where ng is the number of mono- stant (L ­mol−1), K2 is the energetic constant of the interac-
mers in the pharmaceuticals aggregates. tion between adsorbed molecules (KJ m ­ ol−1), R is the gas
The adsorption of PCT and NFA onto CLC can also be constant (kJ ­mol−1 ­K−1), and T is the temperature (K).
described using Frumkin–Fowler–Guggenheim (FFG) and The plots of the Frumkin–Fowler–Guggenheim (Fig. 7c)
Hill–de Boer (HdB) isotherm models. These models also and Hill–de Boer (Fig. 7d) equations were carried out tak-
take into account lateral interaction between adsorbed mole- ing into account values of surface coverage, θ < 0.6. The
cules on the adsorbent surface. The linear forms of FFG and models fit well the data and straight lines were obtained
HdB models are expressed by Eqs. (6) and (7), respectively: with high correlation coefficients indicating that lateral
[( )( )] interaction between adsorbed species is significant for the
𝜃 1 adsorption of PCT and NFA. From Table 3, the positive
ln = ln KF + 2a𝜃 (linear form) (6)
(1 − 𝜃 Ce values obtained for parameter a (1.592 and 1.705 for PCT
and NFA, respectively) indicate a cooperative adsorption
mechanism and reflect that the lateral interactions are attrac-
[ ]
Ce (1 − 𝜃 𝜃 K
ln − = − ln K1 − 2 𝜃 (linear form) tive in nature for both drugs. PCT shows relatively weaker
𝜃 1−𝜃 RT
(7) lateral interaction than NFA. The calculated negative values

13
Chemical Papers

Table 3  Adsorption parameters for PCT and NFA sorption onto CLC Table 4  Comparison of PCTsorption capacities with other adsorbent
Isotherm Model PCT NFA Adsorbent qm (mg ­g−1) References

Gu–Zhu Rice husk ash 7.65 Thakur et al. (2020)


q∞ ­104 (mol ­g−1) 2.031a 1.442a Grape stalk 2.18 Villaescusa et al. (2011)
ng 2.53 4.25 Champignon stalks 338.1 Menk et al. (2019)
KGZ 2.78 × ­1010 1.27 × ­1018 Shiitake stalks 34.2 Menk et al. (2019)
−ΔG0(KJ ­mol−1) 19.49 19.71 Beverage sludge AC 145.4 Streit et al. (2021)
RMSE 0.174 0.196 Babassu coconut AC 71.39 Ferreira et al. (2015)
R2adj 0.990 0.979 Dende coconut AC 70.62 Ferreira et al. (2015)
Frumkin–Fowler–Guggenheim Oak acorn AC 45.45 Nourmoradi et al. (2018)
K ­(10−3) 2.835 3.102 Cotton cloth wastes AC 105 Boudrahem et al. (2017)
a 1.592 1.705 CLC 30.7 This work
−ΔG0ads(KJ ­mol−1) 19.49 19.71
RMSE 0.038 0.074
R2adj 0.992 0.975 2004; Baccar et al. 2012). The sorption studies were carried
Hill de Boer out at pH 6.8, a pH at which the surface of CLC sorbent is
K1 ­(10−3) (L ­mol−1) 2.08 2.44 negatively charged ­(pHpzc = 3.12). As discussed in “Effect of
K2 (KJ ­mol−1) 14.44 14.32 pH” section, the sorption data indicate that neither the solu-
RMSE 0.063 0.040 tion pH nor the ionic strength have a substantial effect on the
R2adj 0.994 0.996 drugs removal. Furthermore, our results demonstrated that
a
electrostatic interactions is unlikely to prevail between the
 Experimental qmax determined from the adsorption isotherms pla-
CLC surface and the molecules of paracetamol (­ pKa ~ 9.5),
teaus
which is not dissociated at this pH. On the other hand, seeing
the ­pKa value it was predictable that ampholyte NFA mol-
of ΔG0ads (− 19.49 and − 19.71 kJ ­mol−1), for PCT and NFA, ecules would be the least adsorbed due to the highest electro-
respectively, indicate the spontaneity and are in agreement static repulsive interactions between the deprotonated form
with physical adsorption process for both drugs (Cheminski and the negatively charged surface of the adsorbent. Besides,
et al. 2019; El-Aila et al. 2016). As can be seen in Table 3, the presence of a strong electron-withdrawing group such
similar trend is observed when the Hill-de Boer equation is as trifluoromethyl in the chemical structure of NFA would
used. The parameter K2, known as the interaction energy, is certainly favor aggregation of the drug molecules (4.2 mono-
positive for both PCT and NFA, indicating that the drugs mers), thus resulting in a low NFA uptake. The number of
exhibit attractive adsorbate–adsorbate interaction. On the monomers in the pharmaceuticals aggregates appeared to be
other hand, for both FFG and HdB models, the NFA adsorp- decisive in drugs uptake as lower number (n ~ 2.5) for par-
tion constant is slightly higher than those obtained for PCT, acetamol provided the higher qmax (206 ppm ­g−1) compared
which denotes the relatively better affinity of CLC for NFA. to that of niflumic acid (145 ppm ­g−1), with n ~ 4.2. This
To sum up, these results strongly suggest that the adsorption denotes that PCT has easy access to diffuse in all available
of PCT and NFA drugs onto CLC is cooperative, leading to sorption sites on the surface of the hyper-crosslinked cel-
a one-step formation of aggregates containing n molecules lulose adsorbent.
of drugs. Table 4 depicts some PCT sorption capacities
reported in the literature with other adsorbents. Maximum Thermodynamic parameters
adsorption capacity provides information on how efficient is
an adsorbent when compared to others. Although qm was not The removal of PCT and NFA by CLC sorbent was stud-
obtained under the same experimental conditions, one can ied in the temperature range (298–318 K) to determine the
realize that the efficiency obtained with CLC material is sig- thermodynamic parameters, and the results are summarized
nificantly lower if compared with activated carbons. On the in Table 5. The drugs uptake decreased by increasing the
other hand, it can also be said that the adsorption capacity temperature, indicating an exothermic process. The thermo-
of CLC is more or less better than those of other adsorbents. dynamic parameters ΔG0 , ΔH 0 , and ΔS0 were calculated by
It is well established that parameters such as molecular using the following equation (Zhu et al. 2010):
size, water solubility, pKa, and the nature of the substituents
ΔG0 ΔS0 ΔH 0
are among the factors that mainly affect the adsorption pro- log K = − = − (8)
RT R RT
cess of aromatic hydrocarbons molecules (Moreno-Castilla

13
Chemical Papers

Table 5  Thermodynamic parameters for PCT and NFA sorption onto (qe, exp). On the other hand, the experimental data were well
CLC fitted to the pseudo-second-order model. The plots of t/qt
ΔH0 (KJ m
­ ol–1) ΔS0 (J ­K–1 ­mol–1) ΔG0 (KJ ­mol–1) versus t (Fig. 8b) gave excellent results with connection to
the criteria of R2adj and RMSE (Table 6), which confirm the
298 K 308 K 318 K
applicability of the pseudo-second-order equation. In full
PCT –24.76 –20.46 –18.67 –18.45 –18.25 agreement with the experimental data, the value of the
NFA –21.13 8.30 –23.60 –23.69 –23.77 pseudo-second-order rate constant (KS) is also more than
three times greater for the adsorption of PCT than that of
NFA. The half-life time (t1/2) and the initial rate of adsorp-
where T is the temperature (K), R is the ideal gas constant tion (h) parameters estimated according to the equations
(8.314 J ­mol−1 ­K−1), and K is the equilibrium constant (L t1∕2 = 1∕KS qe and h = KS q2e (Ho 2006; Tsai et al. 2006),
­mg−1). where qe is the total amount adsorbed, have revealed that the
The vant’Hoff plot of ln K versus 1/T gave straight lines. elapsed time for half of the PCT adsorption (1.15  min)
The calculated slope and intercept from the plot were used to increased by fourfold (4.71 min) for the adsorption of NFA.
determine ΔH 0 and ΔS0 , respectively (Table 5). The negative The same trend is noticed as the initial rates of adsorption
value of ΔG0 (− 23.77 to − 18.25 kJ·mol−1) at each tempera- were found to be 15.98 and 3.48 mg ­g−1 ­min−1 for PCT and
ture implies a favorable and spontaneous adsorption process NFA, respectively.
and confirms the affinity of CLC material for PCT and NFA Thus, in summary, it can be deduced that even though the
drugs. These values are in accordance with those found ear- mean diameter of the drugs molecules (1.16 and 1.35 nm for
lier with GZ and FFG models (“Adsorption isotherms” sec- PCT and NFA, respectively) is of the same order of magni-
tion). In good agreement with the FFG and HdB models, the tude, the sorption of aggregates containing ~ 4.25 NFA mol-
affinity of CLC adsorbent towards NFA is again confirmed ecules caused a deceleration in its removal rate compared to
by the positive value of ΔS0 , which indicates an increase in PCT which forms lower aggregates (n ~ 2.5). Although most
the degree of freedom of the adsorbed NFA molecules and of the pores in CLC are in the size range of 27 nm, which are
also suggests the increased randomness at the solid/solution large enough to fully support the different arrangements of
interface with some structural changes in the adsorbate and both PCT and NFA aggregates, the adsorbent showed how-
the adsorbent. The negative values of Δ H 0 indicate that the ever better sorption efficiency towards PCT. This could be
adsorption is exothermic and also suggest that the sorption explained by the fact that the NFA molecules might just be
process is a physical adsorption. clustered on the side of the pores while the PCT molecules
are spread out inside the pores of the CLC adsorbent.
Kinetics of PCT and NFA removal by CLC

The effect of contact time on the removal of drugs by CLC Analysis of intraparticle diffusivity mechanisms
is shown in Fig. 8a. The results revealed that the initial con-
centration decreased rapidly with time, indicating the avail- Sorption processes of the drugs onto functionalized CLC
ability of plenty of readily accessible sorption sites. The material have been further examined by intraparticle diffu-
process was uniform which suggests strong interactions sion model. Figure 8c shows the plots of the Weber–Morris
between pharmaceuticals and the crosslinked cellulose. equation for the experimental data of the drugs sorption.
Maximum PCT was sequestered from the solution within The plot for NFA adsorption gives a straight line over the
30 min of contact time while for NFA, the equilibrium point whole sorption period suggesting that the sorption process
was reached only after 50 min. After that, the concentration is controlled by intraparticle diffusion. However, deviation
of pharmaceuticals in the liquid phase remained nearly con- of the plot from the origin suggests the contribution from
stant. This was due to the less active sites being available film diffusion process (Ru-Ling et al. 2003), whereas for
on CLC surface at the end of the sorption process. Sixty PCT, the data exhibit two linear plots indicating the simul-
minutes proved to be enough to reach adsorption equilibrium taneous occurrence of two adsorption stages. The first
for both PCT and NFA drugs. straight portion illustrates external pore diffusion and the
The experimental data were fitted to the Lagergren model, second which is attributed to the final equilibrium stage,
but the linearity of log(qe − qt) versus t plots is seen only for and where internal pore diffusion may possibly occur (Val-
the initial adsorption period, in the region where rapid derrama et al. 2007). The slopes of the intraparticle diffu-
adsorption took place (0–25 min for PCT and 0–35 min for sion plots give the value of kid and extrapolation of the plots
NFA), and thereafter it deviates from theory. In addition, the back to the y-axis gives the intercepts, which provide the
calculated adsorption capacity values (qe, cal) from Lager- measure of the boundary layer thickness. The order of sorp-
gren model are not in agreement with the experimental ones tion rate for PCT was in the first stage (kid1) higher than the

13
Chemical Papers

(a) (b)
4
20

3
15

t/qt (min mg g-1)


qt (mg g-1)

10 2

5 NFA 1
PCT
PSO NFA
PFO PCT
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60 70
Time (min.) Time (min)
(c) (d)
20 5

15
3
qt(mg g-1)

Bt

10
1

NFA NFA
0 PCT
PCT
5
1 2 3 4 5 6 7 8 0 10 20 30 40 50

Time 0.5
(min )0.5
Time (min)

Fig. 8  a Adsorption kinetics of PCT and NFA onto CLC; Conditions: ond-order kinetic model. c Intraparticle diffusion plots for the adsorp-
initial drug concentration of 30  mg/L, adsorption dose of 1  g/L, pH tion of PCT and NFA onto CLC. d B(t) versus t plots of PCT and
6.8, contact time of 60 min. b Kinetic data fitted with the pseudo-sec- NFA adsorption onto CLC of bead size 180–250 μm

Table 6  Kinetic data of paracetamol and niflumic acid adsorption second (kid2) with values of 1.68 and 0.30 mg ­g−1 ­min−1/2,
onto CLC fitted to the pseudo-second order and the pseudo-first order respectively. As mentioned before, NFA sorption proceeded
Adsorption kinetic models Non linear Linear merely in a single step and provided a sorption rate value
PCT NFA PCT NFA of 0.96 mg ­g−1 ­min−1/2. Therefore, at the beginning of the
sorption process whether for PCT or NFA, the drugs are
Pseudo-second-order
sorbed by the external pores available on the surface of the
qe (mg ­g−1) 19.21 15.91 18.63 16.70
exterior structure of CLC adsorbent so the sorption rates
KS (g mg ­min−1) 0.029 0.020 0.047 0.013
were very high. When the saturation of these sites is reached,
RMSE 0.051 0.166 0.009 0.060
and taking into account the molecular size along with the
R2adj 0.999 0.999 0.999 0.996
aggregates arrangement of the drugs, only PCT molecules
Pseudo-first-order (smaller size) migrate to the inner tridimensional network
qe (mg ­g−1) 17.95 14.55 9.99 5.81 and diffuse in the internal surface pores and then are sorbed
Kf ­(min−1) 0.289 0.195 0.171 0.043 at the internal surface of the crosslinked polymer. The dif-
RMSE 0.118 0.286 0.045 0.028 fusion resistance increased gradually as PCT molecules dif-
R2adj 0.996 0.998 0.990 0.968 fuse in the inner pores, thus resulting in a decrease in the

13
Chemical Papers

diffusion rate until the equilibrium is reached. Therefore, ln (1 − F) = −kfd t (10)


the discrepancy between kid1 and kid2values is attributed to
the sorption stages of the external and internal surfaces. In where F = qt/qe, is the fractional attainment of equilibrium
summary, these results demonstrate that the sorption process and kfd is the film diffusion rate constant.
of PCT and NFA which are featured by faster rate kid1 is A linear plot of ln(1 − F) versus t passing through the
undoubtedly controlled by intraparticle diffusion and to a origin suggests that the adsorption kinetics is controlled by
lesser extent by film diffusion. diffusion through liquid film surrounding the solid adsor-
The sorption data were further analyzed by the kinetic bent. Although good straight lines could be obtained for the
expression for particle diffusion written as: film diffusion model, they do not pass through the origin,
∞ 2
however (Fig. 9).
qt 6 ∑ e−n Bt The results of the linear regression analysis for PCT and
F= =1− 2 (9)
q∞ 𝜋 n=1 n2 NFA drugs are summarized in Table 7. The intercept values
for both PCT (0.598) and NFA (0.903) are higher than zero,
where F is the fractional attainment of equilibrium; B is a which suggest that the process is not exclusively controlled
constant ( B = 𝜋 2 Di ∕r2 ), where Di is the effective diffusion by film diffusion and that the process may be impeded by the
coefficient and r is the radius of the particle assuming that aggregation of drugs. Furthermore, the liquid film diffusion
all the adsorbent particles are uniform spheres. rate constant (kfd) values show that PCT (0.17) adsorption
From the experimental F values, the corresponding Bt was faster than that of NFA (0.049) which can also be related
values were obtained from Reichenberg’s table and were to the number of aggregates of these drugs. In summary, the
then plotted against the experimental values of t. As shown global adsorption rate of PCT and NFA onto CLC seemed
in Fig. 8d, the Bt versus t plot for NFA gave a relatively good to be controlled by intraparticle diffusion simultaneously in
straight line over the whole range but does not pass through external and internal pores of the hyper-crosslinked cellu-
the origin, indicating that the rate-controlling process may lose adsorbent. It appears also that the low number of drugs
possibly be film diffusion. However, the prevalence of intra- aggregates promotes the diffusion and migration of PCT
particle diffusion in the NFA adsorption process is quite molecules into the internal surface pores of CLC. This was
feasible given the intercept value (0.429) of the plot which
passes close to the origin. As previously expected for PCT,
the plot did indeed provide a good straight line passing 5
through the origin, thus indicating that particle diffusion is
the rate controlling step. The R2adj coefficients and RMSE
4
values indicate a good fit for both drugs. The values of effec-
tive diffusion coefficients (Di) were calculated from the slope
(B value) and are reported in Table 7. The average radius of 3
ln(1-F)

the particle was estimated to be 0.0115 cm as the CLC par-


ticles were in 210–250 μm size range. The calculated Di
2
coefficients of the drugs molecules involved in the sorption
process overlap with those reported in the literature for
organic compounds sorbed by macroporous and hyper- 1 NFA
PCT
crosslinked polymeric sorbents (Valderrama et al. 2007;
Linear liquid film diffusion model
Azanova and Hradil 1999).
0
0 5 10 15 20 25 30 35 40 45
Liquid film diffusion model Time (min)

If liquid film diffusion controls the rate of sorption, the fol-


Fig. 9  Linear liquid film diffusion plots for the sorption of PCT and
lowing expression [49] can be used: NFA onto CLC

Table 7  Kinetic data of Intraparticle diffusion model Liquid film diffusion model


paracetamol and niflumic acid
−1 2 −1
adsorption onto CLC fitted kid (mg A (mg g­ ) RMSE R2adj Di ­(m ­s ) kfd Intercept RMSE R2adj
to the intraparticle diffusion ­g−1 ­min−1/2)
models
PCT 1.68 10.61 0.308 0.953 2.25 ­10−14 0.598 0.171 0.103 0.990
NFA 0.96 8.89 0.200 0.980 6.27 ­10−15 0.903 0.049 0.117 0.950

13
Chemical Papers

cross-checked by desorption assays of the drugs saturated study. Langmuir 33:11146–11155. https://​doi.​org/​10.​1021/​acs.​
CLC. Thus, NFA was entirely recovered by vigorous stir- langm​uir.​7b019​67
Bakatula EN, Richard D, Neculita CM, Zagury GJ (2018) Deter-
ring with pH 1.2 aqueous solution for 30 min, while 90% of mination of point of zero charge of natural organic materials.
adsorbed PCT was recovered after the first wash with pH 4 Environ Sci Pollut Res 25:7823–7833. https://​doi.​org/​10.​1007/​
aqueous solution, reaching complete recovery only after a s11356-​017-​1115-7
subsequent wash. Bendjelloul M, Elandaloussi EH, de Ménorval LC, Bentouami A
(2017) Quaternized triethanolamine-sebacoyl moieties in highly
branched polymer architecture as a host for the entrapment of acid
dyes in aqueous solutions. J Water Reuse Desalin 7:53–65. https://​
doi.​org/​10.​2166/​wrd.​2016.​191
Conclusions Bhatnagar A, Sillanpää M, Witek-Krowiak A (2015) Agricultural waste
peels as versatile biomass for water purification—a review. Chem
The presented work extends the applicability of a novel Eng J 270:244–271. https://​doi.​org/​10.​1016/j.​cej.​2015.​01.​135
crosslinked cellulose adsorbent to investigate adsorption Boudrahem N, Delpeux-Ouldriane S, Khenniche L, Boudrahem F,
Aissani-Benissad F, Gineys M (2017) Single and mixture adsorp-
processes of pharmaceuticals in aqueous media. Based on tion of clofibric acid, tetracycline and paracetamol onto activated
the equilibrium adsorption studies, it was found that the sur- carbon developed from cotton cloth residue. Process Saf Environ
face interaction of paracetamol and niflumic acid drugs can Prot 111:544–559. https://​doi.​org/​10.​1016/j.​psep.​2017.​08.​025
be well described by the GZ, FFG and HdB models, indicat- Box KJ, Comer JEA (2008) Using measured pKa, LogP and solubility
to investigate supersaturation and predict BCS class. Curr Drug
ing that lateral interactions is dominant for the adsorption Metab 9:869–878. https://​doi.​org/​10.​2174/​13892​00087​86485​155
of these species onto the CLC surface. Both drugs exhibit Box KJ, Völgyi G, Baka E, Stuart M, Takács-Novák K, Comer JEA
attractive lateral interactions and their sorption onto CLC (2006) Equilibrium versus kinetic measurements of aqueous solu-
are cooperative, leading to the formation of aggregates of bility, and the ability of compounds to supersaturate in solution—
a validation study. J Pharm Sci 95:1298–1307. https://​doi.​org/​10.​
n monomers. The lesser n value is the better for the drugs 1002/​jps.​20613
uptake, so paracetamol shows a higher uptake unlike nif- Cheminski T, de Figueiredo NT, Silva PM, Guimarães CH, Prediger P
lumic acid, which presents higher n. Likewise, the global (2019) Insertion of phenyl ethyleneglycol units on graphene oxide
adsorption rate of NFA is decelerated as a result of its higher as stabilizers and its application for surfactant removal. J Environ
Chem Eng 7:102976. https://​doi.​org/​10.​1016/j.​jece.​2019.​102976
number of monomers aggregates. Regardless the huge size Ciğeroğlu Z, Küçükyıldız G, Erim B, Alp E (2021) Easy preparation of
of the pores in the crosslinked cellulose, it appears again magnetic nanoparticles-rGO-chitosan composite beads: optimiza-
that n equally controls the intraparticle diffusion step so that tion study on cefixime removal based on RSM and ANN by using
NFA molecules are clustered on the side of the pores, while Genetic Algorithm Approach. J Mol Struct 1224:129182. https://​
doi.​org/​10.​1016/j.​molst​ruc.​2020.​129182
the PCT molecules are spread out inside the pores of the Coelho AD, Sans C, Esplugas S, Dezotti M (2010) Ozonation of
CLC adsorbent. This was further confirmed by desorption NSAID: a biodegradability and toxicity study. Ozone Sci Eng
tests where the NFA molecules were totally recovered from 32:91–98. https://​doi.​org/​10.​1080/​01919​51090​35081​62
the first wash while PCT required two successive washes El-Aila HJ, Elsousy KM, Hartany KA (2016) Kinetics, equilibrium,
and isotherm of the adsorption of cyanide by MDFSD. Arab J
for full recovery. Chem 9:S198–S203. https://d​ oi.o​ rg/1​ 0.1​ 016/j.a​ rabjc.2​ 011.0​ 3.0​ 02
Escapa C, Coimbra RN, Paniagua S, García AI, Otero M (2017) Par-
acetamol and salicylic acid removal from contaminated water by
microalgae. J Environ Manage 203:799–806. https://​doi.​org/​10.​
References 1016/j.​jenvm​an.​2016.​06.​051
Ferreira RC, de Lima HHC, Candido AA, Couto Junior OM, Arroyo
Alygizakis NA, Gago-Ferrero P, Borova VL, Pavlidou A, Hatzianestis PA, de Carvalho KQ, Gauze GF, Barros MASD (2015) Adsorp-
I, Thomaidis NS (2016) Occurrence and spatial distribution of tion of paracetamol using activated carbon of dende and babassu
158 pharmaceuticals, drugs of abuse and related metabolites in coconut mesocarp. Int J Biol Biomol Agric Food Biotechnol Eng
offshore seawater. Sci Total Environ 541:1097–1105. https://​doi.​ 9:575–580
org/​10.​1016/j.​scito​tenv.​2015.​09.​145 Foston MB, Hubbell CA, Ragauskas AJ (2011) Cellulose isolation
Azanova VV, Hradil J (1999) Sorption properties of macroporous and methodology for NMR analysis of cellulose ultrastructure. Materi-
hypercrosslinked copolymers. React Funct Polym 41:163–175. als 4:1985–2002. https://​doi.​org/​10.​3390/​ma411​1985
https://​doi.​org/​10.​1016/​S1381-​5148(99)​00029-2 Franca MD, Santos LM, Silva TA, Borges KA, Silva VM, Patrocinio
Babaei AA, Alavi SN, Akbarifar M, Ahmadi K, Esfahani AR, Kaka- AOT, Trovo AG, Machado AEH (2016) Efficient mineralization of
vandi B (2016) Experimental and modeling study on adsorption of paracetamol using the nanocomposite ­TiO2/Zn(II) phthalocyanine
cationic methylene blue dye onto mesoporous biochars prepared as photocatalyst. J Braz Chem Soc 27:1094–1102. https://​doi.​org/​
from agrowaste. Desalin Water Treat 57:27199–27212. https://d​ oi.​ 10.​5935/​0103-​5053.​20160​007
org/​10.​1080/​19443​994.​2016.​11637​36 Gómez V, Larrechi MS, Callao MP (2007) Kinetic and adsorption
Baccar R, Sarrà M, Bouzid J, Feki M, Blánquez P (2012) Removal of study of acid dye removal using activated carbon. Chemosphere
pharmaceutical compounds by activated carbon prepared from 69:1151–1158. https://​doi.​org/​10.​1016/j.​chemo​sphere.​2007.​03.​
agricultural by-product. Chem Eng J 211–212:310–317. https://​ 076
doi.​org/​10.​1016/j.​cej.​2012.​09.​099 Ho YS (2006) Review of second-order models for adsorption systems.
Bahamon D, Vega LF (2017) Pharmaceutical removal from water efflu- J Hazard Mater B136:681–689. https://​doi.​org/​10.​1016/j.​jhazm​
ents by adsorption on activated carbons: a monte carlo simulation at.​2005.​12.​043

13
Chemical Papers

Hokkanen S, Bhatnagar A, Sillanpää M (2016) A review on modifica- Robberson KA, Waghe AB, Sabatini DA, Butler EC (2006) Adsorption
tion methods to cellulose-based adsorbents to improve adsorption of the quinolone antibiotic nalidixic acid onto anion-exchange and
capacity. Water Res 91:156–173. https://​doi.​org/​10.​1016/j.​watres.​ neutral polymers. Chemosphere 63:934–941. https://​doi.​org/​10.​
2016.​01.​008 1016/j.​chemo​sphere.​2005.​09.​047
Ibanez M, Borova V, Boix C, Aalizadeh R, Bade R, Thomaidis NS, Her- Ru-Ling T, Feng-Chin W, Ruey-Shin J (2003) Liquid-phase adsorption of
nandez F (2016) UHPLC-QTOF MS screening of pharmaceuticals and dyes and phenols using pinewood-based activated carbons. Carbon
their metabolites in treated wastewater samples from Athens. J Hazard 41:487–495. https://​doi.​org/​10.​1016/​S0008-​6223(02)​00367-6
Mater 323:26–35. https://​doi.​org/​10.​1016/j.​jhazm​at.​2016.​03.​078 Selkälä T, Suopajärvi T, Sirviö JA, Luukkonen T, Lorite GS, Kalliola S,
Jandura P, Kokta BV, Riedl B (2000) Fibrous long-chain organic acid Sillanpää M, Liimatainen H (2018) Rapid uptake of pharmaceutical
cellulose esters and their characterization by diffuse reflectance salbutamol from aqueous solutions with anionic cellulose nanofi-
FTIR spectroscopy, solid-state CP/MAS 13C-NMR, and X-ray dif- brils: the importance of pH and colloidal stability in the interaction
fraction. J Appl Polym Sci 78:1354–1365. https://​doi.​org/​10.​1002/​ with ionizable pollutants. Chem Eng J 350:378–385. https://​doi.​org/​
1097-​4628(20001​114)​78:7%​3c135​4::​AID-​APP60%​3e3.0.​CO;2-V 10.​1016/j.​cej.​2018.​05.​163
Ji H, Xiang Z, Qi H, Han T, Pranovich A, Song T (2019) Strategy towards Sim WJ, Lee JW, Lee ES, Shin SK, Hwang SR, Oh JE (2011) Occurrence
one-step preparation of carboxylic cellulose nanocrystals and nanofi- and distribution of pharmaceuticals in wastewater from households,
brils with high yield, carboxylation and highly stable dispersibility livestock farms, hospitals and pharmaceutical manufactures. Che-
using innocuous citric acid. Green Chem 21:1956–1964. https://​doi.​ mosphere 82:179–186. https://​doi.​org/​10.​1016/j.​chemo​sphere.​2010.​
org/​10.​1039/​C8GC0​3493A 10.​026
Joss A, Zabczynski S, Göbel A, Hoffmann B, Löffler D, McArdell CS, Streit AFM, Collazzo GC, Druzian SP, Verdi RS, Foletto EL, Oliveira
Ternes TA, Thomsen A, Siegrist H (2006) Biological degradation LFS, Dotto GL (2021) Adsorption of ibuprofen, ketoprofen, and
of pharmaceuticals in municipal wastewater treatment: Proposing a paracetamol onto activated carbon prepared from effluent treatment
classification scheme. Water Res 40:1686–1696. https://​doi.​org/​10.​ plant sludge of the beverage industry. Chemosphere 262:128322.
1016/j.​watres.​2006.​02.​014 https://​doi.​org/​10.​1016/j.​chemo​sphere.​2020.​128322
Larsson PT, Westermark U, Iversen T (1995) Determination of the cel- Thakur A, Sharma N, Mann A (2020) Removal of ofloxacin hydrochlo-
lulose Iα allomorph content in a tunicate cellulose by CP/MAS 13C- ride and paracetamol from aqueous solutions: binary mixtures and
NMR spectroscopy. Carbohydr Res 278:339–343. https://​doi.​org/​10.​ competitive adsorption. Mater Today Proc 28:1514–1519. https://​
1016/​0008-​6215(95)​00248-0 doi.​org/​10.​1016/j.​matpr.​2020.​04.​833
Larsson DGJ, de Pedro C, Paxeus N (2007) Effluent from drug manufac- Torrellas SA, Lovera RG, Escalona N, Sepúlveda C, Sotelo JL, García
tures contains extremely high levels of pharmaceuticals. J Hazard J (2015) Chemical-activated carbons from peach stones for the
Mater 148:751–755. https://​doi.​org/​10.​1016/j.​jhazm​at.​2007.​07.​008 adsorption of emerging contaminants in aqueous solutions. Chem
Li B, Pan Y, Zhang Q, Huang Z, Liu J, Xiao H (2019a) Porous cellulose Eng J 279:788–798. https://​doi.​org/​10.​1016/j.​cej.​2015.​05.​104
beads reconstituted from ionic liquid for adsorption of heavy metal Tsai WT, Lai CW, Su TY (2006) Adsorption of bisphenol-A from
ions from aqueous solutions. Cellulose 26:9163–9178. https://​doi.​ aqueous solution onto minerals and carbon adsorbents. J Haz Mat
org/​10.​1007/​s10570-​019-​02687-4 B134:169–175. https://​doi.​org/​10.​1016/j.​jhazm​at.​2005.​10.​055
Li Y, Xiao H, Pan Y, Zhang M, Jin Y (2019b) Thermal and pH dual- Valderrama C, Gamisans X, de las Heras FX, Cortina JL, Farrán A (2007)
responsive cellulose microfilament spheres for dye removal in single Kinetics of polycyclic aromatic hydrocarbons removal using hyper-
and binary systems. J Hazard Mater 377:88–97. https://​doi.​org/​10.​ cross-linked polymeric sorbents Macronet Hypersol MN200. React
1016/j.​jhazm​at.​2019.​05.​033 Funct Polym 67:1515–1529. https://​doi.​org/​10.​1016/j.​react​funct​
Menk JJ, Nascimento AIS, Leite FG, Oliveira RA, Jozala AF, Oliveira polym.​2007.​07.​020
JM Jr, Chaud MV, Grotto D (2019) Biosorption of pharmaceuti- Villaescusa I, Fiol N, Poch J, Bianchi A, Bazzicalupi C (2011) Mecha-
cal products by mushroom stem waste. Chemosphere 237:124515. nism of paracetamol removal by vegetable wastes: The contribu-
https://​doi.​org/​10.​1016/j.​chemo​sphere.​2019.​124515 tion of π–π interactions, hydrogen bonding and hydrophobic effect.
Miege C, Choubert JM, Ribeiro L, Eusebe M, Coquery M (2009) Fate Desalination 270:135–142. https://​doi.​org/​10.​1016/j.​desal.​2010.​11.​
of pharmaceuticals and personal care products in wastewater treat- 037
ment plants-conception of a database and first results. Envir Pollut Wang D (2019) A critical review of cellulose-based nanomaterials for
157:1721–1726. https://​doi.​org/​10.​1016/j.​envpol.​2008.​11.​045 water purification in industrial processes. Cellulose 26:687–701.
Mila E, Nika MC, Thomaidis NS (2019) Identification of first and sec- https://​doi.​org/​10.​1007/​s10570-​018-​2143-2
ond generation ozonation transformation products of niflumic acid Xiong P, Hu J (2017) Decomposition of acetaminophen (Ace) using
by LC-QToF-MS. J Hazard Mater 365:804–812. https://​doi.​org/​10.​ ­TiO2/UVA/LED system. Catal Today 282:48–56. https://​doi.​org/​
1016/j.​jhazm​at.​2018.​11.​046 10.​1016/j.​cattod.​2016.​03.​015
Moreno-Castilla C (2004) Adsorption of organic molecules from aqueous Zhou Y, Min Y, Qiao H, Huang Q, Wang E, Ma T (2015) Improved
solutions on carbon materials. Carbon 42:83–94. https://​doi.​org/​10.​ removal of malachite green from aqueous solution using chemically
1016/j.​carbon.​2003.​09.​022 modified cellulose by anhydride. Int J Biol Macromol 74:271–277.
Nourmoradi H, Moghadam KF, Jafari A, Kamarehie B (2018) Removal https://​doi.​org/​10.​1016/j.​ijbio​mac.​2014.​12.​020
of acetaminophen and ibuprofen from aqueous solutions by acti- Zhu HY, Jiang R, Xiao L, Zeng GM (2010) Preparation, characteriza-
vated carbon derived from Quercus Brantii (Oak) acorn as a low- tion, adsorption kinetics and thermodynamics of novel magnetic
cost biosorbent. J Environ Chem Eng 6:6807–6815. https://​doi.​org/​ chitosan enwrapping nanosized γ-Fe2O3 and multi-walled carbon
10.​1016/j.​jece.​2018.​10.​047 nanotubes with enhanced adsorption properties for methyl orange.
Petrie B, Barden R, Kasprzyk-Hordern B (2015) A review on emerging Bioresour Technol 101:5063–5069. https://​doi.​org/​10.​1016/j.​biort​
contaminants in wastewaters and the environment: current knowledge, ech.​2010.​01.​107
understudied areas and recommendations for future monitoring. Water
Res 72:3–27. https://​doi.​org/​10.​1016/j.​watres.​2014.​08.​053 Publisher’s Note Springer Nature remains neutral with regard to
Rivera-Utrilla J, Sánchez-Polo M, Ferro-García MÁ, Prados-Joya G, jurisdictional claims in published maps and institutional affiliations.
Ocampo-Pérez R (2013) Pharmaceuticals as emerging contaminants
and their removal from water. A review. Chemosphere 93:1268–
1287. https://​doi.​org/​10.​1016/j.​chemo​sphere.​2013.​07.​059

13

You might also like