Environmental friendly technology for the removal of pharmaceutical contaminants from wastewaters using modified chitosan adsorbents

Accepted Manuscript Environmental friendly technology for the removal of pharmaceutical contam‐ inants from wastewaters using modified chitosan adsorbents George Z. Kyzas, Margaritis Kostoglou, Nikolaos K. Lazaridis, Dimitra A. Lambropoulou, Dimitrios N. Bikiaris PII: DOI: Reference:

S1385-8947(13)00214-3 http://dx.doi.org/10.1016/j.cej.2013.02.048 CEJ 10399

To appear in:

Chemical Engineering Journal

Received Date: Revised Date: Accepted Date:

7 January 2013 14 February 2013 16 February 2013

Please cite this article as: G.Z. Kyzas, M. Kostoglou, N.K. Lazaridis, D.A. Lambropoulou, D.N. Bikiaris, Environmental friendly technology for the removal of pharmaceutical contaminants from wastewaters using modified chitosan adsorbents, Chemical Engineering Journal (2013), doi: http://dx.doi.org/10.1016/j.cej. 2013.02.048

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Environmental friendly technology for the removal of pharmaceutical contaminants from wastewaters using modified chitosan adsorbents

George Z. Kyzas1, Margaritis Kostoglou1, Nikolaos K. Lazaridis1,*, Dimitra A. Lambropoulou2, Dimitrios N. Bikiaris3,

1

Laboratory of General & Inorganic Chemical Technology, Department of Chemistry, Aristotle University

of Thessaloniki, GR–541 24 Thessaloniki, Greece 2

Laboratory of Environmental Pollution Control, Department of Chemistry, Aristotle University of

Thessaloniki, GR–541 24 Thessaloniki, Greece 3

Laboratory of Polymer Chemistry and Technology, Department of Chemistry, Aristotle University of

Thessaloniki, GR–541 24 Thessaloniki, Greece

Corresponding author. Tel.: +30 2310 997812; fax: +30 2310 997859. E–mail address: [email protected] (D.N. Bikiaris). Corresponding author. Tel.: +30 2310 997807; fax: +30 2310 997859. E–mail address: [email protected] (N.K. Lazaridis). 1

ABSTRACT In the present study, new modified chitosans, cross–linked with glutaraldehyde and grafted with sulfonate (CsSLF) or N–(2–carboxybenzyl) groups (CsNCB), were synthesized and investigated as efficient and environmental friendly adsorbents for removing pharmaceuticals from polluted water matrices. To test the performance of these innovative chitosan–based sorbents, pramipexole dihydrochloride (PRM), a recently available non–ergot dopamine agonist was selected as model compound. Non–grafted chitosan was also prepared and used as reference adsorbent material for comparison with cross–linked (Cs) chitosan. Their characterization was realized via swelling tests, FTIR, SEM, and BET analysis. Alkaline conditions (pH=10) were found to be the optimum for the adsorption process, while the reverse conditions (pH=2) were optimum for desorption. The adsorption mechanism was also examined. Kinetic experiments were performed to study the effect of contact time on adsorption and the experimental findings were fitted to common kinetic models (pseudo–first, –second order and Elovich equations), in line with a detailed kinetic model, incorporating diffusion and localized adsorption and desorption steps. Isotherms showed the effect of initial PRM concentration and temperature (25, 45, 65 °C) on adsorption. Modified chitosans have a better behavior for PRM adsorption and the rate follows the order: CsSLF>CsNCB>Cs. The equilibrium data were fitted to the Langmuir–Freundlich model. The reuse of adsorbents synthesized was evaluated via sequential adsorption–desorption cycles.

Keywords: Chitosan; Environmental friendly adsorbent; Grafting; Modeling; Pharmaceutical wastewaters; Pramipexole.

2

1. Introduction Pharmaceuticals are of scientific and public concern as newly recognized classes of environmental pollutants and are receiving considerable attention with respect to their environmental fate and toxicological properties over the last 15 years [1,2]. Many of these pharmaceutical compounds, which often are pharmacologically active or endocrine modulating across multiple levels of biological organization, are not completely removed by wastewater treatment plants (WWTPs) and municipal effluents as well as effluents from hospitals and pharmaceutical manufacturing facilities have been identified as important sources. Consequently, a vast number of these compounds have been detected in WWTP effluents, surface waters and, less frequently, in ground and drinking water all over the world [3-8]. The above considerations reflect the need for the complete removal of pharmaceuticals and their transformation products (TPs) from aquatic systems to avoid their potential toxicity and other possible dangerous health effects. As conventional water and wastewater treatment processes are unable to act as a reliable barrier towards some recalcitrant pharmaceutical compounds, it is necessary to introduce and apply specific treatment methodologies for pharmaceutical wastewaters, especially for those generated from pharmaceutical industry in which large volumes of wastewater are produced with specific nature and high pharmaceutical loading. Removal of pharmaceuticals by adsorption is one of the most promising techniques, due to its convenience once applied into current water treatment processes. Up to now, activated carbon [9-12] and other inorganic materials such as zeolites [13,14] are among the most commonly used effective adsorptive materials that have been tested for the treatment of wastewaters. Despite its high efficiency, several problems occur by using these adsorbents such as stability and recycling which requires long and costly regeneration procedures. To overcome these limitations, a great deal of effort has been devoted to synthesize novel, effective and environmentally friendly adsorbents based on low–cost and natural polymeric materials.

3

Chitosan (poly– –(1 4)–2–amino–2–deoxy–D–glucose) is the deacetylated derivative of chitin (the natural polymer extracted from shrimp or crab shells) [15] and the second most plentiful natural biopolymer. Owing to its biocompatibility, high biodegradability, non–toxicity and adsorption properties, chitosan has been proved as a promising environmental friendly adsorbent material and has been extensively used for the removal of dyes [16] and heavy metals [17,18] from aqueous solutions. However, to the best of our knowledge, studies and applications on adsorption of emerging organic contaminants such as pharmaceutical compounds on synthesized cross–linked chitosan materials from aqueous media such as treated wastewaters are extremely limited [19,20]. Moreover, although a few specialized reports are available on the adsorption of pharmaceutically active substances by using new synthetic zeolites and mesoporous materials [9-11], it is worth to point out that such synthesized adsorbents have relatively high preparation cost. Therefore, further efforts should be made to develop more effective and low–cost adsorbents that can achieve higher efficacy and better selectivity in the removal of pharmaceuticals from a variety of aqueous matrices, in order to demonstrate adsorption as effective and environmental friendly treatment technology. Bearing in mind the aforementioned information, after the successful synthesis and application of some modified (grafted or/and cross–linked) forms of chitosan for organic (dyes) [21-25] and inorganic species (ions) [26,27] by our research team, the purpose of this study is to investigate the possible use of new modified chitosan based materials as effective, low–cost and environmental friendly adsorbents for removal of pharmaceutical compounds from environmental wastewaters. Pramipexole dihydrochloride ((6S)–N6–propyl–4,5,6,7–tetrahydro–1,3–benzothiazole–2,6–diamine,

PRM),

a

novel

non–ergoline

dopamine agonist, was selected as model compound. PMR initially introduced for treatment of the signs and symptoms of idiopathic Parkinson’s disease [28] and recently approved in US and Europe also for the treatment of idiopathic restless legs syndrome in adults [29]. It is being used widely all over the world for its unique pharmaceutical activity and on the basis of recent drug usage trends [30]; it seems likely that the use of this recently available non–ergot dopamine agonist will continue to increase in the immediate future, 4

as primary care physicians (PCP) become more familiar with it. Some pharmaceutical industries and hospitals are discharging PRM in their effluents resulting into the contamination of our natural water resources. Therefore, treatment of wastewater by high polluted levels of PRM is required and urgent needed. Given the cationic nature of PRM, the derivatives of chitosan prepared were grafted with anionic groups (sulfonate and N–(2–carboxybenzyl)) to enhance their adsorption capacity and cross–linked (glutaraldehyde) to improve their resistance in extreme alkaline or acidic conditions. Non–grafted chitosan was also prepared and used as reference adsorbent material for comparison with cross–linked (Cs) chitosan. Their characterization was realized via swelling tests, FTIR, SEM, and BET analysis, revealing many possible interactions of pharmaceutical–polymer system. Through the latter interactions, the possible adsorption mechanism was elucidated and explained. The effect of key operational factors, i.e., solution pH, initial pharmaceutical concentration, contact time, temperature and regeneration (desorption pH, cycles reuse) has been systematically investigated. In the current study, apart from the fitting to the common kinetic models (pseudo–first, –second order and Elovich equations), a detailed kinetic model incorporating diffusion and localized adsorption and desorption steps was used. The latter was realized in order to extract information on the system dynamics, employing the experimental kinetic curves. By fitting the model to the experimental data and comparing the pore diffusivity of the solute with its corresponding bulk value, the relative contributions of the pore and surface diffusion on the overall diffusion process, are assessed. The connection between the findings for the adsorption mechanism and the chemical structure of the system is also discussed.

2. Materials and methods 2.1. Materials Pramipexole dihydrochloride (C10H21Cl2N3OS, MW=302.26 g/mol) was purchased from Amino Chemicals Ltd. (Malta) (assay 99.2%). Stock solutions of PRM (1000 mg/L) were prepared by weighing 5

and dissolving the suitable amount of the corresponding substance in water; stock solutions were stored at −20 oC and were stable for at least 1 week, as assessed by spectrophotometric assays. Working solutions of PRM were prepared daily by diluting the corresponding stock solutions in water. High molecular weight chitosan (ChHMW) was purchased by Sigma–Aldrich (purification in Soxhlet apparatus by extraction overnight with acetone). Then, the drying of chitosan particles was carried out under vacuum at 20 °C. Its average molecular weight was estimated at 3.55 105 g/mol and the degree of deacetylation was 85 wt%, according to the FT–IR method described in literature [31,32]. Glutaraldehyde (GLA, 50 wt% in water), formamide (•99.5%), chlorosulfonic acid (•99%), sodium borohydride (SBH), and sodium carbonate (•99%) were purchased from Sigma–Aldrich. Dichloroacetic acid (•98.5%) was obtained from Fluka, while 2–Carboxybenzaldehyde (CBA) (99%) was obtained from (Acros Organics). The chemical structure of PRM is given in Fig. 1.

2.2. Synthesis of adsorbents For the sulfonate–grafted chitosan adsorbent (CsSLF) [33], a mixture of dichloroacetic acid (5 mL) and formamide (50 mL) was added into chitosan (4.0 g) and stirred to be an homogenized solution. This was then mixed with a complex of chlorosulfonic acid–dimethylformamide and stirred for 1 h in a water bath at 50 °C. The reaction mixture was then diluted by a small quantity of deionized water, filtered, and precipitated by pouring into ethanol 95% (400 mL). The precipitate was dissolved in deionized water, neutralized by a saturated Na2CO3 solution, and dialyzed against deionized water. After dialysis, the product was dried and stored in a desiccator. Then, a cross–linking procedure was realized with GLA as reagent (0.5 wt%) at 60 °C for 1.5 h. The final grafting degree (GD) was determined on the basis of the percentage weight increase of the final product relative to the initial weight of chitosan GD = (W2–W1)/W1 (where W1 and W2 denote the weight of chitosan before and after grafting reaction, respectively). So, the grafting degree was found 2.2. For the N–(2–carboxybenzyl)–grafted chitosan adsorbent (abbreviated as CsNCB) [27,34], chitosan

6

powder (4.0 g) was gradually added, under constant mechanical stirring, into 500 mL of aqueous acetic acid 0.7% (v/v). After complete dissolution, 30 mL of CBA 17% (w/v) solution in ethanol were added dropwise. The amount of CBA used corresponded to 2:1 molar ratio of reactive groups CHO/NH2. After stirring in a water bath at 80 °C for 5 h, the mixture was cooled down to room temperature and poured into an excess of acetone to isolate the Schiff base derivative. 2.0 g of the purified Schiff base was dissolved in 250 mL of aqueous acetic acid 0.7% (v/v) where 10 mL of SBH aqueous solution 8% (w/v) was added dropwise, followed by a 2 h stirring period. The N–(2–carboxybenzyl) chitosan derivatives produced were precipitated by adding an adequate amount of acetone. Schiff bases and N–alkyl chitosan derivatives were purified by soaking in acetone for 24 h and dialyzed against deionized water for 72 h. The purpose of using a nitrogen purge was the protection from oxygen, since reactions such as depolymerization and oxidation are favoured at temperatures higher than 55 °C. After the grafting reactions, the cross–linking ones were followed in order to prepare the derivatives in the form of powder. CsNCB derivatives in solution 0.4% (w/v) were prepared in 0.5 L of aqueous acetic acid 0.7% (v/v). Under intense stirring, 6 mL of GLA aqueous solution 1% v/v was added dropwise (in the case of powder form). The amount of GLA used was such that it could have reacted with 15% of the amino groups of a zero degree–substituted polysaccharide. After a 5 min stirring period, the bubble–free solution was poured into a glass beaker and left there for 12 h allowing formation of the gel network. The hydrogel was dried at 60 °C till stable weight, milled and the fraction 100–200 µm was collected. This fraction was purified by extraction with acetone in a Soxhlet apparatus for 24 h, followed by drying under vacuum at room temperature (25 °C). The grafting degree of CsNCB was found 2.1. Non–grafted chitosan (just cross–linked as described above) was also prepared (denoted hereafter as Cs) [25] and used as reference adsorbent.

2.3. Characterization techniques Scanning electron microscopy (SEM) images were performed at Zeiss Supra 55 VP. The accelerating voltage was 15.00 kV and the scanning was performed in situ on a sample powder. The FTIR 7

spectra of the samples were taken with a PerkinElmer–2000 FTIR spectrometer using KBr disks prepared by mixing 0.5% of finely ground adsorbent sample in KBr. Pellet made of pure KBr was used as the reference sample for background measurements. The spectra were recorded from 4000 to 400 cm–1 at a resolution of 4 cm–1. The spectra presented are baseline corrected and converted to the transmittance mode. The measurement of surface areas of all adsorbent were realized with a Micromeritics BET surface area analyzer (TriStar 3000), by means of adsorption of ultra pure nitrogen at 77 K. Swelling experiments were performed at pH=10 (optimum pH value after adsorption experiments). 1 g of material was immersed in deionized water in order to be swollen for 24 h. The pH adjustment was achieved with NaOH. The material was allowed to be completely swollen, until there was no further weight increase. The swollen samples were weighted and the swelling percentage S (%) was calculated by Eq. (1), where Wt (g) is the weight of the swollen sample at time t, and W0 (g) is the initial weight of the sample before swelling.

⎛ W − W0 ⎞ S=⎜ t ⎟ ⋅100% ⎝ W0 ⎠

(1)

The need of the swelling determination was on the basis of use pre–swollen particles for all adsorption experiments performed in the current study.

2.4. Adsorption–desorption experiments Adsorption/desorption experiments were conducted in 20–mL amber vials using a batch approach developed in our previous studies [21,24,25]. All experiments were run in duplicate. Samples were taken at predetermined time intervals and filtered using 50 µm pore size filtration membrane (purchased by Schleicher&Schuell–MicroScience).

The

residual

concentration

of

PRM

was

measured

spectrophotometrically by monitoring its UV absorbance at 263 nm (model U–2000, Hitachi). Preliminary experiments were carried out in order to evaluate the effect of pH change on UV absorbance of PRM. The obtained results showed that the change was neglected (~2%). The calibration graph of absorbance versus concentration obeyed a linear Beer–Lambert relationship.

8

The variation of the PRM concentration versus time in the supernatant aliquot has been observed under various experimental conditions. A detailed description is given below, where [PRM]0 (mg/L) is the initial PRM concentration, pH is the pH of the aqueous solutions (fixed with micro–additions of HCl or NaOH), T (°C) is the temperature, m (g) is the mass of the adsorbent used, V (mL) is the volume of adsorbate, N (revolutions or full rotations per minute, abbreviated as rpm ) is the agitation rate of the shaking machine and t (h) is the contact time. It is selected to keep a constant value of 1 g of adsorbent’s mass per 1 L of aqueous solution for each experiment. So, given the use of 20 mL as adsorbate volume, 0.02 g of adsorbent was needed for all experiments. (i) Effect of pH on adsorption: [PRM]0=200 mg/L; pH=2–12; m=0.02 g; V=20 mL; T=25 °C; N=160 rpm; t=24 h. (ii) Effect of contact time (kinetics): [PRM]0=200 mg/L; pH=10 (optimum value found from (i)); m=0.02 g; V=20 mL; T=25 °C; N=160 rpm; t=0–24 h. Pseudo–first [35], –second order [36] and Elovich equations [37,38] were selected to fit the experimental kinetic data. (iii) Effect of initial PRM concentration: [PRM]0=0–500 mg/L; pH=10 (optimum value found from (i)); m=0.02 g; V=20 mL; T=25 °C; N=160 rpm; t=24 h. The equilibrium data resulted were fitted to the Langmuir–Freundlich (L–F) isotherm model [39]: Qe =

Q m bCe1/n 1+ bCe1/n

(2)

where Qe (adsorbed PRM weight/adsorbent weight) is the equilibrium concentration in the solid phase; Qm is the maximum amount of adsorption (adsorbed PRM weight/adsorbent weight); b is the Langmuir– Freundlich constant; n is the Langmuir–Freundlich heterogeneity constant. The adsorption capacity in equilibrium (Qe) was calculated using the mass balance equation:

Qe =

(C

b0

)

− Ce V m

(3)

where Cb0 and Ce (PRM weight/liquid volume) are the initial and equilibrium PRM concentrations in the

9

liquid phase, respectively. (iv) Effect of temperature (isotherms): [PRM]0=0–500 mg/L; pH=10 (optimum value found from (i)); m=0.02 g; V=20 mL; T=25, 45, 65 °C; N=160 rpm; t=24 h. Desorption of PRM from synthesized chitosan adsorbents was investigated to study the interaction between pharmaceutical and material surface. Before desorption, adsorbents have to be loaded in an adsorption step. In this stage (adsorption), 0.02 g of adsorbent was added in conical flasks with 20 mL of 200 mg/L PRM at pH=10. The agitation rate was 160 rpm and the contact time was 24 h at 25 °C. Afterwards, the samples were collected and filtered, using fixed pore-sized membranes. A small fraction of the PRM (1–2 %) and the adsorbent (1%) were retained on the filter membrane; these small variations due to filtration were neglected. Desorption experiments were realized by mixing the collected, after adsorption, amount of PRM-loaded chitosan adsorbents (0.02 g) with aqueous solutions of 20 mL (same volume as in the adsorption step) over a pH range between 2 and 12, at 25 °C for 24 h (agitation rate=160 rpm). The evaluation of desorption was realized as desorption percentage, calculated from the difference between the loaded amount of PRM on adsorbent after adsorption and the amount of PRM in solution after desorption. This procedure was realized to determine the optimum desorption pH value of the PRM-loaded adsorbents.

2.5. Kinetic theory – modeling development

There are two type of kinetic modeling approaches used in the adsorption literature. The first approach is based on the use of simple models like the first–, second–order models and the Elovich equation [35-38]. These models may be successful in fitting the experimental data, but they are just semi– empirical relations, which do not say anything about the underlying physics of the adsorption process. The extracted kinetic constants are not directly related to fundamental quantities so they can be served only for adsorbent screening procedures. It is noted that these models do not take into account the reduction of the driving force for adsorption in batch experiments, where the bulk solute concentrations decreases in time. 10

An alternative to the above modeling approach (also widely employed in literature [39], but in less extent than the first approach), is the use of phenomenological models attempting to describe the physics of the process. Information about the mechanism of the adsorption can be extracted using this type of models and employing the kinetic experimental data and the corresponding isotherms data. From measurements of the water uptake, it was found that the adsorbent particles have very high water content (typically more than 80%). The fate of a solute molecule in the adsorbent particle is determined considering its association/dissociation to the active center of the adsorbent particle. The nature of the forces leading to this interaction will be discussed in the next section. The association–dissociation kinetics is infinitely fast, so it can be considered that there is always equilibrium between the solute in the water phase and this on the active centers. There are two mechanisms of transfer of the solute towards the interior of the sorbent particle: (i) diffusion of the free particles in the water phase of the particle, and (ii) diffusion of the associated molecules from one association site to another. A mathematical model based on the above physical picture will be developed here. The swollen adsorbent particles are assumed to be spheres with radius R and density ρp (mass of dry material/volume of wet particle). The radial coordinate in the particle is denoted as r. The local adsorbed solute concentration per mass of dry material is denoted as Q. According to the experimentally derived isotherms, the quantity of the adsorbed solute per dry gram of adsorbent material is of the order of 1000 times larger than the solute quantity in the same volume of water. Even for wet particles with 90% water content, the amount of solute in the intraparticle liquid phase is at about 1% of the amount of the associated solute, according to a simple mass balance. For this reason, it can be easily ignored. The experiments were performed at high degree of agitation (N=160 rpm), so the results are independent from the agitation rate. This means that any kinetic limitation imposed by mass transfer of the solute from the bulk of the liquid to particles surface can be ignored. Based on the above assumptions, the governing equation for the spatiotemporal evolution of Q in the adsorbent particle can be written as follows (where t is time, C is the local solute concentration in the 11

water phase of the particle, Dp is the effective diffusivity of the solute in the water phase of the particle, and Ds is the diffusivity corresponding to the solute transfer from the one associate site to the other):

∂Q 1 ∂ 2 ⎛ D p ∂C ∂Q ⎞ = 2 r ⎜ + Ds ⎟ ∂t r ∂r ⎝⎜ ρ p ∂r ∂r ⎠⎟

(4)

The local concentration of the liquid phase solute C is related to the local concentration of the associated solute Q through the equilibrium relation Q=f(C). The function f is the same derived by employing adsorption isotherms i.e.:

f(C) = Q =

Q m bC1/n 1+ bC1/n

(5)

The boundary conditions for Eqs. (4), (5) are: spherical symmetry boundary condition ⎛ ∂Q ⎞ ⎜ ⎟ =0 ⎝ ∂r ⎠r=0

(6)

In the absence of mass transfer limitation from bulk liquid to particle: C=Cb at r=R

(7)

where Cb is the instantaneous concentration of the solute in the bulk. The average adsorbed solute concentration Q can be found as: R

Qave =

3 Qr 2 dr 3 ∫ R 0

(8)

For batch experiments, the concentration of the solute in bulk liquid Cb decreases with time due to adsorption. A global mass balance of the solute must be considered in order to close the mathematical model:

Cb =Cb0 −

m Qave V

(9)

The system of Eqs. (4)–(9) cannot be solved analytically due to the nonlinearity imposed by the adsorption 12

isotherms so a numerical solution is needed. The partial derivatives in equation are discretized using second–order finite differences and the resulting system of ordinary differential equations is integrated using a Runge–Kutta based explicit integrator. It is noted that several phenomenological models used in the literature based e.g. on a constant overall diffusion coefficient, on a loading dependent surface diffusion coefficient or on the shrinking core dynamics of the adsorbent particle, are sub–cases of the general model proposed here and they can be derived from Eq. (4) based on physical or mathematical approximations.

3. Results and Discussion 3.1. Characterization

The effect of grafting reactions to the physical structure and the appearance of chitosan can be observed from SEM images. All derivatives of chitosan (Cs, CsNCB and CsSLF) had an irregular shape owing to the grinding. The nearly total smooth surface of Cs (Fig. 2a) was changed to grafted derivatives. It is readily observed that the drying method caused the collapse of any porous microstructure of the particles [40]. This is possibly due to hydrophilic interactions between the water molecules and the carboxyl, hydroxyl and amino groups on the macromolecular chains of the prepared materials [25]. Also, CsSLF

showed rougher surface (Fig. 2c) than CsNCB (Fig. 2b), which can be originated from the slight different grafting degree (CsSLF, 2.2; CsNCB, 2.1). The aforementioned loss of porous structure has a significant effect on the surface area of the particles. For CsNCB particles, the surface area was 2.7±0.3 m2/g, while 2

2

for CsSLF was 2.9±0.3 m /g. The non–grafted derivative had only 0.9±0.2 m /g [25]. When the surface area of the materials is very low, i.e below 10 m2/g (as in our case), then the method has the accuracy of approximately 0.3 m2/g based on 3–5 repeated measurements. In this study, 5 measurements were taken and an accuracy of 0.2 m2/g was taken into account. The above values belong to the typical ones of non–porous materials, as bibliographically chitosan is characterized [41]. In the field of swelling, the materials prepared showed swelling ~300% at pH=10. The above percentages are commonly observed in chitosan adsorbents, given the powdered–nature of the adsorbents 13

and the single cross–linking method with GLA followed in the current study (and not dual cross–linking with GLA and some ionic reagent as sodium tripolyphosphate) [21,42]. FTIR spectroscopy (Fig. 3) confirmed the preparation of chitosan derivatives (Cs, CsNCB, and CsSLF) from the initial pure chitosan (ChHMW) and revealed their interactions with PRM molecules in PRM–loaded adsorbents. Typical absorbancies of chitosan–based materials [43,44] are recorded: 1664 cm–1 (amide I) and 1564 cm–1 (amide II); 3350 cm–1 (broad band due to stretching vibration of O–H); 1595 cm–1 (extension vibration of N–H and inter–hydrogen bonds); 1140 cm–1 (asymmetric stretching of the C–O–C bridge); 1074 and 1034 cm–1 (some skeletal vibrations with possible C–O stretching). The spectra of the grafted derivatives (CsSLF and CsNCB) differ from Cs owing to the large presence of the homopolymers (2.2 and 2.1 grafting degree, respectively). CsNCB spectrum (inset of Fig. 3) showed some new peaks as that recorded at 1612 cm–1 (asymmetrical stretching vibrations of • COO• ), shoulder–like band at 1738 cm– 1

due to the carboxyl groups [45], and an increase at 1675 cm–1 due to C=N vibrations, which overlap

chitosan’s amide I (C=O) absorption band (and also confirms the cross–linking reaction and the successful formation of the Schiff’s bases (imines)). PRM has characteristic peaks located at 2512 and 2426 cm–1

(with low intensity), which are attributed to the tertiary amine hydrochloride stretch, as well as at 3069 and 3171 cm–1 due to the primary and secondary amino groups. In order to evaluate the interactions taking place during the adsorption procedure of the used material and PRM drug, all the following shifts of the peaks were observed. In the case of neat Cs, the absorption band at 1664 cm–1 was shifted to 1652 cm–1, indicating that some hydrogen bonding interactions are taking place between the amide I group of chitosan and PRM drug. This was also confirmed from the shift of 3069 cm–1 absorbance of PRM amino groups to 3078 cm–1. However, in the case of grafted chitosan derivatives the differences in peak absorbancies are more characteristics. CsSLF has a characteristic absorption at 1730 cm–1, while when it was conjugated with PRM this peak was shifted to 1696 cm–1, demonstrating the formation of strong interactions between the sulfonate groups of CsSLF and the drug (the peak was exactly shifted by 34 cm–1). These interactions are taking place with the amino 14

groups of PRM, which bands were also shifted to 3124 and 3227 cm–1. Similar shifts were also recorded in CsNCB derivative and PRM, since the characteristic peak of carboxyl groups at 1738 cm–1 was shifted to 1639 cm–1. Hydroxyl groups of CsNCB can also participate in interactions, since some small shifts from 3306 to 3404 cm–1 were also recorded. All these confirm that the interactions between chitosan derivatives and PRM drug are stronger than with neat Cs.

3.2. Effect of pH on adsorption and evaluation of its mechanism

Fig. 4 illustrates the pH–effect of the particular adsorption process. For all adsorbents tested, the increase of pH enhanced the PRM removal. At pH=2, the PRM removal was found 32% for Cs and 11%, 36% for CsNCB and CsSLF, respectively (grafted materials). In general, the amino groups of Cs are protonated at acidic pH values and consequently the material becomes positively charged [46,47]. However, due to the positive charge of the cationic nature of PRM (Fig. 1), strong repulsive coulombic forces are developed between PRM and amino groups of Cs. With the pH increase, these forces are weakened since the amino groups of Cs are deprotonated and PRM uptake increases (from 32% at pH=2 to 50% at pH=10). On the other hand, the grafting of Cs with anionic groups, as N–(2–carboxybenzyl) (CsNCB) and sulfonate (CsSLF), caused a turn on pH–behavior. Fig. 4 also revealed a milder adsorption behaviour of Cs than CsSLF and CsNCB. This can be explained by the non–existence of extra functional groups on chitosan matrix, which can either improve or reduce the adsorption phenomenon more drastically [21,24,31,46]. So, CsSLF and CsNCB presented higher PRM uptake than Cs at alkaline pH conditions. Increasing the solution pH, deprotonation of the anionically grafted derivatives is realized and strong attractive forces, between the positive charged PRM molecule and negatively charged grafted chitosan materials, resulted in high uptakes as follows: Cs, 50%; CsNCB, 78%; CsSLF, 82% (pH=10). The interactions occurred at pH=10, where found to be the optimum pH value, can be described by Fig. 5 (CsNCB) and Fig. 6 (CsSLF). In the case of CsNCB, the most possible interactions are mainly the attraction between the carboxylic ion (resulted from grafting) of chitosan and primary amino group of PRM and secondly among the hydroxyl groups of chitosan and secondary positive amino group of PRM (Fig. 15

5). Those considerations can be also confirmed with FTIR spectra as extensively described in Section 3.1. In the case of CsSLF, the interactions dominated are those between amino groups of PRM and sulfonate groups of the grafted derivative (Fig. 6) [33]. However, the adsorption mechanism is not only based on electrostatic forces, but also on π–π interactions, van der Waals bonds or extra hydrogen bonding originating from the grafted groups [16,22,48,49].

3.3. Isotherms Fig. 7 presents the effect of initial PRM concentration on equilibrium. The experimental data were successfully fitted to the L–F model (R2=0.987–0.999) and the resulted parameters are given in Table 1. The values in each point express the initial PRM concentration used. The increase of the amount of PRM adsorbed is evident, when the initial PRM concentration increases. The adsorption capacities followed the order CsSLF>CsNCB>Cs (337, 307, 181 mg/g at 25 °C, respectively). The lower adsorption capacity of Cs could be explained from the low intense interactions taking place between Cs and PRM, as was verified from FTIR spectroscopy. Furthermore, increasing the temperature from 25 to 65 °C, an improvement of adsorption capacity was presented, indicating the endothermic nature of the process. The latter was also confirmed by a brief thermodynamic analysis (data not shown), according to which the change of enthalpy was positive. The adsorption process in the solid–liquid system is a combination of two processes: (i) the desorption of the solvent (water) molecules previously adsorbed, and (ii) the adsorption of the adsorbate species. In an endothermic process, the adsorbate species has to displace more than one water molecule for 0 their adsorption and this result in the endothermicity of the adsorption process. Therefore H will be

positive as found. The maximum adsorption capacity for CsSLF was found to be 367 mg/g at 65 °C, while for the same temperature the respective capacities of CsNCB and Cs were 345 and 211 mg/g, respectively. The increase of adsorption found is mainly due to the augmentation of the number of adsorption sites caused by breaking of some of the internal bonds near the edge of the active surface sites of chitosan [50]. The difference of Qm in grafted derivatives can be attributed to the fact that: (i) all amino groups of PRM can 16

interact with CsSLF via strong electrostatic forces, while the strength of carboxyl bonds is lower; (ii) CsSLF had 2.2 of grafting degree, while 2.1 was the respective degree of CsNCB, which means more in number active groups on the sulfonate derivative; (iii) CsSLF can easily interact both with primary and secondary amino groups, and (iv) it may happen increased penetration of PRM molecules inside the possible micropores or apparent surface cavities at higher temperatures or the creation of new active sites [40,50]. All the above confirms that grafted chitosan derivatives can be appropriate adsorbents for pharmaceutical contaminants removal from wastewaters. Their advantage of these materials is that chitosan is produced in large amounts in nature, is a fully biodegradable material while the whole procedure for pharmaceutical contaminates removal can be characterised as environmentally friendly procedure. This is because no toxic materials were used, there are no remaining toxic by–products and the drug removal is very high. Furthermore, a comparative table (Table 2) presents the superiority of the chitosan materials prepared in the current study than some other used in literature for removing pharmaceutical compounds [9,13,51].

3.4. Kinetics – Modeling

Fig. 8 (a, b, c) shows the effect of contact time on adsorption. The fitting was performed using pseudo–first, –second order and Elovich equations. Table 3 presents the kinetic parameters resulted from the fitting. Fig. 8a shows the plot of linearization of pseudo–first order model [35], where the slope (–

k1/2.303) and intercept log(Qe) of plot log(Qe–Qt) versus t was used to determine the pseudo–first order constant k1 and the equilibrium adsorption density Qecal. However, the experimental data deviated considerably from the theoretical data. The correlation coefficients (R2) obtained were not as high as those for pseudo–second equation (R2Cs=0.969; R2CsNCB=0.990; R2CsSLF=0.977). Also, the adsorption equilibrium values (Qecal) found gave significant deviation for all adsorbents (Table 3). These findings suggest that this adsorption system is not a pseudo–first order reaction. Furthermore, the experimental data fitted to the pseudo–second order equation (Fig. 8b), calculating the respective parameters [36]. The slope (1/Qe) and intercept (1/k2Qe2) of plot (t/Qt) versus t were used to calculate the parameters of k2 and Qecal. The 17

straight lines in plots of Fig. 8b showed an excellent agreement of experimental data with this model. The correlation coefficients for all adsorbents were equal to 0.999. Also, the calculated Qecal values also agree very well with the experimental data, as presented in Table 3. These findings indicate that the adsorption system studied belongs to the second order kinetic model. Fig. 8c shows a plot of linearization of Elovich model [37,38]. The slope and intercept of plots of Qt versus ln(t) were used to determine the constant βel and the initial adsorption rate α. However, the experimental data deviated considerably from the theoretical data. A comparison of the results with the correlation coefficients is shown in Table 3. The correlation coefficients for the Elovich kinetic model obtained at all the studies concentrations were low (R2Cs=0.834; R2CsNCB=0.864; R2CsSLF=0.843). This suggests that this adsorption system is not an acceptable for this system. The failure of the Elovich model to fit the data is trivial since this model cannot describe the approach to the equilibrium shown by the data. The only reason for its use is for completeness, since it is usually used in the literature despite its shortcomings. It is noted that the linear fitting approached followed here may led to poorer fitting quality than the one would be achieved by non-linear fitting. Nevertheless, any small change of the fitting quality would not change the overall picture described above. In this study, making an advanced step, the above kinetic experimental data were fitted to the model described in Section 2.4., which was developed from our research group. The phenomenological model, which takes into account the solute–adsorbent association and the intraparticle diffusion, will be applied next to the experimental data. Let us denote as β the ratio of the mass of water adsorbed by the particles to the mass of the dry particles as measured experimentally. The radius and density of the dry particles are denoted as R0 and ρ0, respectively and the water density is denoted as ρw. The swollen particles radius (Eq. (10)) and the corresponding density (Eq. (11)) are found as: 1/ 3

⎛ ρ β⎞ R = R 0 ⎜1 + 0 ⎟ ρw ⎠ ⎝ ⎛ ρ β⎞ ρ p = ρ0 ⎜ 1 + 0 ⎟ ρw ⎠ ⎝

(10)

−1

(11)

18

According to the above computations, the water volume fraction ( =intraparticle water volume/particle volume) in the particles is found about =0.82. Considering the possible uncertainty of the experimental measurements, it can be asserted that the computed values (ρp=273 kg dry material per m3 of

wet particle; Rp=172 µm; =0.82) area approximately the same for the three types of particles used (Cs, CsSLF, and CsNCB). Initially, it is assumed that only the diffusion in the water phase is responsible for the transfer of solute in the adsorbent particles. The model is implemented using the corresponding experimental conditions and the value of Dp, which leads to coincidence between the theoretical and experimental kinetic curves, is found as shown in Fig. 9. The following values of Dp were found:

DpCs=3.94×10–10 m2/s, DpCsSLF=4.13×10–10 m2/s, and DpCsSLF=4.08×10–10 m2/s (R2>0.998 for all cases). The diffusion coefficient of the particular solute in the bulk is computed as Dp∞=7.87×10–10 m2/s, based on Wilke–Chang correlation [52]. Considering again possible experimental uncertainties, it can be certainly stated that the ratio of Dp/Dp∞ is practically the same for the three materials and equal to 0.51. This value validates the assumption that the pore diffusion is the only active diffusion mechanism. At first, the adsorption capacity (as described by the experimental isotherms) has to do with the interactions between the adsorbent material and the solute. The difference in this interaction for the three materials used here has been discussed in Section 3.2. The kinetic of the adsorption is dominated by the intraparticle diffusion. The site-to-site (denoted as “surface” diffusion) diffusion, in general, depends on the interaction between the adsorbent and the solute (i.e it is adsorbent specific) but the pore diffusion depends only on the geometrical structure of the adsorbent. The pore diffusion is always present and the pore diffusivity cannot exceed the corresponding bulk diffusivity value. In the present case, the only diffusion mechanism is the pore one (water phase diffusion). This can be explained from the low density (ρp) of the active sites in the particular materials. The distance between two sites is large so the transition of the associated solute through surface diffusion is difficult. The domination of the pore diffusion is compatible with fact that the three sorbents exhibit similar kinetics irrespectively of their different isotherms. Adsorbents with the same water volume fraction should have the same kinetic behavior in case of pore 19

diffusion. The ratio of Dp/Dp∞ is a function of the geometric structure of the adsorbent particle and for particles with the same water volume fraction is expected to take the same value, as it is indeed the case here. In terms of the relation Dp/Dp∞=ε/τ used for porous material [39], a very reasonable value of tortuosity τ=1.6 arises here. The kinetic model developed here has the following differences with respect to the empirical models presented previously: (i) permits the determination of the physical mechanism underlying the process, and (ii) it can be used to predict the adsorption kinetics for the same materials but different conditions like particle of different sizes or different modes of adsorption e.g. continuous feeding of the solute in the tank or bed adsorption.

3.6. Desorption – Reuse

It has been proven that chitosan derivatives are effective adsorbents for PRM drug removal. However, the possibility of material reuse in adsorption process is of major importance, since it contributes significantly to lowering the operational cost of the wastewater treatment. Thus, it is important to study the desorption process of the PRM adsorbed given (i) the pharmaceutical behavior of the current compound, and (ii) the regeneration potential of materials. Firstly, desorption pH–effect experiments were carried out in order to find the optimum pH–desorption conditions. Fig. 10 showed that the pH–trend of curves were the contrast of that of adsorption. This confirmed that the dominated interactions occurred were electrostatic. At acidic conditions, the bonds between PRM (mainly amino groups) and chitosan are weakened and consequently the desorption percentage is higher: Cs, 85%; CsNCB, 94%; CsSLF, 95% (pH=2). At pH=10, where the adsorption is favoured, desorption occurs in low level (Cs, 15%; CsNCB, 21%; CsSLF, 20%). Fig. 11 shows the reuse potential of the materials studied via sequential cycles of adsorption– desorption. The reduction in adsorption percentages from the 1st to 4th cycle was 12% for Cs (from 50 to 37%), 6% for CsNCB (from 78% to 72%) and 6% for CsSLF (from 82% to 76%). However, apart from the reuse ability revealed for the grafted materials, the reuse–behavior of Cs was abnormal. While the grafted 20

materials demonstrated a relative reduction of their adsorption percentages during the cycles (approximately 2% loss in each cycle), Cs showed a strong decrease of PRM removal during the first cycle and a milder decrease during the other 3 cycles (50, 39, 38, 37% for the 1st, 2nd, 3rd and 4th cycle, respectively). This behavior may be explained by the non–grafted structure of the polymeric network [40,46]. The intense alternation of pH–conditions during cycles (from alkaline of adsorption to acidic of desorption) can be considered to cause a “shock” in the structure of Cs [53,54]. On the other hand, the grafted materials because of their complex network presented milder adjustment to conditions existed. In general, the decrease of the adsorption efficiency occurred can be attributed to several reasons as: (i) a

progressive saturation of the active sites/groups of the adsorbent by PRM molecules, (ii) a degradation of material due to extreme pH conditions, and (iii) a progressive blocking of the active sites of adsorbents by possible impurities caused a slight decrease in the adsorption potential. However, in chitosan derivatives, the reuse ability is very high since even after the 4th cycle the reduction presentence adsorption in CsNCB, and CsSLF was only 6 %, implying that they are promising candidates for the practical use in wastewater

treatment technologies.

4. Conclusions In the present study a new modified chitosan material, cross–linked with glutaraldehyde and grafted with CsSLF or CsNCB groups was synthesized and extensively investigated as low–cost and environmental friendly adsorbent for removal of PMR from wastewaters. According to SEM images, Cs presented smooth surface, while CsSLF showed rougher surface than CsNCB, which can be originated from the slight

different grafting degree (CsSLF, 2.2; CsNCB, 2.1). The aforementioned loss of porous structure has a significant effect on the surface area of the particles. BET analysis showed that CsNCB particles presented 2.7 m /g surface area, while CsSLF had 2.9 m /g. The non–grafted derivative (Cs) had only 0.9 m /g. The 2

2

2

nearly total smooth surface of Cs was changed to grafted derivatives. The interaction between PRM and modified chitosans were confirmed via FTIR spectroscopy. From the pH–effect experiments, the optimum 21

adsorption–pH values was 10, while pH=2 was found as the optimum for desorption. The increase of initial PRM concentration caused an enhancement of the adsorption capacity of adsorbents. The adsorption capacities resulted followed the order CsSLF>CsNCB>Cs after fitting to L–F isotherm model (337, 307, 181 mg/g at 25 °C, respectively). Also, the increase of temperature (from 25 to 65 °C) improved the

capacities (CsSLF, 367 mg/g; CsNCB, 345 mg/g; Cs, 211 mg/g at 65 °C, respectively). The experimental kinetic data were fitted to pseudo–first, –second order and Elovich equations. Only in the case of pseudo– second order model, the fitting was really successful (R2~0.999). A detailed adsorption kinetic model is

also used to fit the kinetic data. It was shown that the diffusion in the intraparticle water phase is the dominant mode of the solute transport in the adsorbent particle. The reduction in adsorption percentages from the 1st to 4th cycle was only 12%, 6%, and 6% for Cs, CsNCB, and CsSLF, respectively, revealing the high reuse ability of the prepared adsorbents. Overall, the findings of the present study demonstrated that the new synthesized chitosan materials can be used as promising, low cost and environmental friendly adsorbent materials in pharmaceutical wastewater reclamation.

22

References

[1] K. Kümmerer, Antibiotics in the aquatic environment - A review - Part I, Chemosphere 75 (2009) 417434. [2] D. Fatta-Kassinos, S. Meric, A. Nikolaou, Pharmaceutical residues in environmental waters and wastewater: Current state of knowledge and future research, Analytical and Bioanalytical Chemistry 399 (2011) 251-275. [3] M. Hörsing, A. Ledin, R. Grabic, J. Fick, M. Tysklind, J.L.C. Jansen, H.R. Andersen, Determination of sorption of seventy-five pharmaceuticals in sewage sludge, Water Res. 45 (2011) 4470-4482. [4] B. Yang, G.G. Ying, J.L. Zhao, S. Liu, L.J. Zhou, F. Chen, Removal of selected endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) during ferrate(VI) treatment of secondary wastewater effluents, Water Res. 46 (2012) 2194-2204. [5] F.J. Benitez, J.L. Acero, F.J. Real, G. Roldan, F. Casas, Comparison of different chemical oxidation treatments for the removal of selected pharmaceuticals in water matrices, Chem. Eng. J. 168 (2011) 11491156. [6] C.-M. Dai, J. Zhang, Y.-L. Zhang, X.-F. Zhou, Y.-P. Duan, S.-G. Liu, Selective removal of acidic pharmaceuticals from contaminated lake water using multi-templates molecularly imprinted polymer, Chem. Eng. J. 211–212 (2012) 302-309. [7] F.J. Rivas, O. Gimeno, T. Borallho, Aqueous pharmaceutical compounds removal by potassium monopersulfate. Uncatalyzed and catalyzed semicontinuous experiments, Chem. Eng. J. 192 (2012) 326333. [8] N. Sunsandee, P. Ramakul, N. Thamphiphit, U. Pancharoen, N. Leepipatpiboon, The synergistic effect of selective separation of (S)-amlodipine from pharmaceutical wastewaters via hollow fiber supported liquid membrane, Chem. Eng. J. 209 (2012) 201-214. [9] T.X. Bui, H. Choi, Adsorptive removal of selected pharmaceuticals by mesoporous silica SBA-15, J. 23

Hazard. Mater. 168 (2009) 602-608. [10] O. Lorphensri, J. Intravijit, D.A. Sabatini, T.C.G. Kibbey, K. Osathaphan, C. Saiwan, Sorption of acetaminophen, 17α-ethynyl estradiol, nalidixic acid, and norfloxacin to silica, alumina, and a hydrophobic medium, Water Res. 40 (2006) 1481-1491. [11] A.S. Mestre, J. Pires, J.M.F. Nogueira, A.P. Carvalho, Activated carbons for the adsorption of ibuprofen, Carbon 45 (2007) 1979-1988. [12] R. Baccar, M. Sarrà, J. Bouzid, M. Feki, P. Blánquez, Removal of pharmaceutical compounds by activated carbon prepared from agricultural by-product, Chem. Eng. J. 211–212 (2012) 310-317. [13] S. Fukahori, T. Fujiwara, R. Ito, N. Funamizu, pH-Dependent adsorption of sulfa drugs on high silica zeolite: Modeling and kinetic study, Desalination 275 (2011) 237-242. [14] A. Martucci, L. Pasti, N. Marchetti, A. Cavazzini, F. Dondi, A. Alberti, Adsorption of pharmaceuticals from aqueous solutions on synthetic zeolites, Micropor. Mesopor. Mater. 148 (2012) 174-183. [15] W.S. Wan Ngah, I.M. Isa, Comparison study of copper ion adsorption on chitosan, dowex A-1, and zerolit 225, J. Appl. Polym. Sci. 67 (1998) 1067-1070. [16] G. Crini, P.-M. Badot, Application of chitosan, a natural aminopolysaccharide, for dye removal from aqueous solutions by adsorption processes using batch studies: A review of recent literature, Prog. Polym. Sci. 33 (2008) 399-447. [17] F.-C. Wu, R.-L. Tseng, R.-S. Juang, A review and experimental verification of using chitosan and its derivatives as adsorbents for selected heavy metals, J. Environ. Manage. 91 (2010) 798-806. [18] T. Gotoh, K. Matsushima, K.I. Kikuchi, Preparation of alginate-chitosan hybrid gel beads and adsorption of divalent metal ions, Chemosphere 55 (2004) 135-140. [19] K.A. Alkhamis, M.S. Salem, M.S. Khanfar, The sorption of ketotifen fumarate by chitosan, AAPS PharmSciTech 9 (2008) 866-869. [20] J. Meler, B. Grimling, J. Pluta, Studies of norfloxacin adsorption on chitosan, Progress on Chemistry and Application of Chitin and its Derivatives 2012 (2012) 99-106. 24

[21] G.Z. Kyzas, D.N. Bikiaris, N.K. Lazaridis, Low-swelling chitosan derivatives as biosorbents for basic dyes, Langmuir 24 (2008) 4791-4799. [22] G.Z. Kyzas, M. Kostoglou, N.K. Lazaridis, Relating interactions of dye molecules with chitosan to adsorption kinetic data, Langmuir 26 (2010) 9617-9626. [23] G.Z. Kyzas, M. Kostoglou, A.A. Vassiliou, N.K. Lazaridis, Treatment of real effluents from dyeing reactor: Experimental and modeling approach by adsorption onto chitosan, Chem. Eng. J. 168 (2011) 577585. [24] G.Z. Kyzas, N.K. Lazaridis, Reactive and basic dyes removal by sorption onto chitosan derivatives, J. Colloid Interface Sci. 331 (2009) 32-39. [25] N.K. Lazaridis, G.Z. Kyzas, A.A. Vassiliou, D.N. Bikiaris, Chitosan derivatives as biosorbents for basic dyes, Langmuir 23 (2007) 7634-7643. [26] G.Z. Kyzas, M. Kostoglou, N.K. Lazaridis, Copper and chromium(VI) removal by chitosan derivatives-Equilibrium and kinetic studies, Chem. Eng. J. 152 (2009) 440-448. [27] G.Z. Kyzas, M. Kostoglou, N.K. Lazaridis, D.N. Bikiaris, N-(2-carboxybenzyl)grafted chitosan as adsorptive agent for simultaneous removal of positively and negatively charged toxic metal ions, J. Hazard. Mater. [28] J.P. Hubble, P. Novak, Pramipexole: A nonergot dopamine agonist as drug therapy in Parkinson's disease, Expert Review of Neurotherapeutics 1 (2001) 43-51. [29] P.L. McCormack, M.A.A. Siddiqui, Pramipexole: In restless legs syndrome, CNS Drugs 21 (2007) 429-437. [30] S.A. Hollingworth, A. Rush, W.D. Hall, M.J. Eadie, Utilization of anti-Parkinson drugs in Australia: 1995-2009, Pharmacoepidemiology and Drug Safety 20 (2011) 450-456. [31] M. Rinaudo, Chitin and chitosan: Properties and applications, Prog. Polym. Sci. 31 (2006) 603-632. [32] M. Miya, R. Iwamoto, S. Yoshikawa, S. Mima, IR spectroscopic determination of CONH content in highly deacetylated chitosan, Int. J. Biol. Macromol. 2 (1980) 323-324. 25

[33] J. Miao, G.H. Chen, C.J. Gao, A novel kind of amphoteric composite nanofiltration membrane prepared from sulfated chitosan (SCS), Desalination 181 (2005) 173-183. [34]

K.P.

Koutroumanis,

K.

Avgoustakis,

D.

Bikiaris,

Synthesis

of

cross-linked

N-(2-

carboxybenzyl)chitosan pH sensitive polyelectrolyte and its use for drug controlled delivery, Carbohyd. Polym. 82 (2010) 181-188. [35] S. Lagergren, About the theory of so-called adsorption of soluble substances, Handlingar 24 (1898) 139. [36] Y.S. Ho, J.C.Y. Ng, G. McKay, Kinetics of pollutant sorption by biosorbents: Review, Sep. Purif. Methods 29 (2000) 189-232. [37] S.H. Chien, W.R. Clayton, Application of Elovich equation to the kinetics of phosphate release and sorption in soils, Soil Sci. Soc. Am. J. 44 (1980) 265-268. [38] D.L. Sparks, Kinetics of Reaction in Pure and Mixed Systems, in: D.L. Sparks (Eds.), Soil Physical Chemistry, CRC Press, Boca Raton, Florida, U.S.A., 1986, pp. 63-145. [39] C. Tien, Adsorption Calculations and Modeling, Butterworth-Heinemann, Boston, U.S.A., 1994. [40] J. Berger, M. Reist, J.M. Mayer, O. Felt, N.A. Peppas, R. Gurny, Structure and interactions in covalently and ionically crosslinked chitosan hydrogels for biomedical applications, Eur. J. Pharm. Biopharm. 57 (2004) 19-34. [41] N.G. Stanley-Wood, A. Chatterjee, The comparison of the surface area of porous and non-porous solids as determined by diffusional flow and nitrogen adsorption, Powder Technol. 9 (1974) 7-13. [42] G.Z. Kyzas, N.K. Lazaridis, M. Kostoglou, Modelling the effect of pre-swelling on adsorption dynamics of dyes by chitosan derivatives, Chem. Eng. Sci. 81 (2012) 220-230. [43] A.L.P. Fernandes, W.A. Morais, A.I.B. Santos, A.M.L.d. Araújo, D.E.S.d. Santos, D.S.d. Santos, F.J. Pavinatto, O.N. Oliveira, T.N.C. Dantas, M.R. Pereira, J.L.C. Fonseca, The influence of oxidative degradation on the preparation of chitosan nanoparticles, Colloid. Polym. Sci. 284 (2005) 1-9. [44] C.L. de Vasconcelos, P.M. Bezerril, D.E.S. dos Santos, T.N.C. Dantas, M.R. Pereira, J.L.C. Fonseca, 26

Effect of molecular weight and ionic strength on the formation of polyelectrolyte complexes based on poly(methacrylic acid) and chitosan, Biomacromolecules 7 (2006) 1245-1252. [45] R.A.A. Muzzarelli, F. Tanfani, S. Mariotti, M. Emanuelli, N-(o-carboxybenzyl) chitosans: Novel chelating polyampholytes, Carbohyd. Polym. 2 (1982) 145-157. [46] R.A.A. Muzzarelli, Natural Chelating Polymers: alginic acid, chitin, and chitosan, Pergamon Press, New York, U.S.A., 1973. [47] R.A.A. Muzzarelli, C. Muzzarelli, Chitosan chemistry: Relevance to the biomedical sciences, Adv. Polym. Sci. 186 (2005) 151-209. [48] A.J. Varma, S.V. Deshpande, J.F. Kennedy, Metal complexation by chitosan and its derivatives: a review, Carbohyd. Polym. 55 (2004) 77-93. [49] E. Guibal, Interactions of metal ions with chitosan-based sorbents: a review, Sep. Purif. Technol. 38 (2004) 43-74. [50] H.Z. Mousavi, A. Hosseynifar, V. Jahed, S.A.M. Dehghani, Removal of lead from aqueous solution using waste tire rubber ash as an adsorbent, Braz. J. Chem. Eng. 27 (2010) 79-87. [51] J.L. Sotelo, A.R. Rodríguez, M.M. Mateos, S.D. Hernández, S.A. Torrellas, J.G. Rodríguez, Adsorption of pharmaceutical compounds and an endocrine disruptor from aqueous solutions by carbon materials, Journal of Environmental Science and Health - Part B Pesticides, Food Contaminants, and Agricultural Wastes 47 (2012) 640-652. [52] C.R. Wilke, P. Chang, Correlation of diffusion coefficients in dilute solutions, AIChE J. 1 (1955) 264270. [53] M.N.V. Ravi Kumar, A review of chitin and chitosan applications, React. Funct.l Polym. 46 (2000) 127. [54] K.M. Kim, J.H. Son, S.K. Kim, C.L. Weller, M.A. Hanna, Properties of chitosan films as a function of pH and solvent type, J. Food Sci. 71 (2006) E119-E124.

27

28

1

FIGURE CAPTIONS

2

H2+ClN H3C

S NH3+Cl- H2O N

Fig. 1. Chemical structure of pramipexole dihydrochloride (PRM).

29

3

(a)

(b)

(c)

Fig. 2. SEM images of (a) Cs, (b) CsNCB, and (c) CsSLF. 4

30

5

Cs Cs-PRM CsSLF CsSLF-PRM

Absorbance

CsNCB CsNCB-PRM

4000 3500 3000 2500 2000 1500 1000

4000

3500

3000

PRM ChHMW

500

2500

2000

1500

1000

500

-1

Wavelength (cm ) Fig. 3. FT–IR spectra of: (red line) ChHMW; (black line) PRM; (blue line) Cs; (green line) Cs–PRM; (olive line) CsSLF; (orange line) CsSLF–PRM. (inset: FTIR spectra of CsNCB (black line) and CsNCB–PRM (red line)).

31

6

100 Cs CsNCB CsSULF

90

Adsorption (%)

80 70 60 50 40 30 20 10 0 0

2

4

6

8

10

12

14

pH Fig. 4. Effect of pH on PRM adsorption. 7

32

NH+

S

H3 C

NH3+ N -

OOC

HOH2C *

NH O

O HO

HO

*

O

O HOH2C

N

n

HOH2C

N HO *

O

O

O

O

HO NH

HOH2C

-

H+ N H3 C

*

OOC

n S NH3+ N

8 9

Fig. 5. Interactions between PRM and CsNCB.

10 11 12

33

H2+ N

S

CH3

+

H3N N

OSO3H2 C

GLA -

O3SO

O

O -

O

O

O3SO GLA

H2C OSO3

-

S +

CH3

H3N N

13 14

n

H2+ N

Fig. 6. Interactions between PRM and CsSLF.

15 16 17

34

280 500

240

400 300

Qe (mg/g)

200

500

400 300

200

160

200

120 100

500

400

300 200

80 100 80 60 40 20

40

Cs CsgNCB CsgSLF

0 0

50

100

150

200

250

300

350

400

Ce (mg/L) Fig. 7. Effect of initial PRM concentration on adsorption onto Cs, CsNCB, and CsSLF at 25 oC (the values in each point expresses the initial PRM concentration used). 18 19

35

20 21

2.4

Cs CsNCB CsSLF

2.1

Cs CsNCB CsSLF

14 12

Elovich equation

200 175

1.5 1.2 0.9 0.6

150

10

Qt (mg/g)

t/Qt (min g/mg)

log(Qe− Qt)

1.8

8 6

125 100 75

4

50

0.3

2 st

0.0 pseudo-1 order 0

50

100

pseudo-2 order

0

150

200

t (min)

(a)

250

300

1500

Cs CsNCB CsSLF

25

nd

0

200

400

600

800

t (min)

(b)

1000

1200

1400

1600

0 1

2

3

4

5

6

7

8

ln(t)

(c)

22 Fig. 8. Effect of contact time on adsorption of PRM onto Cs, CsNCB, and CsSLF, fitted to (a) pseudo–first order, (b) pseudo–second order, and (c)

Elovich equations. 23 24 25

36

26

210 Cs CsNCB CsSLF

180

Ct (mg/L)

150 120 90 60 30 0 0

50

100

150

200

250

300 1500

t (h) Fig. 9. Modeling (kinetic) curves of Cs, CsNCB, and CsSLF. 27 28

37

100

Desorption (%)

80

60

40

20

Cs CsNCB CsSLF

0 0

2

4

6

8

10

12

14

pH Fig. 10. Effect of pH on desorption. 29 30 31 32 33

38

Removal of PRM (%)

100 Cs

CsNCB

1

2

CsSLF

80

60

40

20

0 3

4

Cycles of reuse Fig. 11. Cycles of reuse.

39

34 35 Table 1

Equilibrium parameters for the adsorption of PRM onto Cs, CsNCB and CsSLF at 25, 45, and 65 °C, fitted to L–F isotherm model. T

Qm

b

n

R2

Adsorbent

o

C

mg/g

(L/mg)1/n

Cs

25

181

0.039

1.31

0.987

45

195

0.040

1.41

0.996

65

211

0.041

1.50

0.994

25

307

0.156

1.95

0.994

45

322

0.169

1.99

0.996

65

345

0.174

2.02

0.995

25

337

0.165

2.02

0.990

45

349

0.179

2.19

0.996

65

367

0.188

2.23

0.999

CsNCB

CsSLF

36 37

40

38 Table 2

Comparison of adsorption capacities for the removal of pharmaceutical compounds from aqueous solutions using adsorbents. Adsorbent

Compound

Qm (mg/g)

Reference

Mesoporous silica SBA-15

Carbamazepine

0.16

[9]

Mesoporous silica SBA-15

Clofibric acid

0.07

[9]

Mesoporous silica SBA-15

Diclofenac

0.34

[9]

Mesoporous silica SBA-15

Ibuprofen

0.41

[9]

Mesoporous silica SBA-15

Ketoprofen

0.28

[9]

High-silica zeolite HSZ-385

Sulfamethoxazole

237

[13]

High-silica zeolite HSZ-385

Sulfathiazole

402

[13]

High-silica zeolite HSZ-385

Sulfamerazine

302

[13]

High-silica zeolite HSZ-385

Sulfamethizole

267

[13]

High-silica zeolite HSZ-385

Sulfadimidine

278

[13]

Activated carbon

Atenolol

130

[51]

Activated carbon

Diclofenac

280

[51]

Non-grafted cross-linked chitosan (Cs)

Pramipexole

181

In this study

N–(2–carboxybenzyl) grafted chitosan (CsNCB)

Pramipexole

307

In this study

Sulfonate grafted chitosan (CsSLF)

Pramipexole

337

In this study

39 40

41

41 Table 3

Kinetic constants for the adsorption of PRM onto Cs, CsNCB and CsSLF for [PRM]0=200 mg/L. Pseudo–first order model

Pseudo–second order model

⎛ k ⎞ log ( Q e − Q t ) = log ( Q e ) − ⎜ 1 ⎟ t ⎝ 2.303 ⎠

⎛ 1 ⎞ t 1 +⎜ ⎟t = 2 Q t k 2 Qe ⎝ Qe ⎠

Qe,exp

k1

Qe,cal

(mg/g)

(min–1)

(mg/g)

Cs

100

0.0149

80

CsNCB

160

0.0144

CsSLF

164

0.0158

Adsorbent

R2

k2

Qe,cal

(g mg–1 min–1)

(mg/g)

0.969

3.752×10–5

103

134

0.990

2.293×10–5

137

0.977

3.110×10–5

Elovich equation

Qt = R2

1 1 ln ( a ⋅ β el ) + ln ( t ) β el β el R2

a (mg g–1 min–1)

(g/mg)

0.998

13.77

0.0613

0.834

164

0.998

17.02

0.0367

0.864

167

0.999

26.63

0.0386

0.843

k, and k2 are the rate constants for the pseudo–first and –second order kinetic model, respectively. a is the initial adsorption rate and βel is the desorption constant during any experiment exported from Elovich equation.

42

42

Highlights ►Modified chitosan as environmental friendly pharmaceutical adsorbents ►Characterization with swelling tests, FTIR, SEM, and BET analysis ►Better adsorption behavior/kinetic rates for modified chitosans as: CsSLF>CsNCB>Cs ►Equilibrium data were fitted to the Langmuir–Freundlich model ►Evaluation of adsorbents’ reuse via sequential adsorption–desorption cycles