Supported ionic liquids as efficient materials to remove non-steroidal anti-inflammatory drugs from aqueous media

Non-steroidal anti-inflammatory drugs (NSAIDs) are largely consumed worldwide. As a result, NSAIDs were already found in a variety of environmental aqueous samples, in concentrations ranging from ng/L to µg/L. This is due to the inability of the currently used technologies in sewage treatment plants (STPs) and wastewater treatment plants (WWTPs) to completely remove such pollutants/contaminants, thus leading to serious environmental and public health concerns. This work addresses the preparation and application of materials based on silica chemically modified with ionic liquids (SILs) as alternative adsorbents to remove NSAIDs from aqueous media. Modified silica-based materials comprising the 1-methyl-3-propylimidazolium cation combined with six anions were prepared, and chemically and morphologically characterized. Adsorption kinetics, diffusion models and isotherms of sodium diclofenac – as one of the most worldwide consumed NSAIDs – were determined at 298 K. The Boyd’s film diffusion and Webber’s pore diffusion models were used to disclose the rate controlling step affecting the adsorption process. A maximum equilibrium concentration of sodium diclofenac of 0.74 mmol (0.235 g) per g of adsorbent was obtained. Several solvents were tested to remove diclofenac and to regenerate SILs, being the mixture composed of 1-butanol and water (85:15, v:v) identified as the most promising and eco-friendly. After 3 regeneration steps, the material is able to keep up to 75% of its initial adsorption efficiency. Considering the maximum values reported for sodium diclofenac in effluents from WWTPs/STPs, 1 g of the most efficient material is “ideally” able to treat ca. 50,000 L of water. These materials can thus be envisioned as efficient filters to be implemented at domestic environment in countries where the levels of pharmaceuticals are particularly high in drinking water.


Introduction
Due to the large worldwide consumption of active pharmaceutical ingredients (APIs), their global occurrence in the environment is a current matter of remarkable concern [1].
APIs have been found in non-negligible levels (up to µg.L -1 ) in sewage treatment plants (STPs), wastewater treatment plants (WWTPs) and surface water effluents [2,3], leading to serious problems in environmental and public health after long-term exposure [4].
Although the main purpose of pharmaceuticals is to improve humans and animals health, when released into the environment they will affect the entire biota, from primary producers and consumers to top predators [4,5]. Most of these compounds are cytotoxic, genotoxic and/or endocrine disruptors. Since the administered doses of APIs are not completely metabolized by humans or animals, they are excreted in the urine in either conjugated or in unchanged forms, reaching the aquatic environment [6]. Also, unnecessary or expired medications are recurrently disposed (directly) into wastewaters [7]. Albeit WWTPs use advanced processes for water purification, such as membrane filtration, ozonation, chlorination, flocculation/sedimentation and adsorption, none of these strategies was specifically designed to remove APIs [8], explaining why some of these contaminants were already detected even in drinking water [9,10]. According to the Global Water Research Coalition (GWRC) criteria [11] some non-steroidal antiinflammatory drugs (NSAIDs) are part of the top 10 persistent pollutants list, where diclofenac is the most prevalent NSAID used throughout the world [12]. As with most of APIs, the degradation of diclofenac is limited [13] and there is a lack of efficient strategies for its removal from aqueous samples [14]. Accordingly, diclofenac has been detected in rivers, sediments and sludges [10]; more recently, it has been found in drinking water sources [15]. 4 Research works aiming the removal of diclofenac from aqueous samples and the development of alternative treatment strategies which could be implemented in WWTPs have been previously carried out, where materials such as MOFs (metal organic frameworks) [16][17][18], activated carbon [19,20] and silica [21] have been studied.
Removal efficiencies of diclofenac of ca. 93% were achieved by Betrán et al. [22], using ozone and activated carbon. Primary treatments with coagulating and flocculating agents, such as FeCl3 and Al2(SO4)3, have also been reported; yet, these processes are not able to completely remove the drug from wastewater samples (maximum removal of 70%) [23].
Furthermore, conventional activated sludge processes showed a higher efficiency when compared to membrane bioreactors, where a 30-70% of removal of diclofenac was obtained [24].
Although advances have been accomplished in the past years, the development of costefficient techniques able to completely remove diclofenac and other APIs from the aquatic environment is still an urgent requirement of a sustainable modern society. To this end, supported ionic liquids (SILs) can be seen as an alternative class of materials to be employed in removal/adsorption processes. Several SILs, where the IL is covalently attached to the support material, have been reported in the literature to separate inorganic and organic anions, as well as organic compounds [25][26][27]. In addition to these, Myasoedova et al. [28] and Li et al. [29] reported the use of physically immobilized ILs, i.e. supported ionic liquid phases (SILPs), to extract Pt(IV) and Pu(IV), and five phthalates, respectively, from environmental water samples. Furthermore, an hybrid material composed of poly-[N,N-dimethyl-dodecyl-(4-vinylbenzyl)ammonium chloride] and polyoxometalate (PIL/POM) was successfully applied in the separation of ionic dyes [30]. Despite aqueous samples, SILPs comprising physically immobilized ILs have been investigated in the separation of gases [30][31][32]. Supported ionic liquid membranes have 5 also been investigated for separation purposes [33]. All these materials have shown to be promising in a variety of separations due to their chemical functionalization or impregnation with ILs. The presence of ILs in solid phases allows the preparation of materials bearing some properties of ILs, particularly the tunable properties provided by the IL cation/anion chemical structure design. In fact, it is mainly due to their designer solvents feature that ILs have been successfully applied in a variety of fields [31][32][33][34][35].
The combination of the ILs properties [36] with the advantages of solid supports raised the attention on the use of SILs as alternative materials for solid-phase extraction (SPE) [37]. To the best of our knowledge, the use of SILs as adsorbents for the removal of APIs from aqueous media has not been previously reported. In this work, IL-functionalized silica was prepared by chemically bonding 3-chloropropyltrimethoxysilane onto the silica surface, followed by reaction with N-methylimidazolium, resulting in the formation of 1methyl-3-propylimidazolium-based supported silica with chloride as the counter ion Silica-based SILs were chosen since silica is a low-cost and inert material, easily chemically modified, and has a high specific surface area, mechanical strength and thermal stability. All SILs were chemically and morphologically characterized and evaluated as alternative adsorbents to remove sodium diclofenac from aqueous media since it is the most prevalent NSAID used throughout the world, thus recurrently found in aqueous streams and even in drinking water [38]. Adsorption kinetics, diffusion models and isotherms of the drug were determined at 298 K, allowing to infer the materials removal/adsorption performance and diffusion mechanism for APIs. Water and several 6 alcohols and their mixtures were investigated to remove sodium diclofenac and to regenerate SILs.

Materials
The non-steroidal anti-inflammatory drug, sodium diclofenac (CAS# 15307-79-6), was acquired from Sigma-Aldrich. The physicochemical properties of sodium diclofenac are listed in Table SI1

Synthesis of IL-functionalized silica materials
The chloride-based functionalized silica, [Si][C3C1im]Cl, was prepared by the treatment of chloropropyl silanized silica ([Si][C3]Cl) with an excess of N-methylimidazole in anhydrous toluene according to the literature [26], with some optimizations carried out by us described as follows. Silica was first immersed in hydrochloric acid (37 %) for 24h, and then washed with double distilled water and dried under vacuum for 24h at 378 K. This activated silica (5.0 g) was re-suspended in 60 mL of dried toluene, followed by the addition of 5.0 mL of 3-chloropropyltrimethoxysilane, and 0.5 mL of triethylamine (used as catalyst). This suspension was magnetically stirred and refluxed at ca. 366 K for 24h.
After refluxing, the reaction was stopped and cooled down to room temperature, transferred to a vacuum glass filter, and washed with toluene (100 mL), ethanol-water  Fig. 1 Fig. 2

Fourier transform infrared (FTIR) spectra of activated silica, [Si][C3]Cl, and
[Si][C3C1im]Cl were acquired using a Perkin Elmer BX Spectrometer, with a resolution of 5 cm -1 and equipped with a horizontal Golden Gate ATR cell, in the range of 4000-500 cm -1 . Solid state 13 C NMR spectra were recorded at 9.7 T on a Bruker Avance III -400 MHz spectrometer (DSX model) using 4 mm BL cross-polarization magic angle spinning (CPMAS) VTN probes at 100.6 MHz, at room temperature. In order to increase the signal-to-noise ratio of the solid-state spectra, the CPMAS NMR 13 C settings used were the following: ν1 13

Adsorption kinetics, isotherms and particles diffusion
The adsorption kinetics and isotherms of the six SILs were determined using aqueous solutions of sodium diclofenac, as a major representative of the NSAIDs class. For the 11 adsorption kinetics, a solution of sodium diclofenac with a concentration of 0.079 mmol.L -1 was used, while for the adsorption isotherms, concentrations of the drug ranging from 0.003 to 1.446 mmol.L -1 were employed. In these experiments, 2.5 mg of each SIL were mixed with 10 mL of sodium diclofenac aqueous solutions in 50 mL pyrex Erlenmeyers, and then placed in an orbital shaker at 120 rpm and (298.2 ± 0.5) K. For the adsorption kinetics evaluation, samples were taken from 0 to 180 min, while for the adsorption isotherms evaluation samples were taken at 120 min. All samples were centrifuged at 5000 rpm for 5 min, and the amount of sodium diclofenac in the water phase was quantified through UV-spectroscopy, using a Shimadzu UV-1800, Pharma-Spec UV-Vis Spectrophotometer, at a wavelength of 276 nm. At least, three replicates were investigated for each condition.
The equilibrium concentration of adsorbate in the solid phase (qe/mmol.g -1 ) was determined according to Eq. 1: where w is the weight of each SIL (g), V is the volume of the sodium diclofenac aqueous solution (L), C0 is the initial concentration of sodium diclofenac in solution and Ce is the equilibrium concentration of sodium diclofenac after adsorption onto SILs (mmol.L -1 ).
Both pseudo first-order and pseudo second-order models were applied to correlate the experimental data, namely with the linear form of the Lagergren's [39] first order rate equation (Eq. 2) and Ho's [40] second order rate equation (Eq. 3): where t is the time (min), qe is the amount of sodium diclofenac adsorbed onto the adsorbent at equilibrium (mmol.g -1 ), qt is the amount of sodium diclofenac adsorbed onto the adsorbent at different times (mmol.g -1 ), k1 (min -1 ) is the rate constant of the pseudo first-order adsorption, and k2 (gmmol -1 min -1 ) is the rate constant of the pseudo-secondorder adsorption.
The diffusion process can be controlled by one or combined steps: film diffusion, pore diffusion, surface diffusion and sorption in the pore surface. In order to evaluate the rate limiting process, the most common diffusion models, namely the Boyd's film diffusion (Eq. 6 and Eq. 7) and Webber's pore diffusion (Eq. 8) models [43][44][45], were used: where F(t) is the fractional attainment of equilibrium at different times t, in which F(t)= qt/qe and Bt is a function of F(t), and qt and qe are the sodium diclofenac concentrations adsorbed onto the adsorbent (mmol.g -1 ) at time t and at equilibrium, respectively. Kid is the internal diffusion rate constant (mmol.g -1 .min 1/2 ).
The experimental isotherms were fitted with the Langmuir [41] (Eq. 7) and Freundlich [42] (Eq. 8) models: where Ce is the equilibrium concentration of adsorbate (mmol.L -1 ), qe is the equilibrium concentration of adsorbate in the solid phase (mmol.g -1 ), B (L.mmol -1 ) is the Langmuir isotherm constant, qmax (mmol.g -1 ) is the maximum monolayer coverage capacity, and Kf (adsorption capacity; mmol.g -1 ) and n (adsorption intensity) are the constants of the Freundlich equation. The fitting of the data was performed with SigmaPlot®. 13

Desorption and regeneration studies
Envisioning the materials regeneration, desorption studies were performed for the material with higher adsorption capacity for sodium diclofenac, namely  Table SI2 in the Supplementary Material.

Characterization of the prepared SILs
In order to confirm the silica functionalization with ILs, FTIR spectra of activated silica,

SILs
The  In order to study the adsorption mechanisms of sodium diclofenac in the prepared SILs, pseudo first-order kinetic (Eq. 2) and pseudo second-order (Eq. 3) kinetic models were used to correlate the experimental data. The adsorption kinetic parameters are summarized in Table 3 It should be noted that non-functionalized silica was also tested under the same conditions. Although with a higher specific surface area (SBET of 434.545 m 2 /g) and a similar pore size diameter, even for a longer time (up to 180 minutes), no adsorption of diclofenac was verified on silica, at least up to the detection limits of the analytical equipment used. includes zero. These results suggest that, under the investigated conditions, the film diffusion does not control the adsorption process during the initial period.
The numerical results for the Webber' plots obtained for the adsorption of sodium diclofenac onto SILs are presented in Table 4. Two operational stages were defined: the first stage corresponding to the steep-sloped portion of qt vs t 1/2 plots and the second one resembling the linear gentle-sloped portion, with a breakpoint (square root of the time where the two linear segments intersects). These two stages may correspond to the diffusion of sodium diclofenac into the pores of different sizes, and gradually smaller, of the SILs [44]. The values of kid (given in  The experimental data on the adsorption isotherms at 298 K were fitted by the Langmuir [41] (Eq. 7) and the Freundlich [42] (Eq .8) models, using initial concentrations of sodium diclofenac ranging between 0.003 and 1.446 mmol.L -1 . To guarantee that equilibrium was reached, the contact time to carry out the adsorption isotherms studies was set to 120 min.
The relationship between the equilibrium concentrations of sodium diclofenac between the solid and liquid phases is shown in Fig. 6 (the respective representations according to each SIL are given in Fig. SI5, and detailed data are given in Table SI9 in the 25 Supplementary Material). The equilibrium adsorption of sodium diclofenac onto SILs increases with the increase of its initial concentration until saturation, where a plateau is reached at around 0.2 mmol.L -1 of an equilibrium concentration of adsorbate.  promising and less toxic solvents in the desorption step than those commonly used [20].
Methanol was mainly studied to evaluate the molecular structure influence of alcohols for on the sodium diclofenac desorption. Detailed data are given in Table SI10 in the 29 Supplementary Material. As expected, water has a low performance (ca. 10%) to remove sodium diclofenac from the material. Methanol, ethanol, 1-butanol and 2-butanol lead to desorption values ranging from 24.7 to 33.0%. However, mixtures of alcohols and water perform better than the isolated compounds due to a co-solvency effect, with the mixture composed of 1-butanol and water (85:15, v:v) leading to a removal of (61.2 ± 8.7)% of the adsorbed sodium diclofenac.
After the sodium diclofenac desorption, the regeneration and reusability of the The results obtained are depicted in Fig. 7. In the 4 cycles of sodium diclofenac adsorption, the adsorption efficiency of the material decreases as follows: 100 > 92.5 > 86.3 > 75.0%. Although the adsorption performance decreases along the regeneration steps, it should be remarked that after the third regeneration the material still keeps 75% of its initial adsorption efficiency for sodium diclofenac, and at a concentration of ca. 4 mmol.L -1 which is well above the concentrations of the drug found in the real water samples aimed to be treatedcf. discussion below. Overall, the studied SILs display maximum equilibrium concentrations ranging from 0.503 to 0.754 mmol (0.16 to 0.24 g) of sodium diclofenac adsorbed per gram of material.
Diclofenac has been detected in effluents from WWTP/STP at levels ranging from 460 to 3300 ng.L -1 in Europe, from < 0.5 to 177 ng.L -1 in North America, and from 8.8 to 127 ng.L -1 in Asia and Australia [55]. Considering the largest reported value of 3300 ng.L -1 , 1 g of [Si][C3C1im]Cl is "ideally" able to treat around 50,000 L of water, a significantly higher volume than those reported using other strategies involving ionic liquids [56].

Conclusions
Six supported IL-based silica materials were synthesized and characterized, and further evaluated as adsorption materials to remove NSAIDs/APIs from aqueous media. The adsorption kinetics, isotherms and diffusion models of sodium diclofenac, as a major representative of the NSAIDs class, were determined for all materials. The film diffusion is the rate controlling mechanism for the adsorption of sodium diclofenac onto