Metal-Organic Framework MIL-53 ( Fe ) as an Adsorbent for Ibuprofen Drug Removal from Aqueous Solutions : Response Surface Modeling and Optimization

Faculty of Pharmacy, Nguyen Tat anh University, Ho Chi Minh City, Vietnam Institute of Hygiene and Public Health, Ho Chi Minh City, Vietnam Institute of Chemistry, Vietnam Academy of Science and Technology, Hanoi, Vietnam Dong Nai Technology University, Bien Hoa City, Dong Nai Province, Vietnam Institute for Tropical Technology, Vietnam Academy of Science and Technology, Hanoi, Vietnam Ho Chi Minh University of Natural Resources and Environment, Ho Chi Minh City, Vietnam NTT Hi-Tech Institute, Nguyen Tat anh University, Ho Chi Minh City, Vietnam Faculty of Chemical Engineering, Ho Chi Minh City University of Technology, Ho Chi Minh City, Vietnam


Introduction
Ibuprofen ((±R, S)-2-(4-(2-methylpropyl)phenyl)propanoic acid, IBU), a nonsteroidal anti-inflammatory drugs (NSAIDs), has widely been used for the treatment of bacterial infection-related diseases in many countries (its chemical structure is shown in Figure 1) [1][2][3][4].However, there is a rapid acceleration in IBU contamination due to the ineffective treatment of wastewater sources, derived from hospitals and pharmaceutical manufacturers [5][6][7].According to recent environmental reports, varying amounts of the IBU residue have been detected in some rivers (i.e., ames River, UK (0.783 μg/L), and Aura River, Finland (20 μg/L)), significantly exceeding the permissible standards for human health [8].Besides, the accumulation of the IBU pollutant in water may support the antibiotics-resistant bacteria existing in the environment [1,5,9].erefore, techniques for eliminating this contaminant have received a great attention.
anks to the good reusability of absorbents, the overall cost can be reduced.Moreover, the adsorption is highly compatible with organic-derived hazardous wastes including IBU contaminant [4].erefore, the IBU degradation by the adsorption process has been rapidly developed.
anks to the highly porous structure along with open metal sites, these materials have been considered as ideal platforms towards diverse applications in fuel cells, chemical sensing, energy conversion, catalysis, drug delivery, and thin films [23][24][25].However, the synthesis strategies for such materials were carried out via many elaborate steps combined with severe condition-controlled reactions, limiting their widespread applications [26].Several studies reported the use of hazardous and expensive agents (i.e., reversible additionfragmentation chain transfer (RAFT)) as vital components that control the growing of polymeric organic chains in controlled radical polymerizations, thus imposing the adverse impacts on the environment [26].It is evident that finding out the greener and sustainable pathways for the fabrication of nanostructured materials is necessary.
Metal-organic frameworks (MOFs) are advanced materials constructed by metal ions or clusters and organic ligands via the coordination bonds [27][28][29][30].Recently, they are considered as remarkable and promising materials because of good crystallinity, complex topologies, high porosity, tunable chemical properties, and large metal cluster density, and thereby, MOFs have attracted much attention in many fields such as drug delivery, catalysis, gas storage, sensor, and adsorption [31].MIL-53(Fe) or Fe(III)(OH)(1,4-BDC) is a typical class of MOFs generated by a combination between iron(III) cations and 1,4-dicarboxylic acid [32,33].
is structure consists of FeO 6 hexagonal chains connecting with dicarboxylate anions to form the three-dimensional networks or SBUs (secondary building units) [34][35][36].One of the most emergent features of MIL-53(Fe) compared with other MOFs is the "breathing effect," which experiences a breathing transition in the presence of guest molecules, thus creating the structural flexibility [37].Moreover, this material can be easily synthesized via the conventional strategies, such as microwave or solvothermal method [38,39].Unlike the other MOFs such as MIL-88B(Cr) or MIL-101(Cr), which also represents the same "breathing effect," MIL-53(Fe) is chemically stable and constituted of lower toxic metal centers and thus has gained much attention [40].
In this work, we described the MIL-53(Fe) synthesis by the solvothermal method and its application in IBU adsorption.Some techniques including XRD and FT-IR were used to analyze the as-synthesized products.Moreover, the experiments were optimized based on the response surface methodology.e D8 Advance Bruker powder diffractometer was used to record the X-ray powder diffraction (XRD) profiles using Cu-Kα beams as excitation sources.e FT-IR spectra were recorded on the Nicolet 6700 spectrophotometer to explore the functional groups.e UV-Vis spectrophotometer was used to determine the IBU concentration at wavelength 222 nm.All analytic samples were reactivated at 105 °C under nitrogen atmosphere.

Synthesis of MIL-53(Fe).
e MIL-53(Fe) was solvothermally prepared according to a recent study [37].Firstly, 1.35 g of FeCl 3 •6H 2 O and 0.83 g of H 2 BDC were dissolved in 25 mL DMF. e mixture was then transferred into a Teflon-lined autoclave and heated up at 150 °C for 6 h.e yellow solid was extracted, refluxed with DMF overnight, washed with C 2 H 5 OH for three times (3 × 10 mL), and then dried at 80 °C for storage in a desiccator.Figure 2 illustrates the synthesis strategy of MIL-53(Fe).

Experimental Batches.
e stock solutions (20 mg/L) were prepared by dissolving the IBU substrate in distilled water.Various levels of concentration (1.8-18.4mg/L) for the adsorption were generated by consecutively diluting the initial stock solutions.A series of HCl and KOH solutions was utilized to adjust the pH indexes.Herein, twenty experimental runs were randomly performed at room temperature following the response surface methodology (RSM) [42].Firstly, MIL-53(Fe) samples (0.16-1.84 g/L) were mixed with 50 mL of ibuprofen solutions (1.8-18.4mg/L) at various ranges of pH.
e flasks were sealed and placed in the shaking tables (200 rpm).After the adsorption process reached the equilibrium state at 120 min, the adsorbent solids were dispended and separated using the simple centrifugation.
e IBU residual concentrations were identified using UV-Vis spectroscopy at the wavelength of

2
Journal of Chemistry 222 nm. e percentage of removal y (%) and adsorption capacity Q (mg/g) can be mathematically defined as follows: where C 0 , C t , and C e are denoted for the initial concentration, concentration at time t (min), and equilibrium concentrations (mg/L) and m (g) and V (mL) are denoted for adsorbent mass and solution volume, respectively.

Optimization Model with Response Surface Methodology.
Response surface methodology is the empirically statistical method, which allows optimizing the target (y) or "response" based on the evaluation of chosen variables (x i ) [43].In this study, three factors as input parameters were selected to investigate including IBU concentration (x 1 ), MIL-53 dosage (x 2 ), and pH (x 3 ).eir random experimental runs were designed according to the guide of the Design-Expert version 10 (DX10) program (Table 1).e "response" of the model was the removal efficiency (y), and thus, the target of this model is to maximize the "response"; the function of three variables is given by the following equation: where y is the predicted response and x i and x j are the independent variables (i, j � 1, 2, 3, 4, . . ., k). e parameter β o is the offset coefficient, β i is the linear coefficient, β ii is the second-order coefficient, and β ij is the interaction coefficient [44].

Error Analysis.
In this study, we explored the nonlinear kinetic and isotherm models using error functions, which were fitted by the Origin 9.0 program.Error analysis aims to determine the suitability of any models, and their confidence level could be assessed via commonplace error functions [45].us, coefficient of determination (R 2 ), mean relative error (MRE), and sum square error (SSE) were used, and their mathematical forms were shown in equations ( 3)- (5).Moreover, Q i,cal and Q i, exp are the respective calculated and experimental adsorption capacity values: (5) is observation was also commensurate with several previous reports [33,46,47], suggesting that the MIL-53(Fe) was solvothermally synthesized.

Structural Characterization of MIL-53(Fe)
Moreover, the FT-IR spectrum in Figure 3(b) reveals the collection of functional groups that features the structure of MIL-53(Fe) (Table 2) [48].In detail, a broad vibration at around 3420 cm −1 was ascribed to the presence of O-H groups, which adsorbed water existing in the surface of MIL-53(Fe) [49].
e asymmetric (c as C-O) and symmetric (c s C-O) stretching of dicarboxylate linkers could be verified  by the appearance of sharp vibrations at 1550 cm −1 and 1390 cm −1 , respectively, [37].Moreover, the carbonyl groups (C�O) of carboxylate ligand (COO) were visible at 1678 cm −1 , whereas a very sharp peak at 745 cm −1 was consistently corresponding to Csp 2 -H bending vibrations, which belong to the aromatic rings of carboxylates [50].
Especially, the characteristic coordination bonds between Fe 3+ cations and -OOC-C 6 H 4 -COO-carboxylate anions were observed at the low wavenumber (540 cm −1 ), revealing the existence of a Fe-oxo bond inherent in the structure of MIL-53(Fe) [51].erefore, a combination of two characterization results demonstrates that the crystals of MIL-53(Fe) were successfully fabricated.e morphological properties of MIL-53(Fe) can be analyzed by the SEM technique.Figure 4 showed the MIL-53(Fe) particles at 100 and 10 µm, which generally welldescribed the polyhedron-like crystalline structure [32].However, a clear difference of MIL-53(Fe) samples could be observed between before and after purification.Figure 4(b) demonstrated that a certain part of as-synthesized MIL-53(Fe) morphology was not homogeneous, which may be attributable to the residual 1,4-BDC components.A previous study also implied that this precursor may encapsulate the assynthesized MIL-53(Fe) crystals [52].Meanwhile, Figure 4(c) showed a relatively smooth surface of refined MIL-53(Fe) along with the complete disappearance of amorphous solid phases, suggesting that the refinement of as-synthesized MIL-53(Fe) was necessary to eliminate the 1,4-BDC trace [52].
Figure 5(a) diagnosed the textual feature of MIL-53(Fe) using the Raman spectrum.Clearly, a shape peak could be observed at the Raman shift of 464 cm −1 , revealing the existence of gamma bond of Fe-oxo in the structure of MIL-53(Fe).Also, typical peaks of the aromatic C-H bond were found at 632 and 865 cm −1 , while the single peak at 1141 cm −1 was ascribed for the bonds between two sp 2hydrid carbon atoms of benzene rings and carboxylate groups.Importantly, the asymmetric (c as C-O) and symmetric (c s C-O) footprints of COO-linkers were again confirmed via a couple of consecutive peaks at 1451 and 1501 cm −1 , respectively.Finally, ](C�C) was a kind of typical bond of aromatic cycles, which were emergent at 1616 cm −1 .
To gain the insights into the nature of porosity and pore size in MIL-53(Fe), the pore size distribution was measured and shown in Figure 5(b).It is evident that pore diameters of MIL-53(Fe) exhibited a wide range of sizes, revealing the existence of both micropore and macropore in the structure, but macrosize pore (more than 50 nm) was utterly dominant.Notice that there was an infinitesimal amount of micropore, which was less than 2 nm in length.Moreover, the surface area of MIL-53(Fe) was determined by the N 2 adsorption-desorption isotherm at 77 K, which was found to be 5.5 m 2 /g.Tuan et al. also reported the very low surface area, at about 14 m 2 /g, which may be attributable to the closed pores of MIL-53(Fe) that prevent the arrival of nitrogen molecules [49].

Optimization of the Ibuprofen Absorbability of MIL-53(Fe) by RSM Tool.
Herein, we applied the RSM to optimize the removal of IBU in water using the MIL-53(Fe) as an adsorbent.All experiments were designed and analyzed statistically using the DX10 [11].In detail, three independent variables including initial IBU concentration, MIL-53(Fe) adsorbent dosage, and pH solution at various five levels were investigated.
e relationship between IBU removal efficiency and three independent variables can be described by the following equation:  e actual and predicted values for the experimental matrix were listed in Table 3. Table 4 illustrates the coefficients of ANOVA data including the sum of squares, degree of freedom, mean square, F value, and Prob.> F [43].
Interestingly, determination coefficients (R 2 ) and P values of the regression model were found to be 0.9621 and <0.0001 with the confidence level of 95%, respectively.us, the proposed quadratic model was statistically significant and could be used to assess the effect of factors (initial IBU concentration, MIL-53(Fe) adsorbent dosage and pH) on the IBU absorbability of MIL-53(Fe) [48]. Figure 6(a) also indicates the high compatibility of experimental and predicted values as their data points were closely distributed to the 45degree line.
Figures 6(b)-6(d) showing the three-dimensional response surfaces reveal the consistent impacts of three factors on the IBU drug removal efficiencies of MIL-53(Fe).Generally, the removal process of IBU was conducive with an increase in the amount of the MIL-53 adsorbent or a decrease in pH values of IBU solutions.Meanwhile, the initial concentration of IBU solution shows a negligible effect on the IBU removal efficiency.Understandably, the improvement of IBU removal percentage may be ascribed to the presence of "adsorption sites" [53].When an addition of adsorbent was considerable, "adsorption sites" containing the surface functional groups became more dominant, thus increasing the number of IBU molecules captured [54,55].By contrast, a decline in the amount of the MIL-53(Fe) adsorbent may result in the rapid saturation of adsorption sites, possibly rendering the treatment of IBU drug ineffective.
Importantly, the decontamination of the IBU pollutant reached the excellent results in the acidic media, while the figures for the basic/neutral solutions were clearly unfavorable.In detail, Figure 6(d) indicates that nearly 100% of IBU could be removed from water at the very low pH values (i.e., pH � 2.6-4.3) and high adsorbent dosage (1.5-1.84 g/L).
By applying the RSM via model confirmation, several optimal conditions were proposed including the factors along with the predicted response value (Table 5).To test the precision and confidence of proposed optimized conditions, a verification experiment was performed using the above proposed conditions.e observed results were in good agreement with the predicted results because there was an ineligible error between predicted and tested results, and "desirability" values were obtained up to 100%, reflecting the excellent closeness of response (y) to their ideal values [56].In other words, the model for the optimization of IBU adsorption over MIL-53(Fe) in this study achieved the high compatibility and thus could be used to design and optimize the experimental conditions.

Adsorption Studies.
To gain deeper insights into the adsorption mechanism and behavior of IBU over MIL-53(Fe), herein, the nonlinear kinetic and isotherm models could be adopted.All experiments were conducted at room temperature, and other experimental conditions were chosen from model optimization (Table 5), at initial concentration (10 mg/L), dosage (1.0 g/L), and pH 2.6.Note that samples were extracted for regular time intervals (0, 10, 20, 40, 60, 80, 100, and 120 min) to investigate the kinetic study, while IBU concentrations were in range from 5 to 20 mg/L.Final IBU content could be measured by using a UV-Vis spectrophotometer at 222 nm.
Any selected model needs to meet the requirements in terms of R 2 , MRE (%), and SSE.A model with a high R 2 value indicates that the predicted data are fitted well with actual data [45].Meanwhile, lower MRE and SSE values reveal the negligible magnitude of error functions.To assess the models, therefore, terms of R 2 , MRE (%), and SSE were added herein.
For the kinetic models, we optioned four equations including pseudo-first-order, pseudo-second-order, Elovich, and Bangham, which their mathematical forms were illustrated in the equations given in Table 6. Figure 7(a) illustrates the effect of uptake capacity on contact time.It is evident that the adsorption process reached an equilibrium nature after 90 min.According to Table 6, all kinetic models obtained an excellent fitness based on R 2 values (0.981-0.992), indicating the high compatibility between actual and proposed data.Nevertheless, among the models, the pseudosecond-order equation offered the highest degree of suitability because of the highest R 2 , smallest SSE, and very low MRE (%).In fact, Elovich model showed the lowest MRE (3.15%), but its SSE value was so far higher than the pseudosecond-order model, 7.44 compared with 0.39, respectively.erefore, the pseudo-second-order model could be used to explore the adsorption mechanism of the IBU species onto the adsorbent, which chemisorption plays a crucial role.
For the isotherm models, four equations including Langmuir, Freundlich, Temkin, and Dubinin and Radushkevich (D-R) were applied.Similar to the kinetic models, their mathematical forms were explained in equations given in Table 7. Obviously, the Langmuir demonstrated the most appropriate model via the highest R 2 and lowest MRE and SSE values, suggesting that the monolayer mechanism is more inclining to others.e maximum adsorption capacity (Q m ) obtained from the Langmuir equation was calculated at 10.67 mg/g, revealing the favorable adsorption and good capacity of IBU onto MIL-53(Fe).

Conclusion
e MIL-53(Fe) material was successfully synthesized from the precursors including FeCl 3 •6H 2 O and 1,4-BDC via the solvothermal method and structurally characterized using the X-ray diffraction patterns, FT-IR spectra, SEM, pore size distribution, and Raman.e characterization results implied that the MIL-53(Fe) possessed the crystalline structure, along with a low surface area (5.5 m 2 /g).Moreover, important functional groups were found (Fe-O, COO, C�O, C-O, C-H, and O-H), proving the presence of these chemical bonds in the structure of MIL-53(Fe).For the optimization 6 Journal of Chemistry experiment of ibuprofen over MIL-53(Fe), the RSM results indicated that all variables (IBU initial concentration, MIL-53(Fe) dosage, and pH had a vital impact on the IBU removal efficiency through the very high determination of coefficient (R 2 > 0.9) and very low P value (P < 0.0001).Also, the proposed model was statistically significant and could be used to find out the optimal conditions.e confirmation tests also proved that the model was highly compatible with experimental data because the observed values was consistent with the predicted values and high removal efficiency

Figure 6 :
Figure 6: Actual versus predicted plots (a) and three-dimensional response surfaces (b)-(d) presenting the effect of factors on the response.

Table 1 :
List of independent variables and their levels.

Table 3 :
Experimental design used for investigation of IBU removal over MIL-53(Fe).

Table 4 :
ANOVA data for the IBU removal models.

Table 5 :
Optimal conditions and confirmation runs.

Table 6 :
Nonlinear kinetic constants for the adsorption of IBU over MIL-53(Fe).