Physicochemical Modeling of the Adsorption of Pharmaceuticals on MIL-100-Fe and MIL-101-Fe MOFs

The adsorption of naproxen (NAP), diclofenac (DFC), and acetaminophen (APAP) molecules from aqueous solutions using MIL-100-Fe and MIL-101-Fe metal organic frameworks (MOFs) has been analyzed and modeled. Adsorption isotherms of these pharmaceuticals were experimentally quantified at 30 and 40°C and pH 7. Textural parameters and surface chemistry of these MOFs were analyzed, and results were utilized to explain the pharmaceutical adsorption mechanism. Density Functional Theory (DFT) calculations were performed to understand the reactivity of pharmaceutical molecules, and a statistical physics model was employed to calculate the main physicochemical parameters related to the adsorption mechanism. Results showed that the adsorption of these pharmaceuticals on MOFs was multimolecular and exothermic. Both MOFs displayed the highest adsorption capacities, up to 2.19 and 1.71 mmol/g, for NAP and DFC molecules, respectively. MIL-101-Fe showed better pharmaceutical adsorption properties than MIL-100-Fe due to its highest content of Fe-O clusters and mesopore volume. Adsorption mechanism of these organic molecules could involve hydrogen bond, van der Waals forces, and electrostatic interactions with MOF surfaces. In particular, MIL-101-Fe MOF is a promising material to prepare composites with competitive adsorption capacities for facing the water pollution caused by pharmaceutical compounds.


Introduction
In recent years, the literature has documented an increment of the water pollution caused by emerging compounds [1]. They are unregulated pollutants by the environmental legislation that include an extensive variety of chemicals such as pharmaceuticals, personal hygiene products, and surfactants [1]. In particular, the release of pharmaceutical molecules in the environment is a relevant issue since they can affect significantly the human health causing, for example, carcinogenesis, teratogenesis, and mutagenicity even at very low concentrations (i.e., from μg/L to ng/L) [2][3][4][5]. Diclofenac (DFC), naproxen (NAP), and acetaminophen (APAP) are nonsteroidal anti-inflammatory drugs that stand out due to their worldwide prescription for human healthcare [6,7]. NAP is employed for the treatment of osteoarthritis, rheumatoid arthritis, and migraine and to reduce inflammation and fever [7,8]. DFC is a widely used pharmaceutical to minimize pain and inflammation caused by ankylosing spondylitis, rheumatoid arthritis, and osteoporosis [9]. This drug is commonly identified as a water pollutant in environmental samples [7]. APAP is also an anti-inflammatory, antipyretic, and analgesic drug [10]. Overall, these pharmaceutical molecules can reach wastewater via the human excreta and by their inappropriate disposal after the drug expiration. Different studies have concluded that these organic molecules can persist in the environment for a long time due to their resistance to biodegradability and stability to heat and light [11,12].
The control and reduction of concentrations of pharmaceutical pollutants in the environment, and especially in water resources, can be performed via adsorption [13,14] or using other treatment methods like photodegradation [15,16]. Particularly, the removal of pharmaceutical molecules via adsorption processes can offer additional economical and technical advantages, and consequently, it is necessary to study the application of novel adsorbents to consolidate its application at industrial level. In this direction, the metal organic frameworks (MOFs) are interesting adsorbents with promising potential for wastewater treatment including the pharmaceutical depollution due to their surface area, controllable pore size (in micro and mesoporous domains), structural versatility, and composition [17][18][19]. There is a wide spectrum of metals and organic ligands that can be used for the preparation of MOFs, thus offering the possibility to obtain materials with different physicochemical characteristics and adsorption properties. Research on MOFs has indicated that their adsorption capacity to remove organic molecules is determined by their textural parameters (surface area and pore size) and surface functionalities. These physicochemical properties influence the π-π interactions between the aromatic parts of the linker on the MOF structure and the organic molecules besides the electrostatic adsorbent-adsorbate interactions [20,21]. Particularly, MIL-100-Fe MOF has been suggested as a promising material for liquid-phase adsorption due to its hydrothermal stability, surface area, and pore volume [22]. It has been utilized successfully as an adsorbent of organic pollutants and heavy metals from liquid phase [23,24]. In this direction, MIL-101-Fe is another interesting MOF that has been proved as an adsorbent of various toxic chemicals due to the affinity of saturated or unsaturated metal sites [24]. Several studies on the water pollutant adsorption via MOFs have focused mainly on the assessment and improvement of the adsorption capacities of these materials including the preparation of composites. However, the discussion of the adsorbentadsorbate interactions and physicochemical parameters related to the adsorption mechanism is usually excluded in these studies. Note that the structure of adsorbate molecules (e.g., molecular size, functional groups, polarity, and solubility) could also have a significant impact on the MOF adsorption performance [25,26]. Therefore, the analysis and interpretation of the adsorption mechanism of pharmaceuticals on MOFs and the impact of their molecular properties are necessary to improve the application of these materials and to tailor their properties with the aim of enhancing the adsorption capacities in water depollution.
Under this perspective, this study reports the modeling and analysis of the physicochemical parameters of the adsorption of pharmaceutical molecules on MIL-100-Fe and MIL-101-Fe MOFs. These materials were synthesized and employed to adsorb DFC, NAP, and APAP as target pharmaceutical molecules from aqueous solutions. These MOFs were characterized and their adsorption properties were determined experimentally at 30-40°C and pH 7. The physicochemical parameters associated to the adsorption mechanism of these pharmaceuticals were calculated via the statistical physics-based modeling. Therefore, this study contributes with new experimental data and theoretical insights on the application of MOFs for the adsorption of pharmaceuticals molecules from aqueous solutions.

Methodology
2.1. Synthesis and Characterization of MOFs. MIL-100-Fe and MIL-101-Fe MOFs were prepared and employed to analyze the adsorption of different pharmaceutical molecules from aqueous solutions. MIL-100-Fe was synthetized with 1.76 g of Fe(NO 3 ) 3 ·9H 2 O and 0.57 g of H 3 BTC (trimesic acid). The organic linker and metal source were dissolved in 20 mL of deionized water and stirred well for 1 h. The final solution was submitted to a thermal treatment at 160°C for 12 h using a Teflon-lined stainless-steel autoclave. The solid product obtained from the reaction was separated via centrifugation, washed several times with deionized water and ethanol, and finally dried at 80°C for 24 h. The preparation of MIL-101-Fe was performed with 0.675 g of FeCl 3 ·6H 2 O and 0.206 g of H 2 BDC (terephthalic acid), which were dissolved in 30 mL of DMF. The thermochemical conversion of this solution was done at 110°C for 20 h with the Teflonlined stainless-steel autoclave. The final solid product was also separated via centrifugation, washed with ethanol, and dried at 80°C and 24 h. Nanosized MOF particles were used in all the pharmaceutical adsorption studies reported in this paper.
Samples of these MOFs were characterized to determine their main surface and textural properties. Surface functional groups were identified by Fourier-transform infrared (FTIR) spectroscopy with a Thermo Nicolet Is10 FTIR spectrometer (Thermo Scientific). Spectra were recorded with KBr-based sample pellets in the 4000-400 cm -1 range with a resolution of 4 cm -1 . Crystallinity analysis was done via the X-ray diffraction patterns using a Malvern-Panalytical X-ray diffractometer. Samples were analyzed at room temperature with copper radiation (λ = 1:5406 Å) in the angle of 5 ≤ 2θ°≤ 60 at 45 kV and 40 mA. The textural parameters were obtained from N 2 adsorption-desorption isotherms at -196°C using a Micromeritics ASAP 2020 equipment. These isotherms were analyzed with suitable models to estimate the main textural parameters of these MOFs. Morphology and surface elemental composition of these adsorbents were obtained by scanning electron microscopy (SEM) analysis using a TM3000 (Hitachi) microscope with an energy dispersion system (EDS) (Nano XFlash Bruker).

Pharmaceutical Adsorption
Studies. NAP (≥98%), DFC (≥98%), and APAP (≥99%) were supplied by Sigma-Aldrich and utilized to prepare the adsorbate solutions with deionized water. The main characteristics and molecular structures of these pharmaceuticals are shown in Table 1. Adsorption isotherms of these pharmaceuticals using MIL-100-Fe and MIL-101-Fe MOFs were experimentally quantified at 30-40°C and pH 7. These isotherms were determined with a MOF ratio (W/V) of 10 g/L at batch adsorption where C e (mmol/L) is the equilibrium concentration of tested pharmaceuticals in the aqueous solution. Concentrations of all pharmaceuticals in the aqueous solutions were quantified by high-performance liquid chromatography (HPLC) (Ultimate 3000TM Thermo Scientific™) at 220 nm with a Thermo Scientific™ Hypersil GOLD™ Aq C18, 3 μm, 4:6 mm × 250 mm reverse-phase HPLC column. A mixture of 0.1% formic acid in water and acetonitrile was the mobile phase with a flow rate of 1 mL/min. The reagents and water used for the mobile phase were HPLC grade. Thermo Scientific™ Chromeleon™ software was utilized to collect and process the pharmaceutical quantification data.

Thermodynamics and
Modeling of the Adsorption of Pharmaceutical Molecules. The thermodynamics of the pharmaceutical adsorption was analyzed calculating the adsorption enthalpy (ΔH°, J/mol) with the van't Hoff approach employing the next equation [27].
where R is the universal ideal gas constant (8.3144 J/mol·K), T is the adsorption temperature (K), ΔS°is the adsorption entropy (J/mol·K), and K c is the adsorption equilibrium constant that was calculated following the procedure reported by Tran et al. [28].
Results of MOF characterization and the analysis of adsorbate molecular structure were utilized to define a statistical-physics-based model to calculate the physicochemical parameters of the adsorption mechanism of pharmaceutical molecules [29][30][31][32]. Therefore, this model was utilized to fit the adsorption data and to determine these parameters. It is convenient to note that the adsorption of tested pharmaceutical molecules on these MOFs could imply two surface functionalities: oxygenated functional groups (e.g., -COOH and -OH) and Fe-O clusters. But the complex molecular structure of MOFs limits the possibility of identifying, with reliability, the specific contribution of these functionalities during the pharmaceutical adsorption. Based on these facts, the statistical physic model assumed that one functional group (i.e., oxygenated functionality with or without Fe) was involved in the adsorption of NAP, DFC, and APAP where an adsorbate monolayer was also formed. This model was defined as [33] q e = n phar N ads where n phar represents the number of pharmaceutical molecules adsorbed for each MOF functional group, N ads is the amount of MOF functional groups (mmol/g) involved in the pharmaceutical adsorption, and C h is the half saturation adsorbate concentration (mmol/L), respectively. The pharmaceutical adsorption capacity at the saturation condition (q sat , mmol/g) of these MOF can be obtained from The adsorption energies (ΔE ads , J/mol) related to the molecular interactions between MOF surface and pharmaceutical molecules were calculated using where S phar is the pharmaceutical solubility (mmol/L) in aqueous solution. Calculated statistical physics parameters were used to interpret the pharmaceutical adsorption mechanism. They were obtained from the isotherm data correlation via a nonlinear regression where the next objective function (F obj ) was minimized where exp and mod refer to the experimental and calculated pharmaceutical adsorption capacities, respectively, and n dat is the number of experimental data used in the correlation.
Finally, Density Functional Theory (DFT) calculations were performed with GAUSSIAN09 software to understand the electrostatic and reactive properties of APAP, DFC, and NAP molecules. The molecule optimization was carried out with a functional hybrid B3LYP with 6-311++G(d, p) basis set. The charge distribution of these pharmaceutical molecules was identified with the molecular electrostatic potential (MEP), which was defined as where |r ′ − r | is the distance to point r, ρðr ′ Þdr ′ is the density of the electronic charge (measured as a volume for each element), |R A − r | is the distance from point r, R A is the position in space of nucleus A, and Z A is the atomic number of nucleus A, respectively [34]. Note that in any charge distribution, the electrons and atomic nuclei in molecules generate an electrostatic potential in the space. Therefore, this descriptor provides the electron density response when a unit of positive charge approaches. showed the characteristic diffraction peak at~11.1°2θ, thus denoting the formation of a pure phase [35,36]. This result was consistent with those reported by Chen et al. [ [45], and Chávez et al. [46]. After the adsorption of the different pharmaceutical molecules, the MOF crystalline structure was stable but the intensity of the diffraction peaks decreased. These changes in the MOF crystallinity were attributed to incorporation of each adsorbed pharmaceutical on the adsorbent surface [47,48]. X-ray diffraction pattern of the MIL-101-Fe showed diffraction peaks at~5.1, 9.2, 16.7, and 19.5°2θ, which confirmed the formation of the crystalline structure characteristic of this MOF [4,[49][50][51][52][53].
After the NAP, DFC, and APAP adsorption, FTIR spectra of these MOFs showed a decrement in the intensity of the absorption band associated with the -OH group; see Figure 2. This result suggested the formation of hydrogen bonds between pharmaceutical molecules and MOF surface [57]. Similarly, a change in the absorption band of Fe-O stretching vibration (~550 cm -1 ) was also observed, which could be an indication that the metal clusters played an important role in the adsorption of these pharmaceuticals [58]. There was also a shift in the absorption band (~1590 cm -1 ) of the stretching vibration of carboxyl groups (O-C=O) of MIL-101-Fe. According to Tomul et al. [59] and Yaah et al. [60], this change in FTIR spectrum could be also associated with the incorporation of pharmaceutical molecules in the MOF surface. To complement the surface chemistry analysis of adsorbent samples, FTIR spectra of the single pharmaceuticals were also included and compared in Figure 2. Results showed that the specific and characteristic absorption bands of pharmaceutical molecules were also identified in the spectra of MOF samples obtained after the adsorption experiments. Therefore, these findings were an indirect evidence of the adsorption of these molecules on the external surface of these MOFs thus agreeing the results reported in other studies [59,60]. Table 2 and Figures 3 and 4 show the results obtained from the elemental analysis, SEM images, and textural parameters of tested MOFs. In general, these adsorbents were mainly composed of carbon and oxygen. The presence of iron was confirmed in both MIL-100-Fe (4.9 wt%) and MIL-101-Fe (24.2%) thus providing additional evidence of the successful synthesis of these organometallic structures. Note that chlorine was also identified in the MIL-101-Fe sample because FeCl 3 ·6H 2 O was employed as precursor in its synthesis. SEM micrographs of these MOFS indicated that the size of MIL-100-Fe crystals varied from 0.05 to 0.5 μm, while the MIL-101-Fe particles showed an average diameter of 0.5-1 μm; see Figure 3. These results were consistent with other studies [36,43,53,56,61]. It was also observed that these MOFs presented an octahedral structure; however, the octahedral form of MIL-101-Fe MOF was imperfect [50-52, 62, 63]. N 2 adsorption-desorption isotherms of these MOFs are reported in Figure 4. These N 2 isotherms can be categorized between types I and IV of IUPAC classification, which are typical of micro-and mesoporous materials [23,37,41,43,52,53]. The specific surface area, total pore, micropore, and mesopore volumes of these adsorbents are given in Table 2. BET surface area of MIL-100-Fe was 768.18 m 2 /g with a total pore volume of 0.58 cm 3 /g. This surface area was similar to that reported by Nehra et al. [41] (S BET = 790:5 m 2 /g) and Bezverkhyy et al. [64] (S BET = 750 m 2 /g). However, the total pore volume differed from that obtained by these authors (i.e., 0.34 and 0.41 cm 3 /g). BET surface area and pore volume of MIL-101-Fe were~450 m 2 /g and 1.02 cm 3 /g, respectively. Similar textural parameters for this MOF have been reported by Li et al. [49] and Jiang and Li [65]. Herein, it should be noted that the MOF textural parameters depend on the synthesis route because these materials can tune their size by changing the fraction of organic connectivity and the inorganic part, which in turn is associated to the preparation conditions of each organometallic compound.  Table 3 reports and compares the adsorption capacities of MIL-100-Fe, MIL-101-Fe, and other MOFs used in the removal of different pharmaceutical from aqueous solutions [2,4,8,18,[67][68][69]. Several adsorption capacities reported in the literature were lower than those obtained for NAP, DFC, and APAP with MIL-100-Fe and MIL-101-Fe MOFs. Therefore, these materials could be considered as an alternative separation medium for the adsorption of these three pharmaceuticals in water depollution. In general, MIL-101-Fe showed a better performance than MIL-100-Fe for the adsorption of NAP and APAP, while the highest DFC adsorption capacities were obtained with MIL-100-Fe. The pharmaceutical adsorption capacities of MIL-100-Fe and MIL-101-Fe followed the next trends: APAP < <NAP < DFC and APAP < <DFC < NAP, 7 Adsorption Science & Technology respectively. It was clear that these adsorption capacities depended on both MOF and pharmaceutical properties. As stated, BET surface area of MIL-100-Fe was higher than that of MIL-101-Fe. For both MOFs, it was expected that the pharmaceutical adsorption was performed mainly on the external adsorbent surface due to the size of these organic molecules. However, the volume of mesopores of MIL-100-Fe was lower than that of MIL-100-Fe; see   Table 2. These results suggested that the adsorption of these pharmaceutical molecules was favored by large pores in MOF structure. Also, the contribution of Fe-O clusters on these MOFs played a relevant role in the adsorption of APAP and NAP pharmaceuticals where the adsorbent with the highest Fe content (i.e., MIL-101-Fe) also showed the highest removal. The solution temperature affected the pharmaceutical adsorption with these MOFs; see Figure 5. All the maximum adsorption capacities ranged from 0.14 to 2.19 mmol/g and decreased in 7.9-27% when the solution temperature increased from 30 to 40°C. Pharmaceutical adsorption on both MOFs was affected by the solution temperature according to the next trend: NAP < APAP < DFC. In particular, the removal performance of MIL-101-Fe was more sensitive to the solution temperature where its pharmaceutical adsorption capacities reduced from 13.7% for NAP to 27% for DFC. This exothermic adsorption was partially associated to the solubility of these pharmaceuticals in water and the energy exchange that occurred during the separation process [70]. Specifically, the solubility of these pharmaceutical molecules increased with the solution temperature thus affecting the intermolecular attractive forces between the MOF structure and these adsorbates [4,70]. Calculated enthalpies for the exothermic adsorption of tested pharmaceuticals on these MOFs ranged from 19.1 to 65.1 kJ/mol. These enthalpy values could correspond to an adsorption caused by hydrogen bonds, van der Waals forces, and electrostatic interactions [71][72][73].
With respect to the pharmaceutical molecular properties, the literature indicates that the molecular weight and hydrophobicity of these organic compounds could impact their adsorption on MOFs [6,8]. Figure 6 shows the MEPs of NAP, APAP, and DFC molecules that were calculated with DFT. These electrostatic potentials are illustrated via a color mapping where the scale indicates from the most reactive to the least reactive zone of the molecule (i.e., red > orange > yellow > green > blue). APAP molecule had negative charges located on the oxygen atoms, which were identified in the hydroxyl group in the para position and the oxygen of the acetate. This pharmaceutical molecule had two positive charges located on the hydrogen atoms of the secondary amine and on the hydrogen atom located in the -OH group in the para position. The remaining of this molecule showed a behavior mainly oriented towards the repulsion, which was identified by its green color [74,75]. DFC molecule showed a high chemical reactivity where the area of interest of this adsorbate corresponded to the -COOH group. The molecular area of least reactivity (blue color) was identified in the hydroxyl group of the carboxyl (mainly in the H atom), and the most reactive area (red color) was found on the oxygen atom attached to the carbon from the carboxyl group. The rest of this molecule showed an intermediate reactivity (i.e., yellow zones) [76,77]. Finally, the highest reactivity zone of NAP molecule was the oxygen of the methoxyl and -COOH groups, while the less reactivity zone was identified in the hydrogen of the -COOH group. The rest of this pharmaceutical molecule had a low reactivity [78,79]. DFT calculations also supported that the more reactive zones of these pharmaceuticals molecules could interact mainly with Fe-O clusters on MOF structure during the adsorption.  Figure 7. This statistical physics model showed R 2 ≥ 0:97 for all pharmaceutical isotherms. Modeling results indicated that the number of pharmaceutical molecules adsorbed per functional groups of MIL-100-Fe and MIL-101-Fe ranged from 1.3 to 1.9 and 1.1 to 1.6, respectively. A multimolecular adsorption of these pharmaceuticals occurred in both MOFs especially for MIL-101-Fe. As stated, MIL-101-Fe showed the highest Fe content and it could be expected that its Fe-O clusters could interact simultaneously with more than one pharmaceutical molecule during the adsorption. N ads ranged from 0.1 to 1.2 mmol/g for MIL-100-Fe and from 0.4 to 1.4 mmol/g for MIL-101-Fe, respectively. The increment of adsorption temperature reduced the number of functional groups involved in the pharmaceutical adsorption due to some adsorbate-functional group interactions were broken caused by the thermal agitation effect [33]. The saturation adsorption capacities were 0.17-1.69 mmol/g for MIL-100-Fe and 0.44-2.26 mmol/g for MIL-101-Fe. The lowest adsorption capacities at saturation were obtained for APAP with both MOFs. In particular, MIL-100-Fe showed the highest saturation adsorption capacities for DFC, while MIL-101-Fe had the highest saturation adsorption capacities for NAP. Calculated adsorption interaction energies were 10.6-22.6 kJ/mol for MIL-100-Fe and 11.4-28.6 kJ/mol for MIL-101-Fe. These pharmaceutical-MOF interaction energies were consistent with the calculated adsorption entropies thus confirming that hydrogen bonds, van der Waals forces, and electrostatic interactions could be present in the adsorption mechanism of these organic molecules.

Conclusions
The adsorption of three relevant pharmaceuticals on MIL-100-Fe and MIL-101-Fe MOFs from aqueous solutions has been experimentally studied and modeled to understand their adsorption mechanism. Results showed that the pharmaceutical adsorption capacities of these MOFs were mainly related to the presence of Fe-O clusters where their mesopore structure contributed for the removal of these organic pollutants. Adsorption properties of MIL-101-Fe outperformed those obtained for MIL-100-Fe. Overall, both MOFs showed the lowest adsorption capacities for acetaminophen molecules. The adsorption of naproxen, diclofenac, and acetaminophen on these MOFs was a multimolecular and exothermic process where more than one pharmaceutical molecule can interact with one functional group from the adsorbent surface. The pharmaceutical adsorption properties of MIL-101-Fe MOF were more sensitive to increments of solution temperature and decreased up to 27% for diclofenac molecules. MOF characterization, DFT, and statistical physics calculations indicated that hydrogen bonds, van der Waals and electrostatic forces could be the main interactions involved in the pharmaceutical adsorption on tested MOFs. MIL-101-Fe is an interesting material with better pharmaceutical adsorption capacities than other MOFs reported in literature, and it can be utilized to prepare composites and other materials for water depollution. Therefore, the experimental and theoretical results reported in this study can contribute to enhance and consolidate the application of MOFs as adsorbents of emerging compounds for water treatment and purification.

Data Availability
Data of this paper are available on request to the corresponding author.

Conflicts of Interest
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.