Use of Photocatalytically Active Supramolecular Organic–Inorganic Magnetic Composites as Efﬁcient Route to Remove β -Lactam Antibiotics from Water

: Considerable efforts have been made in recent years to identify an optimal treatment method for the removal of antibiotics from wastewaters. A series of supramolecular organic-inorganic magnetic composites containing Zn-modiﬁed MgAl LDHs and Cu-phthalocyanine as photosensitizers were prepared with the aim of removing β -lactam antibiotics from aqueous solutions. The characterization of these materials conﬁrmed the anchorage of Cu-phthalocyanine onto the edges of the LDH lamellae, with a negligible part inserted in the interlayer space. The removal of the β -lactam antibiotics occurred via concerted adsorption and photocatalytic degradation. The efﬁciency of the composites depended on ( i ) the LDH: magnetic nanoparticle (MP) ratio, which was strongly correlated with the textural properties of the catalysts, and ( ii ) the phthalocyanine loading in the ﬁnal composite. The maximum efﬁciency was achieved with a removal of ~93% of the antibiotics after 2 h of reaction.


Introduction
Antibiotics are by far the most successful class of drugs used to fight against bacterial infections and, therefore, they are extensively used in both the human and veterinary medicine. They are produced by both microorganisms and synthetic or semi-synthetic routes, where, regardless of the pathway, the building of the complex chemical structures generates pollutant residues [1]. Furthermore, their excretion after their use and the disposal of unused medicinal compounds are additional sources of pollution. Some antibiotics, such as penicillin, can be easily degraded, while others, such as fluoroquinolones or tetracyclines, are considerably more persistent. Accordingly, these predominate for a longer time in the environment and accumulate in higher concentrations [1]. Thus, their potential role in promoting the development of a high level of resistance to human and animal pathogens enhances the concern about their presence in the environment.
Considerable efforts have been made in recent years to identify an optimal treatment method for the removal of antibiotics from wastewaters. With this aim, advanced oxidation processes (AOP) emerged as a more promising tool compared to traditional routes, such as adsorption on activated carbon, air stripping, or reverse osmosis [2]. Indeed, many of these techniques only transfer pollutants from one phase to another, without complete destruction up to mineralization. Biological treatment is not toxic to organic cultures, but it has some limitations [3]. The most important is related to its applicability to wastewaters containing biodegradable compounds.
with their characterization and catalytic behavior, for the photodegradation of β-lactam antibiotics under visible light. Each component in the system contributes to the photocatalysts' efficiency in the removal of antibiotics: LDH serves as a support for Pc, and is also responsible for the adsorption of the antibiotics from the environment; Pc serves as photosensitizer, and the MNPs make possible the rapid extraction of the composite from the reaction environment.

Results and Discussion
The coding of the catalysts based on their preparation conditions is presented in Table 1. As previously reported [40], the X-ray diffractogram of the bare LDH shows (Figure 1a), at low angles, narrow and intense diffraction lines and, at high diffraction angles, wide and less intense lines characteristic of the LDH materials [41]. In addition to these lines, additional lines corresponding to the zincite phase (ZnO) appeared, as minor impurities, in the 2 theta range 31-38 • [40]. The reconstruction by memory effect occurred with the preservation of the lines of the stable ZnO phase (ICDD 005-0664). The X-ray diffractograms corresponding to P1-P8 catalysts are presented in Figure  1b. The magnetite phase was observed in almost all the samples and was most predominant for P1, P2, P5, and P6. Apart from LDH, Fe3O4 (ICDD 19-0629), and small impurities of ZnO, no other phases or impurities were detected.
The network parameter "a" (Table 2), calculated as a = 2·d110, indicating the average cation--cation distance in the layered network, showed a decrease in inter-layer space, as 10 Figure 1. X-ray diffractograms corresponding to (a) as-prepared and rehydrated LDH after deposition of CuPc and (b) P1-P8 catalysts.
The X-ray diffractograms corresponding to P1-P8 catalysts are presented in Figure 1b. The magnetite phase was observed in almost all the samples and was most predominant for P1, P2, P5, and P6. Apart from LDH, Fe 3 O 4 (ICDD 19-0629), and small impurities of ZnO, no other phases or impurities were detected.
The network parameter "a" (Table 2), calculated as a = 2·d 110 , indicating the average cation-cation distance in the layered network, showed a decrease in inter-layer space, as an effect of the hydration associated with the elimination, by calcination, of the larger carbonate anions that were larger, and the replacement of these by hydroxyl anions, which are much smaller in size. The network parameter "c", calculated as c = 3/2·(d 003 + 2d 006 ), keeps the trend of IFS variation caused by an increase in the electrostatic forces between the layers and the interlayer anions. According to the literature, the presence of heavy interlayer molecular species led to a change in the electron density of the 00l harmonics, i.e., organic anions [42], metal nanoparticles [43], or oxometalates [44]. Considering the "a" and "c" parameters, we can conclude that the majority of phthalocyanine is anchored on the edge of the hydroxide lamellae, while a smaller part is placed in the interplanar space. The crystallite size, which in this case was an average of the entire composite, depended on the LDH: MP ratio. The bare LDH showed a wide absorption band in the range of 240-380 nm, with maxima at 348 and 359 nm corresponding to a zincite phase in the layered structure ( Figure 2a) [40]. The additional band at 212 nm corresponds to Mg(OH) 2 /MgO [40,46], while those in the region 550-750 nm to the strong Q bands, assigned to the CuPc [47]. The DR-UV-Vis spectra of the phthalocyanine-LDH composites are presented in Figure 2b. These disclose B-bands of CuPc (300-350 nm) [48] and magnetite 500 nm [49].

Diffuse-Reflectance Infrared Spectroscopy (DRIFT)
The intercalation of phthalocyanine was also confirmed by the DRIFT analysis ( Figure 3). Regarding the LDH (Figure 3a), the wide absorption band located in the 3700-3400 cm −1 range corresponds to the vibrations of the OH, ν(O-H) groups in the interplanar space, while the band at 3000 cm −1 was assigned to the hydrogen bonds between the water and the carbonate anion [50]. The band at 1638-1650 cm −1 corresponded to the vibrations of the water molecules in the layered structure, that in the range 1100 and 650 cm −1 to the vibrations of the CO 3 2− groups, and those before 600 cm −1 to the Mg-O, Zn-O and Al-O bonds. The infrared spectra also showed some bands associated with the presence of LDH-anchored phthalocyanine [51] (Figure 3b), although their intensity was very low. For example, the band at approximately 1650 cm −1 was be assigned to the C=N bond, that at 1120 cm −1 to the S=O bond, and those at 730, 1033 and 1090 cm −1 to the C-H bond. The bare LDH showed a wide absorption band in the range of 240-380 nm, with maxima at 348 and 359 nm corresponding to a zincite phase in the layered structure ( Figure 2a) [40]. The additional band at 212 nm corresponds to Mg(OH)2/MgO [40,46], while those in the region 550-750 nm to the strong Q bands, assigned to the CuPc [47]. The DR-UV-Vis spectra of the phthalocyanine-LDH composites are presented in Figure  2b. These disclose B-bands of CuPc (300-350 nm) [48] and magnetite 500 nm [49].

Diffuse-Reflectance Infrared Spectroscopy (DRIFT)
The intercalation of phthalocyanine was also confirmed by the DRIFT analysis ( Figure 3). Regarding the LDH (Figure 3a), the wide absorption band located in the 3700-3400 cm −1 range corresponds to the vibrations of the OH, ν(O-H) groups in the interplanar space, while the band at 3000 cm −1 was assigned to the hydrogen bonds between the water and the carbonate anion [50]. The band at 1638-1650 cm −1 corresponded to the vibrations of the water molecules in the layered structure, that in the range 1100 and 650 cm −1 to the vibrations of the CO3 2− groups, and those before 600 cm −1 to the Mg-O, Zn-O and Al-O bonds. The infrared spectra also showed some bands associated with the presence of LDH-anchored phthalocyanine [51] (Figure 3b), although their intensity was very low. For example, the band at approximately 1650 cm −1 was be assigned to the C=N bond, that at 1120 cm −1 to the S=O bond, and those at 730, 1033 and 1090 cm −

Textural Properties
The surface areas of the P1-P8 catalysts ( Table 2) varied in a rather narrow range, i.e., from 110 (P4) to 160 m 2 ·g −1 (P5). The increase in the ageing time, which also meant an increase in the MNP size, led to a decrease in the surface area (samples P1 and P2). The decrease of the surface area compared to the parent LDH was due to the tendency to clusterize in large conglomerates systems confirmed by TEM (see Supplementary Materials) [52]. All the samples presented a Type IV adsorption-desorption isotherm, characteristic of mesoporous materials ( Figure 4). However, depending on the LDH: MNP ratio and the nature of MNP precursors, the variation in the pore size was larger (Table 3, Figure 4).  Figure 3. DRIFT spectra collected for (a) as-prepared and rehydrated LDH and (b) P1-P8 catalysts and commercial CuPc tetrasulfate.

Textural Properties
The surface areas of the P1-P8 catalysts ( Table 2) varied in a rather narrow range, i.e., from 110 (P4) to 160 m 2 ·g −1 (P5). The increase in the ageing time, which also meant an increase in the MNP size, led to a decrease in the surface area (samples P1 and P2). The decrease of the surface area compared to the parent LDH was due to the tendency to clusterize in large conglomerates systems confirmed by TEM (see Supplementary Materials) [52]. All the samples presented a Type IV adsorption-desorption isotherm, characteristic of mesoporous materials (Figure 4). However, depending on the LDH: MNP ratio and the nature of MNP precursors, the variation in the pore size was larger (Table 3, Figure 4).

Degradation of Antibiotics
The experiments carried out in this study for the removal of antibiotics from water using the organic-inorganic magnetic composite photocatalysts indicated two parallel processes, namely, adsorption and photocatalytic degradation ( Figures 5 and 6, respectively).
Simple adsorption measurements (i.e., in the absence of photo-irradiation) indicated a removal of ampicillin of 9-87%. Depending on the catalyst, the removal of the ampicillin varied in the following order: P7 > P3 > P8 > P5 > P6 > P4 > P2~P1, i.e., in good concordance with their textural properties (pore volume and interlayer diameter). Although P3 had the highest pore volume in the series, it presented some interlayer limitation disfavoring the adsorption of ampicillin. For mesoporous supports with larger pores (200-700 nm), the loading of the adsorbed ampicillin does not depend on the pore size, and it was related only to the surface charge density (σ) of the sorbent surface [53]. For the LDHs, the ampicillin-related amoxicillin was found to adsorb predominantly by a chemisorption mechanism, following pseudo-second-order kinetics [54]. Indeed, as shown in Table 3, the interlayer distance did not influence the mass transfer or the access to the chemisorption sites.
The photo-irradiation of the system changed the order established by the adsorption. However, the photo-catalytic degradation could not be separated from the adsorption of the ampicillin and degraded fragments, and it was also dependent on the photo-irradiation energy. Under the LED irradiation (the initial concentration of ampicillin in solution at t = 0), the maximum degradation was achieved for P7, with a removal of ~93% after 2 h of reaction. Overall, summarizing the effects of the adsorption and photocatalytic degradation, the removal decreased in the following order: P7 > P5 > P8~P3 > P6 > P4 > P2 > P1 ( Figure 5). The performances of these catalysts depended both on the LDH:MNP ratio (which was also associated with the amount of CuPc) and the textural properties of the composite.

Degradation of Antibiotics
The experiments carried out in this study for the removal of antibiotics from water using the organic-inorganic magnetic composite photocatalysts indicated two parallel processes, namely, adsorption and photocatalytic degradation ( Figures 5 and 6, respectively).
Simple adsorption measurements (i.e., in the absence of photo-irradiation) indicated a removal of ampicillin of 9-87%. Depending on the catalyst, the removal of the ampicillin varied in the following order: P7 > P3 > P8 > P5 > P6 > P4 > P2~P1, i.e., in good concordance with their textural properties (pore volume and interlayer diameter). Although P3 had the highest pore volume in the series, it presented some interlayer limitation disfavoring the adsorption of ampicillin. For mesoporous supports with larger pores (200-700 nm), the loading of the adsorbed ampicillin does not depend on the pore size, and it was related only to the surface charge density (σ) of the sorbent surface [53]. For the LDHs, the ampicillinrelated amoxicillin was found to adsorb predominantly by a chemisorption mechanism, following pseudo-second-order kinetics [54]. Indeed, as shown in Table 3, the interlayer distance did not influence the mass transfer or the access to the chemisorption sites. Moving to a solar simulator as an irradiation source ( Figure 6) afforded a higher removal efficiency, which was correlated with the higher photocatalytic efficiency attributed to the absorption of the light via the Q bands of the phthalocyanine. The stability and reusability of catalysts are very important factors in the characterization of their efficiency and the limits of their applicability in real applications. Based on these considerations, the composite P7 demonstrated an enhanced efficiency. After three runs (the catalyst was magnetically removed from the reaction environment washed thoroughly with deionized water, and then dried at 120 °C for 4 h) the conversion of the ampicillin was conserved at 98%. After each run, the DRIFT analysis con-  Moving to a solar simulator as an irradiation source ( Figure 6) afforded a higher removal efficiency, which was correlated with the higher photocatalytic efficiency attributed to the absorption of the light via the Q bands of the phthalocyanine. The stability and reusability of catalysts are very important factors in the characterization of their efficiency and the limits of their applicability in real applications. Based on these considerations, the composite P7 demonstrated an enhanced efficiency. After three runs (the catalyst was magnetically removed from the reaction environment, washed thoroughly with deionized water, and then dried at 120 °C for 4 h) the conversion of the ampicillin was conserved at 98%. After each run, the DRIFT analysis confirmed no leaching of the phthalocyanine (Figure 7).  The photo-irradiation of the system changed the order established by the adsorption. However, the photo-catalytic degradation could not be separated from the adsorption of the ampicillin and degraded fragments, and it was also dependent on the photo-irradiation energy. Under the LED irradiation (the initial concentration of ampicillin in solution at t = 0), the maximum degradation was achieved for P7, with a removal of~93% after 2 h of reaction. Overall, summarizing the effects of the adsorption and photocatalytic degradation, the removal decreased in the following order: P7 > P5 > P8~P3 > P6 > P4 > P2 > P1 ( Figure 5). The performances of these catalysts depended both on the LDH:MNP ratio (which was also associated with the amount of CuPc) and the textural properties of the composite.
Moving to a solar simulator as an irradiation source ( Figure 6) afforded a higher removal efficiency, which was correlated with the higher photocatalytic efficiency attributed to the absorption of the light via the Q bands of the phthalocyanine.
The stability and reusability of catalysts are very important factors in the characterization of their efficiency and the limits of their applicability in real applications. Based on these considerations, the composite P7 demonstrated an enhanced efficiency. After three runs (the catalyst was magnetically removed from the reaction environment, washed thoroughly with deionized water, and then dried at 120 • C for 4 h) the conversion of the ampicillin was conserved at 98%. After each run, the DRIFT analysis confirmed no leaching of the phthalocyanine (Figure 7). Similar tests on the removal of amoxicillin under simulated solar-light irradiation ( Figure 8) using selected composites confirmed the order of activity determined for the ampicillin. However, both the absorption part and the photocatalytic behavior showed some differences. These were directly related to the structural features of these antibiotics (Figure 8), which can affect both the adsorption of molecules and photocatalytic degradation processes. Thus, the degradation of β-lactam antibiotics can follow two routes: hydroxylation and opening of the β-lactam ring [55]. The ratio between these routes is influenced by the structure of the antibiotic and the nature of the catalyst. For example, hydroxylation can be enhanced for amoxicillin compared to ampicillin by the population and the strength of the hydroxyl groups able to activate the aromatic ring, making it more susceptible to their attachment. However, the mechanism of the photo-degradation of β-lactam antibiotics is still under discussion.  Similar tests on the removal of amoxicillin under simulated solar-light irradiation ( Figure 8) using selected composites confirmed the order of activity determined for the ampicillin. However, both the absorption part and the photocatalytic behavior showed some differences. These were directly related to the structural features of these antibiotics (Figure 8), which can affect both the adsorption of molecules and photocatalytic degradation processes. Thus, the degradation of β-lactam antibiotics can follow two routes: hydroxylation and opening of the β-lactam ring [55]. The ratio between these routes is influenced by the structure of the antibiotic and the nature of the catalyst. For example, hydroxylation can be enhanced for amoxicillin compared to ampicillin by the population and the strength of the hydroxyl groups able to activate the aromatic ring, making it more susceptible to their attachment. However, the mechanism of the photo-degradation of β-lactam antibiotics is still under discussion.
FTIR analysis of the amoxicillin solution in the presence of P7 sample at each reaction time shows an increase in the amount of bonds involving oxygen (Figure 9a-c), which back up both reaction pathways. dation processes. Thus, the degradation of β-lactam antibiotics can follow two routes: hydroxylation and opening of the β-lactam ring [55]. The ratio between these routes is influenced by the structure of the antibiotic and the nature of the catalyst. For example, hydroxylation can be enhanced for amoxicillin compared to ampicillin by the population and the strength of the hydroxyl groups able to activate the aromatic ring, making it more susceptible to their attachment. However, the mechanism of the photo-degradation of β-lactam antibiotics is still under discussion.

Synthesis
The synthesis of LDH Mg0.375Zn0.375Al0.25 was carried out following a previously reported procedure [40]. Therefore, two solutions (A and B), were mixed at an inlet flow of 1 cm 3 ·min −1 , at RT under stirring (600 rpm). Solution A of a concentration of 1.5M con-

Synthesis
The synthesis of LDH Mg 0.375 Zn 0.375 Al 0.25 was carried out following a previously reported procedure [40]. Therefore, two solutions (A and B), were mixed at an inlet flow of 1 cm 3 ·min −1 , at RT under stirring (600 rpm). Solution A of a concentration of 1.5M contained the nitrate-like precursors of Mg, Zn, and Al in the molar ratio Mg/Zn/Al = 0.375/0.375/0.25. The base solution contained 0.23 moles NaOH and 0.092 moles Na 2 CO 3 at a concentration of 1M in Na 2 CO 3 . Co-precipitation of these took place for 1.5 h at the pH of 10 ± 0.2 (continuously monitored by a pH-meter, pH315i), after which the resulted precipitate was aged for 18 h, at 80 • C, in air atmosphere at the same stirring rate. The suspension was then cooled at room temperature, filtered under vacuum, and washed with distilled water until reaching a pH of 7. The precipitate was dried at 120 • C for 24 h in air atmosphere, resulting in LDH-MgZnAl. The insertion of phthalocyanine into the structure of LDH started from the calcination of LDH for 18 h at 460 • C (10 • C/min), followed by stirring of the resulting mixed oxide with a solution of Co-phthalocyanine 3,4 ,4 ,4 -tetrasulfonic acid (0.1 M) for 12 h, filtration of the composite phthalocyanine-LDH, washing of the precipitate, and drying at 120 • C for 24 h in air. The amount of CuPc determined by elemental analysis was 10 × 10 −5 moles CuPc/g of LDH.
The attachment of the magnetic particles (MPs) was achieved by mixing 100 mg of modified LDH with Cu-phthalocyanine in 100 mL of water. After vigorous stirring for 10 min, the corresponding amounts of FeCl 2 and Fe(NO 3 ) 3 were added, resulting in a MP@CuPc@LDH-MgZnAl composite. The mixture was then heated at 70 • C in an oil bath and treated, by a slow addition, with 7.5 mL of 5M NaOH. The resulting suspension was then left to age for 60 min. Finally, the solid was magnetically removed, washed with water by decantation, and dried at 70 • C for 12 h.
In order to verify the structural influence of the LDH: MNPs ratio and the nature of the MNP precursors, a series of materials containing CuPc@LDH-MgZnAl were prepared using a similar protocol (Table 1).

Specific Surface Area (BET) and Pore-Size Measurements
The investigation of the textural characteristics was performed using the 77K nitrogen adsorption-desorption isotherms collected using a Micromeritics ASAP2020 Surface Area and Porosity Analyzer. The samples were outgassed under vacuum for 24 h at 120 • C.

X-ray Diffraction (XRD)
XRD data were collected at room temperature using a Shimadzu XRD-7000 apparatus taking the monochromatic Cu K α radiation (λ = 1.5406 Å, 40 kV, 40 mA), with a scanning rate of 0.1 degrees per minute, in the range 2θ = 5-80 degrees.
The size of the crystallites was estimated using Sherrer's equation: where: k = 0.9 (shape factor) λ = wavelength of the X-ray radiation source (for Cu = 1.54 Å) FWHM = full width at half height (in radians) θ = maximum position (in radians)

Diffuse-Reflectance UV-Vis Spectroscopy (DR-UV-Vis)
The DR-UV-Vis spectra were collected under ambient conditions using Specord 250 equipment (Analytic Jena) with an integrating sphere as a measuring device in the reflectance mode. MgO was used as reference material.

Diffuse-Reflectance Infrared Spectroscopy (DRIFT)
The infrared spectra were collected with a Brucker Tensor II device equipped with a Harrick Praying Mantis Diffuse Reflection accessory. The final spectra averaged 128 scans with a resolution of 4 cm −1 . KBr was used as a reference.

Photocatalytic Tests
The irradiation was generated by a sunlight simulator using a Luzchem LZC-4b source (LED 445-465 nm), or a Sciencetech SF150-A Small Collimated Beam Solar Simulator (equipped with Air Mass AM1.5G Filter and Light-Tight Reaction Chamber). The evolution of the photocatalyzed reaction was followed by means of a liquid chromatograph (HPLC) equipped with Zorbax SB-C18 column (4.6 × 150 mm, 5 microns), mobile phase 25 mM KH 2 PO 4 :ACN = 60:40, mobile phase flow 0.5 mL·min −1 ; analysis time 30 min; column temperature 60 • C, DAD detection 204 nm). In a typical experiment, 15 mL of a prepared 0.015 M solution of antibiotic and 60 mg of catalyst were added into a quartz test tube. To establish the absorption-desorption equilibrium; before turning on the lamp, the samples were kept in the dark, under stirring, for 10 min.

Conclusions
A series of supramolecular organic-inorganic magnetic composites containing Znmodified MgAl-LDHs and Cu-phthalocyanine as photosensitizers were prepared with the aim of removing β-lactam antibiotics from aqueous solutions. The resulting composites were characterized using textural, spectral, and diffractometric techniques. These confirmed the anchorage of Cu-phthalocyanine onto the edges of the LDH lamellae, without any insertion in the interlayer space. The removal of these antibiotics occurred via concerted adsorption and photocatalytic degradation. The efficiency of the composites depended on (i) the LDH:MP ratio, which was strongly correlated with the textural properties of the catalysts, and (ii) the phthalocyanine loading in the final composite. The maximum efficiency was achieved for the P8 photocatalyst, with a removal of~93% ampicillin after 2 h of reaction.