Adsorption Characteristics of Antibiotic Meropenem on Magnetic CoFe2O4@Au Nanoparticles

Adsorption characteristics of the antibiotic meropenem on a novel magnetic material synthesized by surface coating cobalt iron oxide (CFO) with gold nanoparticles (AuNPs) were systematically investigated. The AuNPs can enhance material adsorption capacity by having high affinity towards the thioether and amine groups in the meropenem structure. Au coverage on the CFO surface decreased the saturation magnetization from 55.8 emu/g to 48.8 emu/g, still allowing synthesized CFO@Au nanomaterials to be magnetically recoverable. The CFO@Au nanomaterials showed enhanced adsorption capacity of 25.5 mg/g at optimum conditions of pH 4.0 adsorption time 120 min, and adsorbent mass 0.05 g. Adsorption equilibrium was in accordance with a monolayer Langmuir isotherm, while the adsorption kinetics followed pseudo-first-order kinetics and intraparticle diffusion models. This work provides a simple method to prepare a magnetic composite material with high adsorption efficiency for meropenem and probably other thioether-containing substances.


Introduction
Recently, the combination of metal oxide materials with other functional components or other materials has been extensively studied to produce composite materials with novel properties and unique applicability. Due to the evident advantages of gold nanoparticles (AuNPs), structures of such nanoparticles on core-shell oxides have attracted great attention [1][2][3]. As an inert metal, Au is extremely valuable as a material coating for magnetic nanoparticle protection, surface modification versatility, strong catalytic capability, and biocompatibility [4][5][6]. Applications of Au/metal oxide nanocomposites include removal of toxic substances such as Hg, As [7,8], catalysts for organic compound reduction [9], or CO oxidation [10], analysis of various targets including catechols [11], prostate-specific antigen [10], malachite green [12], and drug carriers for biomedical applications [13,14]. Among different Au-metal oxide composites, Au/magnetic oxide nanoparticles have frequently been studied since these materials can be used in various fields such as protein purification, biological separation, target delivery, magnetic resonance imaging, therapy, and biosensing. Fe 3 O 4 and other iron-containing compounds such as CoFe 2 O 4 attract much attention thanks to the capacity for surface functionalization as well as its excellent magnetic properties. In addition, it is easy to separate these compounds from solution using an external magnetic field [4,[15][16][17][18][19][20].
Many studies have been carried out on Fe 3 O 4 /Au nanocomposites, however, not as many studies have focused on Au and CoFe 2 O 4 (CFO) composites. Among magnetic oxides, CFO has advantages of large anisotropy, high coercive field, moderate saturation magnetization, high electromagnetic performance, chemical stability, and good electrical conductivity [21][22][23]. Studies mostly focus on nanocomposite applications for analytical detection, catalysis, and drug delivery, nevertheless, applications for adsorption of -S-or NH-containing substances, exploiting the strong affinity of gold for sulfur or nitrogen atoms has been scarce.
Meropenem is a pyrrolidinyl dimethyl carbamoyl derivative of thienamycin taken from Streptomyces cattleya. It is an injectable antibiotic which currently has the broadest antibacterial spectrum in the carbapenem group [24]. Meropenem is one of the strongest antibiotics to kill harmful microorganisms which are resistant to penicillin and cephalosporin antibiotics. Particularly, the carbapenems have been used as last-resort antibiotics for patients in cases of infectious diseases in intensive care units who are seriously infected with multidrug-resistant biota [25]. So far, the analogs of carbapenems in biological and pharmaceutical matrices have mostly been determined with common separation methods include liquid-liquid extraction and solid-liquid extraction (SPE). However, such methods require several steps and have limited selectivity [26]. Sample treatment coupled with magnetic nanoparticles can provide faster phase separation by applying an external magnetic field and is easily reused. To enhance the separation efficiency, magnetic particles have been surface-functionalized with small organic molecules, polymers, aptamers, antibodies, and metal nanoparticles [27].
The present study is aimed at combining magnetic CoFe 2 O 4 with AuNPs (CFO@Au) and study the adsorption behavior of meropenem on the composite material, which can later be used for separation and enrichment of meropenem from complicated matrices. Since meropenem molecules contain thioether and amine groups, which strongly bind with the Au surface, adsorption of meropenem on Au-modified CFO particles should be enhanced in comparison with bare CFO material. Studies on the adsorptive removal of meropenem have been reported on adsorbents such as multiwalled carbon nanotubes [28] and rice husk functionalized with Mg/Fe-layered double hydroxides [29]. In these cases, the interaction between meropenem and adsorbents is mainly due to electrostatic, π-π EDA interactions, hydrophobic interactions, and hydrogen bonding to -OH and -NH-group. To the best of our knowledge, there is as yet no research examining adsorption of meropenem on CFO@Au.
The synthesis of Au/metal oxide composites usually involves mixing gold (III) salts with a suspension containing oxide nanoparticles and then reducing gold (III) by a suitable reductant such as sodium citrate [8], extract of the Allium Sp plant [9], or tetrakis (hydroxymethyl) phosphonium chloride (THPC) [12]. Aniline and dithiothreitol are used to stabilize AuNPs while substances like (3-aminopropyl) triethoxysilane (APTES) or polyethyleneimine dithiocarbamate (PEI-DTC) are used to functionalize the magnetic oxide surface to form better linkages with Au particles. In our study, CFO was prepared by a hydrothermal method, while AuNPs were prepared and stabilized on the CFO surface by using the conventional sodium citrate or sodium borohydride (NaBH 4 ) reductants and poly (diallyldimethylammonium chloride) PDADMAC polycation as a stabilizer.

Materials and Methods
2.1. Reagents and Materials. All chemicals used were analytical grade and were directly used without further purification. Meropenem sodium salt 98% ( Figure S1 3 .9H 2 O with the Co 2+ : Fe 3+ molar ratio of 1 : 2.2 was added to a beaker and adjusted to pH 12 with NaOH. This mixture was then transferred to a thermo-hydrolysis vessel and hydrolyzed at 150°C for 2 h. After being hydrolyzed, the sample was taken out and washed several times with distilled water. The precipitate was collected using magnets and dried at 80°C to produce CoFe 2 O 4 product as a black powder.

Synthesis of CoFe 2 O 4 @Au Composites
(1) PDADMAC-Stabilized Au. AuNPs were obtained by reducing HAuCl 4 with NaBH 4 in the presence of PDAD-MAC as a stabilizer [30]. The freshly prepared dispersion of CFO (50 mg in 50 mL of water) was acidified with 0.1 M of HCl at pH 6.5. While continuous stirring, successively, 152 μL of (3-aminopropyl) triethoxysilane (APTES) and 200 μL of 50 mM HAuCl 4 were added. Then, 20 μL of PDADMAC was added to the formed dark brown colloidal nanoparticle solution, and the mixture was stirred for 10 minutes. After that, 2 mL of a freshly prepared solution of NaBH 4 (0.05 M) was added dropwise into the formed reddish-brown colloidal nanoparticle solution, and then, the mixture was stirred for 3 h. The obtained substance was collected with magnets, washed with deionized water and alcohol, and then dried at 80°C for 12 h. Adsorption Science & Technology 50 mM sodium citrate was added. The solution was further heated for 20 min to form a reddish-brown colloidal solution. The obtained substance was collected with magnets, washed with deionized water and alcohol, and then was dried at 80°C for 12 h.

Material Characterization.
The crystal structures of CFO and CFO@Au were investigated by X-ray diffraction (XRD) using a D8 Advance X-ray diffractometer (Bruker) with wavelength Cu Kα = 1:5406 Å and the 2θ angle range of 20°to 80°. The morphology and particle size distribution of CFO and CFO@Au were analyzed on a JEM 1010 transmission electron microscope (TEM, JEOL) operated at an accelerating voltage of 100 keV. The magnetic hysteresis loops were measured by a vibrating sample magnetometer (VSM, Lakeshore) at room temperature and in an external field up to 5 kG. The infrared spectra were recorded on a Fourier transform infrared spectrometer (FTIR-1S, Shimadzu). The absorption spectra of AuNPs, CFO, and CFO@Au were obtained with a UV-Vis spectrophotometer (UV 1601 PC, Shimadzu) using two 1 cm matched quartz cells at room temperature.

Meropenem Determination.
The concentrations of meropenem before and after adsorption on adsorbent material are determined by capillary electrophoresis with a capacitively coupled contactless conductivity detector (CE-C 4 D).
The CE instrument was built-in house with a commercial C 4 D (ER815, eDAQ, Denistone East, NSW, Australia). Uncoated fused silica capillary (ID of 50 μm and OD of 375 μm), total length 60 cm, effective length 53 cm, was used with buffered eluent composed of 10 mM Tris adjusted to pH 8.0 with acetic acid. Injection was carried out by siphoning at a height of 20 cm for 20 s. Separation voltage was +20 kV [31].

Batch Adsorption of Meropenem on CFO@Au.
To evaluate the adsorption capacity, the adsorbent (0.03-0.1 g) was mixed with 10 mL of meropenem solution (pH from 3.0 to 8.0) having different initial concentrations (30 to 350 mg/L) in 15 mL falcon tubes. These falcon tubes were then shaken in the range of 60 to 300 min. The parameters investigated include solution pH, adsorption time, adsorbent mass, and initial concentration of meropenem. The concentrations of meropenem after adsorption were determined by using CE-C 4 D measurement. The adsorption efficiency (H % ) was evaluated by the following equation: The adsorption capacity q e (mg/g) of meropenem onto CFO@Au was calculated by the following equation: where C 0 (mg/L) and C e (mg/L) are the initial concentration and equilibrium concentrations of meropenem, respectively, V (L) is the volume of the solution, and m (g) is the absorbent mass. Kinetics experiments used initial meropenem concentration of 50 ppm, pH = 4:0 adsorbent mass of 0.05 g, and time ranging 0 to 180 min, while experiments to evaluate adsorption isotherms used initial meropenem concentration varying from 30 to 350 ppm, pH = 4:0, adsorbent mass = 0:05 g, time = 120 min.
2.6. Adsorption Isotherms. Langmuir and Freundlich models were investigated for adsorption isotherms. The Langmuir adsorption isotherm is described by the following equation [32][33][34]: where q e and q m are equilibrium and maximum adsorption capacities (mg/g), respectively, and K L is the Langmuir adsorption constant (L/mg) which relates to energy of adsorption. The Freundlich model is as follows [33,35]: where k f is Freundlich adsorption capacity (L/g) and n is an indicator for the degree of surface heterogeneity and describes the distribution of the adsorbed molecules on the adsorbent surface. The Freundlich equation is characteristic for a multilayer adsorption isotherm.

Kinetic Models.
To evaluate the rate of the adsorption process, the pseudofirst-order, pseudosecond-order, and intraparticle diffusion models are frequently applied to fit the kinetics data. The correlation coefficient (R 2 ) is computed to evaluate how successfully the model fits the data and demonstrate the kinetics of adsorption [33].
The pseudofirst-order model is described by equation (5). This model assumes the physical adsorption of one adsorbate molecule onto one active site on the adsorbent surface [33]: The pseudosecond-order model follows equation (6). The pseudosecond-order model assumes the adsorption of one adsorbate molecule onto two active sites on the adsorbent surface, preferentially through chemisorption [33]: where q e and q t are equilibrium adsorption capacity (mg/g) and adsorption capacity at different time intervals (mg/g), respectively. t is adsorption time (min), and k 1 and k 2 are pseudofirst-order (1/s) and pseudosecond-order (g/mg.min) rate constants, respectively.

Adsorption Science & Technology
Intraparticle diffusion model provides information about the diffusion of adsorbate onto the adsorbent material and whether intraparticle diffusion is the rate-limiting step. Equation (7) describes the intraparticle diffusion model: where k d is the intraparticle diffusion rate constant (1/min 1/2 ).

Comparison of Meropenem Adsorptivity between
CFO@Au-Citrate, CFO@Au-PDADMAC, CFO, and C18 Materials. The adsorption efficiencies of meropenem on CFO@Au composites, prepared by 2 methods, stabilized with citrate and stabilized with PDADMAC, were compared with bare CFO and conventional adsorption material C18 (10 mL of 50 ppm meropenem, pH 4.0; contact time 120 min, adsorbent mass 0.10 g) as presented in Figure S2. Among the 4 materials, C18 provides the lowest adsorption efficiency of 24%, followed by CFO (62%), then CFO@Au-PDADMAC (72%), and the highest efficiency was obtained for CFO@Aucitrate (94%). The modification of CFO by AuNPs enhanced the adsorption efficiency of meropenem through affinity interaction between AuNPs on the surface with thioether and amine groups in the meropenem molecular structure. The CFO and Au composites synthesized by 2 routes demonstrate not only different adsorption efficiencies but also different pH-dependent adsorption behaviors. For CFO@Au-PDADMAC, when the solution pH was varied from 3.0 to 8.0, the adsorption capacity changed insignificantly and had a maximum at pH 6.0. According to our previous research on the PDADMAC-Au system [30], the interaction between the analyte and PDADMAC is mainly hydrophobic interaction. Meropenem has isoelectric point  Adsorption Science & Technology pI~6.0, meaning that at pH 6.0, it is charge neutral. Thus, meropenem can be adsorbed and distributed very well on the hydrophobic surface of CFO@Au-PDADMAC. The maximum adsorption at pH 6.0 suggests the dominant contribution of hydrophobic interaction between meropenem and PDADMAC capped-AuNPs over the electrostatic interaction.
Meanwhile, for CFO@Au-citrate, the maximum adsorption capacity was obtained at pH 4.0 (94%). This is explained by the dominant electrostatic interaction between citrate-AuNPs and meropenem which vary with pH. Citrate-AuNPs are negatively charged at pH > pK = 3:1 of citric acid. For meropenem, at low pH, meropenem is positively charged and hence is well kept on the negatively charged surface of citrate-AuNPs. At higher pH, the positive charge is reduced; hence, the electrostatic attraction decreases, leading to decreased adsorption efficiency. Since the CFO@Aucitrate composite provides higher adsorption efficiencies than the CFO@Au-PDADMAC, in succeeding parts, we will focus our investigation only on the CFO@Au-citrate system, which is denoted simply as CFO@Au. Figure 1 [36,37]. This result indicated that the composite of CFO@Au has been successfully synthesized, and both phases are in well-developed crystalline structures.

Surface Morphology.
The TEM images of the CFO material (Figure 1(b)) and the CFO@Au composite (Figure 1(c)) reveal that CFO particles have cubic or rectangular shapes with homogenous sizes from 10-20 nm. After coating with Au, the particles remain the same size but tend to form clusters, covered by additional small and round particles, which may correspond to the morphology of single sphere AuNPs prepared by the citrate method [38,39].

UV-Vis
Spectra. UV-Vis absorption spectra of AuNPs, CFO, and CFO@Au particles dispersed in water were recorded in the wavelength range 400-800 nm ( Figure S4).

Magnetic Properties.
In the CFO@Au composites as well as of CFO@Au-meropenem, CFO plays a role as magnetic support, which helps to easily recover material after using. The magnetic properties of CFO, CFO@Au, and CFO@Au-meropenem were characterized based on magnetic hysteresis loops MðHÞ. The MðHÞ loops provided in Figure 2 show typical soft magnetic behavior. From these curves, characteristic parameters such as remanent magnetizations and coercivities were derived as M r~1 6 emu/g and H c 430 G, respectively. The saturation magnetizations M s differ among these samples as 55.8 emu/g, 51.3 emu/g, and 48.8 emu/g for CFO, CFO@Au, and CFO@Au-meropenem, respectively. The decrease in M s is due to the coverage of nonmagnetic Au as well as meropenem on magnetic core CFO. Despite this change, the composites CFO@Au and CFO@Au-meropenem still possess good magnetic properties.
After meropenem adsorption, CFO@Au powders can be easily separated from the solution after 3 min by using external magnets ( Figure S5).
From the above measurements, we confirm that the CFO@Au nanocomposite was successfully fabricated with particle size ranging from 10-20 nm, and that the material has high crystallinity and good magnetic properties. This material was further used to investigate adsorption behavior of meropenem in batch adsorption experiments.

Adsorption of Meropenem on CFO@Au
3.3.1. Effect of pH. As mentioned in Section 3.1, for CFO@Au-citrate, the highest adsorption efficiency was obtained at pH 4.0 (Figure 3(a)), which is resulted from the strongest attraction force between negatively charged Aucitrate particles and positively charged meropenem at this pH.

Effect of Adsorbent
Mass. The adsorbent mass strongly influences the number of adsorption centers, hence, strongly influences the meropenem adsorption capacity. As the adsorbent mass was varied between 0.007 and 0.07 g (see Figure 3(b)), the adsorption efficiency increased and tended to a maximum at 0.05 g. The increase of adsorption capacity with increasing adsorbent mass can be explained by the increase in the number of adsorption centers. However, when the mass of the material is above a certain amount, the adsorption capacity reaches saturation, which corresponds to the saturated adsorption of meropenem. The optimum mass adsorbent was selected as 0.05 g for further studies.  investigated in the range from 0 to 180 min with an initial meropenem concentration of 50 ppm, pH 4.0, and adsorbent mass of 0.05 g. Figure 3(c) shows that the uptake of meropenem (q t ) (mg/g) increased with increasing adsorption time and remained nearly constant after 120 min. The adsorption is rapid in the initial stages and gradually decreases with the progress of adsorption. All plots are single, smooth, and continuous curves leading to saturation.

Effect of Initial Meropenem Concentration.
Initial meropenem concentration in the solution plays an important role as the movement of the constituents in the solution controls mass transfer between the solution and the adsorbent surface. Systematic investigation on the initial meropenem concentration was further carried out in a wide concentration range from 30 to 350 ppm, at pH 4.0, adsorbent mass of 0.05 g, and time of 120 min. As can be seen in Figure 3(d), the equilibrium adsorption capacity of meropenem increases significantly at low initial meropenem concentrations from 30 to 160 ppm, from 5.43 to 21.47 mg/g, then increases more slowly and reaches a maximum of 24.89 mg/g at meropenem concentration of 350 ppm. These data are further used to deduce the adsorption isotherms of meropenem on CFO@Au.
3.4. Study on the Adsorption Isotherms. The nonlinear regression fit of experimental results with Freundlich and Langmuir adsorption isotherm models are given in Figure 4(a). Comparing the fit parameters of the two models (Table 1), the Langmuir model provides a higher correlation coefficient R 2 of 0.97693 and lower χ 2 of 1.50 than the Freundlich model (R 2 of 0.91281 and χ 2 of 5.705), indicating better experimental agreement with the Langmuir rather than the Freundlich model. Hence, we conclude that the isothermal adsorption of meropenem on CFO@Au is by monolayer adsorption rather than multilayer adsorption, and the distribution of active sites on the surface of the adsorbent is uniform. The maximum adsorption capacity calculated from the Langmuir isotherm is rather high as 25.5 (mg/g), which suggests the potential of the composite material for meropenem adsorption in biomedical applications.

Study on Kinetic Models of Static Kinetic Adsorption.
Diffusional mechanisms can be studied using three kinetic models including pseudofirst-order, pseudosecond-order, and intraparticle models. The regression correlation coefficient, R 2 , of these models was used to determine the bestfit model.  Figure 4(b) and Table 2. Both pseudofirst-order and pseudosecond-order models have high correlation factors R 2 (0.97178 for pseudofirst-order model and 0.96455 for pseudosecond-order model), indicating that both models could be satisfied and both physical diffusion and chemical adsorption via bonding between SH-group and gold surface are responsible for meropenem adsorption on CFO@Au material. Since the q e value calculated from the fitted pseudofirst-order model (10.572 mg/g) is closer to q e experimental (9.5 mg/g), the data seem to better follow the pseudofirst-order model. According to this model, the adsorption of meropenem on CFO@Au was assumed to proceed by diffusion through a boundary, and one adsorbate molecule was adsorbed onto one active site on the adsorbent surface [33].
Our observation on the simultaneous occurrence of physical diffusion and chemical adsorption agrees with other authors [28,29], however, in their studies, the pseudosecondorder model provides a better account of meropenem adsorption on rice husk functionalized with Mg/Fe layered double hydroxides [29] and on multiwalled carbon nanotubes [28].

The Intraparticle Diffusion Model.
Because the first two models did not convincingly describe the diffusion of meropenem onto the material surface, the intraparticle diffusion model was then tried as it is a widely used approach for the analysis of adsorption kinetics. This model states that adsorption changes roughly proportionally to t 1/2 rather than with contact time t. The adsorbate was most probably transported into adsorbent through an intraparticle diffusion process with three steps. The first step is instantaneous adsorption or external surface adsorption. The next step is the fast adsorption stage, where intraparticle or pore diffusion is rate-limiting, followed by the final equilibrium step, where intraparticle diffusion begins to decelerate due to extremely low adsorbate concentrations remaining in the solutions adjacent to the particles [44,45]. Adsorption data for q t versus t 1/2 is shown in three stages in Figure S6. According to this model, a plot of q t versus t 1/2 should be linear if intraparticle diffusion is involved in the adsorption process. The three steps of the adsorption process are shown by the distinct stages in Figure S6. The slope of the graph is steeper at the start, indicating a boundary layer effect. In this stage, meropenem diffuses fast to the sorbent's surface and into the pores thanks to the driving force of electrostatic Table 2: Pseudofirst-order adsorption and pseudosecond-order adsorption rate constants (C 0 = 50 ppm).

Model
Fit equation R 2 k (g/mg.min) q e (exp) (mg/g) q e (cal) (mg/g) Pseudofirst-order adsorption q t = q e 1e −k 1 t 0.97178 0.01466 9.5 10.572 Pseudosecond-order adsorption q t = q 2 e k 2 t 1 + q e k 2 t 0.96455 0.000874 9.5 14.365 8 Adsorption Science & Technology attraction between meropenem and CFO@Au and the affinity interaction between meropenem and Au surface. There is also external resistance to mass transfer around the particles [46]. The second region is the gradual adsorption stage, representing micropore diffusion. Intraparticle diffusion is the rate-limiting step in this region. There is also the third stage, as indicated in Figure S6, where intraparticle diffusion began to slow, because the adsorbent was saturated and the concentrations of solutes were low.

Conclusions
We have successfully synthesized a magnetic nanocomposite having high adsorptivity (maximum adsorption capacity of 25.5 mg/g) for meropenem, based on magnetic cobalt ferrite (CFO) and gold nanoparticles. Surface modification of CFO by AuNPs significantly enhanced the adsorption capacity efficiency of meropenem in comparison with bare CFO, thanks to affinity interaction between AuNPs on the surface with thioether and amine groups in the molecular structure of meropenem. The optimum conditions for adsorption of meropenem on CFO@Au were found to be pH 4.0, adsorption time 120 minutes, and adsorbent mass 0.05 g. The adsorption isotherm of meropenem CFO@Au follows well the Langmuir model; differences between first and second-order kinetic models are too small to resolve. Despite the presence of nonmagnetic gold particles, the composite still possesses good magnetic properties, which is helpful for quickly collecting material after adsorption. CFO@Au material can be used for the separation and enrichment of meropenem from an aqueous solution and is also a potential candidate for other applications such as biological separation and target drug delivery.

Data Availability
All the data and supporting materials are included within the article.

Conflicts of Interest
All authors declare that there are no conflicts of interest regarding the publication of this paper. Figure S1: chemical structure of meropenem. Figure S2: adsorption efficiencies of meropenem on C18, CFO, CFO@Au-citrate, and CFO@Au-PDADMAC. Figure S3: IR spectra of (a) CFO@Au; (b) CFO@Au@meropenem. Figure  S4: UV-vis absorption spectra of (1) CFO, (2) AuNPs, and (3) CFO@Au solution. Figure S5: separation of CFO@Aumeropenem from aqueous solution using a magnetic stirrer. Figure S6: intraparticle diffusion plots for the adsorption of meropenem. (Supplementary Materials)