Removal of metronidazole antibiotic from aqueous solution by ammonia-modified activated carbon: adsorption isotherm and kinetic study

This article was aimed at investigating the removal of metronidazole (MNZ) from aquatic solutions by modified activated carbon (MAC) with amine groups. The effect of various parameters on the adsorption rate such as the initial pH, adsorbent dose and initial concentration of MNZ and contact time were scrutinized. MAC was characterized by Fourier transform infrared spectroscopy and Brunauer–Emmett–Teller techniques. The obtained results illustrated that under the optimum conditions (pH1⁄4 3, contact time1⁄4 50 min, initial MNZ concentration1⁄4 5 mg/L and MAC dose1⁄4 0.5 g/L), the maximum adsorption efficiency was 95%. Furthermore, the kinetic studies indicated the applicability of the pseudo-second-order kinetic model, whereas the adsorption isotherm fitted well with the Freundlich model (0.996), and the maximum adsorption capacity was 66.22 mg/g. The SBET and the total pure volume of MAC were 706.92 m /g and 0.532 cm/g, respectively. Also, the regeneration tests demonstrated that MAC had good stability after five cycles (73%). It can be concluded that MAC, as an effective adsorbent, has a high ability to remove MNZ from aqueous solutions.

Several methods such as the direct UV photolysis (Dantas et al. 2010), ozonation process (Kermani et al. 2013), solar photoelectro-Fenton (Pérez et al. 2015), bio sorbent (Azarpira & Balarak 2016), photocatalytic process (Wang et al. 2010) and adsorption process (Carrales-Alvarado et al. 2014) have been employed for the removal of MNZ from aqueous solutions. However, most of the mentioned methods are not suitable due to low efficiency, high operational cost and high energy requirement, and some of them are not eco-friendly (Jung et al. 2013). Among these methods, the adsorption process benefits from some unique advantages such as low costs, high flexibility, simple navigation, insensibility, high capability, reusability and eventually recycling of valuable pollutants (Su et al. 2011). Activated carbon as a sorbent is a suitable alternative for the effective removal of organic pollutants from aqueous media along with the porosity structure and high surface area. But the problems such as filtration, dispersion and turbidity, and the high cost of recovery lead to its application restriction on large scales (Ai & Jiang 2010). Recently, the modification of activated carbons with different methods has been applied to overcome its circumscription. The main reason for surface oxidation and the development of oxygen surface groups is the role of these functionalities as an intermediate stage to develop some oxygenated anchoring sites before introducing nitrogen functionalities to carbon surface (Pittman et al. 1997). When oxidized carbons are treated via ammonia at high temperatures, free radicals (such as NH 2 , NH and atomic hydrogen), which are created during ammonia decomposition, may attack the surface oxides and active sites present on the carbon surface to form nitrogen-containing functional groups (Stöhr et al. 1991;Shafeeyan et al. 2011). Numbers of studies have reported on the use of different adsorbents for the removal of MNZ from aqueous solutions. Flores-Cano et al. investigated the removal of MNZ on activated carbons derived from coffee residues and almond shells (Flores-Cano et al. 2016). In another research, Sun et al. studied the adsorption of MNZ onto biochar derived from Sugarcane Bagasse (Sun et al. 2018). However, there is no report on the removal of MNZ by functionalization of commercial activated carbon using ammonia. Therefore, the aim of this study was to investigate the performance of functionalization of commercial activated carbon using ammonia for the removal of MNZ from aquatic solutions.

Reagents, materials and solutions
In this study, all chemicals materials were purchased from Merck, and MNZ of purity 97% (C 6 H 9 N 3 O 3 ; MW ¼ 171.2 g/mol) was bought from Sigma Aldrich Co in Iran. Table 1 shows the characteristics and chemical structure of MNZ (Shemer et al. 2006;Zarei et al. 2018). Also, H 2 SO 4 and NaOH were obtained from Merck Germany laboratory with a degree of purity to adjust the pH of the solution. The stock of synthetic MNZ (1,000 mg/L) was made by dissolving the required amount in deionized water and was kept in a glass container and kept in the dark at below 4°C. It should be noted that in all stages of the experiments, double distilled water was used.

Preparation of modified activated carbon
All of the required activated carbons were prepared from Merk Company. First, for the production of soluble mediator's amine, 98 mL of chlorohydrin solution was contacted with 152 mL of diethylamine solution in 150 mL of dimethylformamide (DMF) into an evaporator at 85°C for 120 min. In the second step, to make modified activated carbon (MAC), 20 g of Water solubility (g/L) 9.5 powdered activated carbon (PAC) was mixed with 60 mL of the prepared amine solution, and 20 mL of pyridine was used as a catalyst. This stage was performed in an evaporator at 80°C for 2 h. Next, for the elimination of residual chemicals, MAC was washed with 0.1 N HCl and NaOH, and after that, distilled water was used for further cleanup of the adsorbent; then, natural pH was achieved. Finally, the synthesized MAC was dried at 50°C in 24 h and then stored for further use (Gholami et al. 2018).

Characterization of MAC
The Fourier transform infrared spectroscopy (FTIR, Perkin Elmer) was prepared by using potassium bromide (KBr) as the reference and analyzed with a 1 cm À1 resolution in a wide range of frequency of 450-4,000 cm À1 . The nitrogen adsorption/desorption isotherms were conducted by the Brunauer-Emmett-Teller (BET) analysis (II BELSORP mini) at 77 K to determine the pore volume and specific surface area of MAC. The point of zero charge (pH pzc ) was investigated for the determination of adsorbent surface charge. The scanning electron microscope (SEM) images of the samples were obtained on a MIRA3 TESCAN at an accelerating voltage of 30 kV. X-ray diffractometer (XRD) analysis was examined with US Intel diffractometer EQuniox 3000 in the 2θ range between 10 and 80°with a scanning speed of 0.01 to ensure the formation of the activated carbon.

Adsorption experiments
The effect of experimental parameters, namely, pH (3, 5, 7, 9 and 11), adsorbent dose (0.1, 0.25, 0.5, 0.75 and 1 g/L), contact time (10, 20, 30, 40, 50, 60 and 70 min), initial concentration (5, 10, 20, 30 and 40 mg/L) and regeneration, was studied. The pH of the solution for each experiment was adjusted by using 0.1 M NaOH or H 2 SO 4 . The solutions were agitated for achieving equilibrium time at room temperature at a constant speed (250 rpm). After the completion of the reaction time, centrifugation was done to isolate the solvent absorber with 5,000 rpm for 10 min and filter using a Whatman paper (0.2 μm), and then the filtered sample was analyzed by a UV-visible spectrophotometer (model DR 5000) at the maximum absorption wavelength to determine the residual MNZ concentration. After equilibrium, the removal efficiency and the adsorption capacity (q e , mg/g) were calculated using the following equations, respectively (Afshin et al. 2019;Rashtbari et al. 2020aRashtbari et al. , 2020b: where C o and C e are the initial dye concentration and the MNZ concentration at the time of equilibrium (mg/L), V is the volume of the solution (L) and M is the mass of the adsorbent (g).

Determination of MNZ concentration
The measurement of MNZ concentration in this study was carried out using a UV-visible spectrophotometer (model DR 5000) at 320 nm wavelength according to the Standard Methods for the Examination of Water and Wastewater (Nasseh et al. 2019).

Determination of pH pzc
One of the most important characteristics of the adsorbents is pH pzc , showing the state of electrical charge dispersion on the adsorbent surface. To determine pH pzc , 30 mL of a 0.1 M solution of sodium salt was poured into 11 100-mL Erlenmeyer flasks, and the pH values of the solutions were adjusted between 2 and 12. Then, 0.05 g of the MAC prepared in each of the flasks was added. Next, the solutions were placed on a shaker at a speed of 250 rpm for 48 h. Later, the final pH of the solutions was measured after the adsorption separation. Finally, pH pzc was determined after plotting the final pH change curve against the initial pH (Rivera-Utrilla et al. 2010).

Determination of the structural characteristics of the adsorbent
To study the nature of the functional groups, present on the surface of the samples, the FTIR analysis was carried out. Figure 1(a) shows the results of FTIR measurements taken for the untreated carbon and modified with amine groups samples.
The broad peaks detected at 1,570-1,700 cm À1 were weaker because of the reduction of the C-O bond. As can be seen, there were a few weak peaks, corresponding to the existence of a few functional groups. The broad bands that appeared at 3,426 cm À1 for PAC could be attributed to O-H vibration in the H 2 O molecule (Huang et al. 2014). Ammonia treatment caused some changes in the spectra of modified samples. In the range of 3,376-3,294, overlapping bands of O-H and N-H stretching vibrations were observed for all the modified samples (Shafeeyan et al. 2011). Absorption at wavenumbers of 2,930 cm À1 for PAC and 2,941 cm À1 for MAC are attributed to the aliphatic C-H interaction with the surface of the adsorbent (Moussavi et al. 2013). MAC displays the presence of one strong peak at 2,360 cm À1 which corresponds to the C-C stretching vibrations in alkyne groups. In addition, the spectrum of MAC exhibits the appearance of new bands related to the N-containing species at 1,542 cm À1 (cyclic amides and 1,052 cm À1 ); C-N of the amine group implies that the use of ammonia modification produced new nitrogen surface complexes. The peak created around the 577.18 cm À1 bands relates to the presence of amine groups (CN tensile vibrations), which are coated on the activated carbon. Thus, the results of FTIR showed that amine groups were well coated on active carbon (Wang et al. 2007;Moussavi et al. 2013). Figure 1(b) shows the XRD patterns of MAC in the 2θ range from 20 to 80°. It was found that there was a series of conventional broad peaks at 23.1, 24.2 and 26.5°w hich are a special characteristic of the activated carbon (Xu et al. 2017). Figure 1(c) and 1(d) indicates the images of MAC and PAC by the SEM (Shemer et al. 2006). The SEM is a primary tool used for the characterization of the surface morphology and fundamental physical properties of adsorbent surfaces. It is useful for the determination of particle size, shape and porosity. The surface morphology of both adsorbents was investigated using SEM micrographs. The SEM micrograph of both carbon types (Figure 1(c)) indicated that PAC has a form like compressed and porous fibers with a series of parallel and long channels, but MAC (Figure 1(d)) had a smooth surface with scattered holes. In this study, the specific surface areas (S BET ) and pore size distribution of the adsorbents were calculated by the multipoint BET and Barrett-Joyner-Halenda methods. The specific surface area (S BET (and pore volume of adsorbents have been compiled in Table 2. The results obtained showed that the SBET and the total pure volume for the MAC and PAC were 706.92, 738.18 m2 g À 1 and 0.532, 0.551 cm 3 g À1 , respectively. It was found that modification slightly decreased the apparent surface area, total pore and micropore volumes. For instance, the surface area of the modified adsorbent was 706.92 m 2 g À1 , which represented 95.7% of the surface area of the untreated adsorbent. The decrease in porous texture properties is probably due to the partial blockage of the micropore entrances by nitrogen functional groups that occur during modification (Pereira et al. 2003). The MAC and PAC showed an average pore diameter of 3 nm, which can allow ions to penetrate easier into their porous surface.

Effect of solution pH
The initial pH is one of the main parameters that affect the adsorption process with the effect on solution chemistry and the activity of functional groups on absorbent surfaces. Figure 2 shows the effect of the pH of the solution on the removal of MNZ by MAC. Note that for the removal, the initial pH of the solutions was adjusted from 3 to 11 to determine the optimum pH. The efficiency declined with increasing the pH value. For example, at pH ¼ 3 and 11, the percentages of MNZ removal were 97 and 65%, respectively. Therefore, pH ¼ 3 was selected for the next experiments. This can be attributed to pH pzc (pH value of zero charges, pH at which the flux transfer occurs from the surface). Normally, adsorption is fully dependent on the pH pzc value of adsorbent, which in this study, was found to be 6.83 for MAC (Figure 2(b)); a positive charge developed on the surfaces at pH below pH pzc . The high removal efficiency of antibiotics in acidic conditions indicates the adsorption of the contaminant on adsorbents through its carboxyl groups (Legnoverde et al. 2014). In basic conditions, the levels of electrostatic discharging strength between antibiotics and negative sites are reduced on the adsorbent surfaces, leading to a decrease in the removal efficiency of pollutants from the solution. The number of OH À groups increased with increasing the pH of the solution, which generally resulted in a decrease in positive reception sites, and consequently, the effectiveness Table 2 | BETspecific surface area (S BET ), total pore volume (V Total ), micropore volume (V micro ) and mesopore volume (V meso ) of PAC and MAC D p (nm) V meso (cm 3 /g) V micro (cm 3 /g) V Total (cm 3 /g) S BET (m 2 /g) Material  of MNZ elimination went down (Pouretedal & Sadegh 2014). The behavior of MNZ sorption with the changing pH in this study was similar to that reported by Wang who showed that the adsorption of MNZ, at pH values around 3.3-5, was higher than that of pH range from 3.33 to 9.30 ).

Effect of adsorbent dosage
As shown in Figure 3, the amount of adsorbent dose is an important factor that affects the percentage of removal efficiency. In this study, the removal percentage of MNZ increased from 75 to 96% by raising the MAC dose from 0.1 to 0.5 g/L. The results of the present work indicated that the percentage of MNZ removal enhanced with increasing adsorbent dose. This can be attributed to the higher adsorbent surface area, as well as the availability of more binding sites due to an increase in the amount of absorbent dosage (Ghaedi et al. 2013). Furthermore, the absorption capacity reduced with an increase in dose, so that the absorption capacity of MNZ reduced from 37.54 to 4.91 mg/g in doses from 0.1 to 1 g/L. Therefore, 0.5 g/L of the adsorbent was selected as the optimal dose. Although with an increase in absorbent dose the removal efficiency increased due to the saturation of some sites on the adsorbent surface, the adsorption capacity decreased (Hoseinzadeh et al. 2013;Khosravi et al. 2018). The results of this study were consistent with the results obtained by Khosravi et al., who used modified montmorillonite with ZnO and TiO 2 nanoparticles in the presence of H 2 O 2 to absorb cephalexin from aqueous solutions; they reported that the percentage of cefalexin, which was %33, increased with increasing the amount of GLM from 0.4 to 4 g/L (Khosravi et al. 2018).

Effect of contact time and initial MNZ concentration
The contact time between absorbent and adsorbate as well as the initial concentration of MNZ are important parameters in the adsorption process. The effect of contact time on MNZ adsorption on MAC was examined at various concentrations of MNZ (Figure 4). The adsorption rate of MNZ at various initial concentrations was high during the first 10 min of the process and that is due to the high free space in the early stages of the process. After 10 min, the removal percentage reduced gradually. Generally, the ideal adsorbent adsorbs contamination at low contact times with high removal performance. The initial concentration is another important factor which affects the amount of adsorption of the MNZ from the solution. The MNZ elimination efficacy decreased by increasing initial concentrations. Accordingly, with an increase in the initial concentration of MNZ from 5 to 40 mg/L, the adsorption percentage decreased from 95 to 69.23% in 50 min; this indicates that the initial concentration strongly affects the removal efficiency. However, the equilibrium absorption capacity increased with an increase in the initial concentration of MNZ. It can also be explained by the fact that more binding sites are available when MNZ concentration is raised. This can lead to an increase in the concentrations of pollutants, which acts as a propulsion to overcome the resistance between the aqueous phase and the adsorbent (Pouretedal & Sadegh 2014). The results were consistent with other studies (Ahmed & Theydan 2012;Samarghandi et al. 2015). Mousavi et al. studied the antibiotic removal of amoxicillin by the NH 4 Cl-MAC. They found that, by increasing the concentration of amoxicillin from 100 to 500 mg/L, the percentage of antibiotic removal of SAC and NAC decreased from 100 to 41% due to the saturation of accessible sites (Moussavi et al. 2013).

Isotherm studies
Adsorption isotherm studies indicate the relationships between the adsorbate and the absorbent; besides, it allows the determination of the theoretical maximum adsorption capacity of an adsorbent to a given adsorbate. To explain the adsorption behavior of MNZ on the MAC, the data from the adsorption equilibrium test stage were tested on the most common models: Langmuir and Freundlich. For the Langmuir isotherm, its linear relationship was used (Equation (3) Langmuir: where C e is the concentration of MNZ adsorption (mg/L), q e is the amount of MNZ adsorbed per unit mass of MAC (mg/g) and K L is the Langmuir constant (L/mg) and expresses the adsorption energy. The fundamental characteristics of the Langmuir isotherm can be expressed by the dimensionless reversible factor, R L (Fazlzadeh et al. 2017): here, if R L is greater than 1, absorbance is unfavorable; if it is equal to 1, it is linear adsorption; if it is between 0 and 1, adsorption is favorable and if it is equal to 0, adsorption is irreversible. The Freundlich isotherm shows heterogeneous and multi-layered adsorbent adsorption. For the Freundlich isotherm, its linear relationship was used (Equation (5)) (Abdollahzadeh et al. 2020): where K F expresses the maximum absorption capacity (mg/g), and 1/n expresses the adsorption intensity. If n is between 1 and 10, it indicates that the adsorption is favorable. Figure 5 represents the Langmuir and Freundlich isotherms derived from the adsorption process data, and Table 3 shows the coefficients of adsorption isotherm models. As can be seen, the amount of R 2 for the MAC adsorbent in the Freundlich isotherm was higher than that in the Langmuir isotherm, indicating that the adsorption process followed the Freundlich model. Also, the value of R L was between 0 and 1, which confirms that MNZ adsorption by the MAC was favorable. On the other hand, in the Freundlich model, the value of n was greater than 1, which indicates that the adsorption of MNZ on the adsorbent was a favorable physical process. The results were consistent with the results obtained by other studies (Liu et al. 2012;Kong et al. 2016). Table 4 shows the adsorption capacity of different types of adsorbents for MNZ. The maximum absorption capacity of MAC was 66.22 mg/g. The results of the maximum adsorption capacity obtained indicated that the MAC had better performance than the other adsorbents applied.

Kinetic studies
The adsorption kinetics indicate the MNZ adsorption rate in the MAC; this adsorption rate controls the equilibrium time. To evaluate the transfer rate of MNZ mass on the adsorbent MAC, adsorption kinetics were used based on the equilibrium test   data from the pseudo-first-order and pseudo-second-order models. The linear form of the pseudo-first-order kinetic model is presented by the following equation (Borna et al. 2016): log (q e À q t ) ¼ log q e À k 1 t 2:303 (6) where q t is the adsorption capacities at time t, and k 1 is the constant coefficient of the pseudo-first-order kinetic model. The pseudo-second-order kinetic model is based on solid-phase adsorption. q e and k 2 are obtained by drawing a plot of t/q e versus t. The linear form of the model is expressed by the following equation (Rashtbari et al. 2020a(Rashtbari et al. , 2020b: where k 2 is the constant coefficient of the pseudo-second-order kinetic model. Regarding the kinetic constants and the correlation coefficients (R 2 ) presented in Table 5 and Figure 6, as well as the comparison of experimental data with the calculated data, it can be stated that the kinetics of the adsorption followed the pseudosecond-order model. Al-Khalisy et al. (2010) showed that the adsorption of cephalexin from aqueous solutions by bentonite and activated carbon followed the pseudo-second-order kinetic model (Al-Khalisy et al. 2010).

Real wastewater samples
Real wastewater containing antibiotics contains many impurities that may interfere with the adsorption process. In this study, a real wastewater sample was obtained from the Daana Pharmaceutical Company. MNZ removal efficiency under the optimal conditions from the real wastewater sample was 64%, a lower value compared with that from the synthetic solutions.
Removal efficiency in the real sample of the industrial wastewater was lower than that determined from the synthetic samples. Interfering factors such as organic and cyclic compounds as well as turbidity inside the wastewater caused the removal efficiency to decline compared with the synthetic samples. Generally, these results are suggestive of the appropriate performance of the composite in MNZ removal from the real samples Zhang et al. 2016). The regeneration and recycling abilities of adsorbents are crucial for their practical application. Spent adsorbents have no benefit and are disposed of as wastes. Disposing of spent adsorbents can be difficult because some of them could be hazardous and need to be incinerated and may be toxic, flammable or even explosive (Gupta et al. 2013). In chemical regeneration, generally, MNZ molecules are dissolved in the eluent or get replaced by an ion-exchange process on the adsorbent surface (Pouretedal & Sadegh 2014;Chieng et al. 2015). The adsorption ability of MAC for the elimination of MNZ from an aqueous solution was examined by using 0.1 M NaOH as a model. As shown in Figure 7, the generation process was carried out five times. Generally, the MAC capacity for the removal of MNZ was reduced with an increasing number of regeneration cycles. Figure 7 shows that, in the first stage of regeneration, the regeneration of the adsorbent decreased to approximately 95%; however, the removal percentage after the fifth regeneration was 73%. Thus, the experimental data indicated that MAC has a high potential to be used for the adsorption of MNZ after five consecutive cycles of regeneration; it will therefore be cost-effective.

CONCLUSIONS
In the present study, MNZ adsorption on the MAC was investigated. The results illustrated that the adsorption of MNZ had high efficiency in acidic conditions and removal performance decreased with increasing initial concentration. The study of adsorption kinetics and isotherm showed that the removal of MNZ was well proportionated with the pseudo-second-order and Freundlich models. Therefore, it can be pointed out that both physisorption and chemisorption occurred in MNZ adsorption. The surface areas of PAC and MAC were 738.18 and 706.92 m 2 /g, and the total pore volumes were 0.551 and 0.532 cm 3 / g. Also, the regeneration of the MAC indicated that this adsorbent has a promising capacity in five regeneration cycles for the removal of MNZ. Thus, it is concluded that MAC has a significant performance in the removal of MNZ from aquatic solutions, and it can be used for further studies of pollutant removal from water and wastewater in the laboratory and industrial scales.