Abstract

Pharmaceutical compounds are essential to preserve human and animal welfare, as well as to prevent illnesses. However, the elevated consumption of drugs, followed by incorrect disposal and inefficient wastewater treatment, may increase their environmental risk. In the case of antibiotics, such as ampicillin, some of the already known consequences are bacterial resistance and some toxic interactions with aquatic organisms. The scope of the present work is to investigate the removal of ampicillin through batch adsorption experiments onto granular activated carbon (GAC). The influence of pH and phase contact time were evaluated. Pseudo-first order, pseudo-second order and intraparticle diffusion models were adjusted to experimental data to determine process kinetics. In order to study adsorption equilibrium and thermodynamics parameters, isotherms at 298 K, 298 K and 308 K were constructed. The models of Langmuir, Freundlich and Sips fitted to experimental data. The best results (73% of removal, residual concentration 5.2 mg L−1) were reached at pH 6 and 120 minutes of contact time. Pseudo-first order model better represented the adsorption kinetics (R2 = 0.99), while the Langmuir equation suited well the experimental isotherms at 288 K and 298 K (R2 = 0.998 and R2 = 0.991) and the Sips equation better represented the system at 308 K (R2 = 0.990). Thermodynamic parameters were estimated as ΔG° = −6,000 J mol−1; −6,700 J mol−1; −7,500 J mol−1 at 288 K, 298 K and 308 K respectively, ΔH° = 14,500 J mol−1 and ΔS° = 71.0 J mol−1 K−1. The results indicate that this process is spontaneous, efficient and potentially applicable in the removal of ampicillin from water.

INTRODUCTION

Contaminants of emerging concern are organic chemicals present in drugs, hormones, pesticides, cleaning and personal care products, amongst others. Several studies point out that this varied group of substances do not have regulatory legislation despite their detected presence in the environment, and the effects upon aquatic organisms and human health are little known yet (Deblonde et al. 2011).

Pharmaceutical compounds are complex molecules developed to preserve human and animal health. However, even though the use of these substances presents obvious benefits, they must be correctly destined after consumption. Still, an expressive part of the ingested molecules is excreted through urine and feces in an unaltered form or as biologically active metabolites (Esplugas et al. 2007).

The advances in analytical techniques have allowed, for the past decades, the detection of these compounds in rivers, water and sewage treatment plants, and in hospital effluents, in concentrations from ng L−1 to μg L−1. Even these low concentrations can present diverse risks to the environment (Souza & Féris 2015). In the case of antibiotics, their occurrence in wastewater and surface waters allows the growth of resistant bacteria, yet many studies confirm that the molecules of this class keep biologically active long after their excretion or disposal, causing unwanted effects in non-target species (Fatta-Kassinos et al. 2011).

Ampicillin (AMP) in its turn, is a β-lactam antibiotic of large spectrum, able to interact with both gram-positive and gram-negative bacteria. Due to this characteristic, AMP is largely applied to treat several kinds of conditions, like gonorrhea, urinary infections and ear, nose and throat infections (Rahardjo et al. 2011). It has been one of the most-consumed antibiotics worldwide, resulting in its detection in surface waters, sewage treatment plants and hospital effluents (Souza & Féris 2017).

Regular water and wastewater treatment may not be able to remove pharmaceutical compounds effectively. Complementary treatment technologies, such as membrane filtration, advanced oxidative processes, bioremediation and adsorption have been intensively studied in the past decade (Teodosiu et al. 2018).

Granular activated carbon (GAC) is a well-known versatile adsorbent. Its application presents such advantages as elevated surface area, pore size, chemical and physical stability and hydrophobicity. Applications vary from the removal of metallic ions to pharmaceuticals (Bhadra et al. 2016).

Some previous studies approached the removal of ampicillin and similar β-lactam antibiotics by adsorption. Weng et al. (2018) studied simultaneous adsorption of amoxicillin, ampicillin and penicillin onto functional bentonite-supported nanoscale Fe/Ni, reaching 80.6% removal in the mixture and 85.1% removal in the individual solution (Co = 20 mg L−1). In the work of De Franco et al. (2017), adsorption of amoxicillin onto activated carbon was studied in batch and fixed bed column systems, where an adsorption capacity of 4.4 mg g−1 was observed. Other researches evaluated the removal of ampicillin in different adsorbents, like ordered mesoporous silica (Nairi et al. 2017) and modified Ponorogo bentonite (Rahardjo et al. 2011).

The goal of the present work is to study the adsorption of ampicillin onto activated carbon, investigating the influence of pH and phase contact time in the process, to evaluate adsorption kinetics, with pseudo-first order, pseudo-second order and intraparticle diffusion models, and adsorption equilibrium, with the construction of equilibrium isotherms at 288 K, 298 K and 308 K. and adjustment of Langmuir, Freundlich and SIPS models to experimental data and estimation of thermodynamic parameters.

METHODS

Reagents

The antibiotic Ampicillin (Table 1) was purchased in analytical standard (>99%, Sigma-Aldrich). A stock solution in the concentration of 1,000 mg L−1 was prepared with distilled water; solutions from experiments were prepared by dilution and pH was adjusted with NaOH (Synth, Diadema, Brasil) and HCl (Synth, Diadema, Brasil) 0.1 M solutions.

Table 1

Ampicillin molecular structure and physical-chemical properties

Parameter Character/value 
Molecular structure  
CAS number 69 − 53 − 4 
Molecular formula C16H19N3O4
Molecular weight 349.40 
Acid dissociation constant (pKa) 2.77a (carboxyl), 7.14a (amine) 
Molecular diameter 16.4 Å 
Parameter Character/value 
Molecular structure  
CAS number 69 − 53 − 4 
Molecular formula C16H19N3O4
Molecular weight 349.40 
Acid dissociation constant (pKa) 2.77a (carboxyl), 7.14a (amine) 
Molecular diameter 16.4 Å 

Adsorbent solid

Vegetal-based granular activated carbon (GAC) (Synth, Diadema, Brazil) was sieved and the fraction between 1.4 and 2 mm was used. BET (Brunauer-Emmett-Teller) surface area was determined as 543.4 m2 g−1, and medium pore size was estimated at 32.2 Å. Zeta potential analysis was performed (Zeta Plus Brookhaven Instruments Corporation, ZEE, model 500) in a sample, and the results indicated that the GAC used in this work is positively charged only at pH values less than 2 and negative in higher pH values, increasing its negative characteristic with the increase of pH. The solid was washed in current distilled water in order to remove particles in smaller sizes and dust.

Adsorption experiments

Adsorption experiments were carried in 250 mL Schott flasks, filled with 100 mL of 20 mg L−1 AMP solution (de Franco et al. 2017; Haro et al. 2017) and pH was adjusted with acid and basic solutions. Activated carbon was previously weighted and added to the solution, and then the flasks were introduced into a Wagner shaker (Marconi, MA160BP) under a rotation of 28 ± 2 rpm. In the case of isotherms, experiments were carried out in a controlled atmosphere orbital shaker (Cientec, CT-712RN) to maintain constant temperature. Samples were taken after the experiments, and AMP concentrations were measured in a UV-Vis spectrometer (Thermo Scientific, Genesis 10S UV-VS) at the wavelength of 204 nm (Elmolla & Chaudhuri 2010). The difference amongst initial and residual concentrations of AMP were used to calculate percentage removal (%) in each experiment. All experiments were performed in triplicate.

Influence of pH

To evaluate the influence of pH in the process, flasks were filled with AMP solution, then pH was adjusted in the studied range (2, 4, 6, 8 and 10) with HCl and NaOH 0.1 M solutions. 5 g L−1 of GAC were added to the flasks and experiments were carried out for 30 minutes. Percentage removal was calculated by the Equation (1): 
formula
(1)
where Co and Ct are initial and final concentrations (mg L−1) of ampicillin in aqueous solution.

Contact time and kinetics

Experiments were taken in the pH that achieved the best removal efficiency in previous tests (item 2.2.1.). In these experiments, contact time varied between 0 and 210 minutes, until equilibrium was observed. AMP and GAC concentrations testes remained as 20 mg L−1 and 5 g L−1, respectively.

The quantity of adsorbed ampicillin onto granular activated carbon at specific contact times is described by mass balance (Equation (2)): 
formula
(2)
where qt is the quantity of ampicillin retained per gram of GAC at a certain time t (mg g−1), Co is the initial concentration of ampicillin and Ct is the concentration of ampicillin at time t (mg L−1), V is the volume (L) of ampicillin solution, and W is the adsorbent mass, in grams (g).
A kinetics study provides information like reaction order and rate constants, important factors in order to apply the adsorption process. The most commonly tested models of adsorption kinetics are pseudo-first order (Equation (3)), pseudo-second order (Equation (4)) and intraparticle diffusion (Equation (5)). 
formula
(3)
 
formula
(4)
 
formula
(5)
where qt and qe are the quantities of ampicillin adsorbed at a certain time t and at equilibrium (mg g−1), respectively, k1 (min−1), k2 (mg g−1 min−1) and kid (mg g−1 min1/2) are the rate constants of pseudo-first order, pseudo-second order and intraparticle diffusion model, respectively.

Adsorption equilibrium isotherms

Isotherms comprise an essential part of adsorption studies. From them, it is possible to evaluate the physical interactions between adsorbate and adsorbent. The experimental curves were obtained in three different temperatures: 288 K, 298 K and 308 K, varying initial AMP concentration between 0 and 750 mg L−1. GAC concentration was kept at 10 g L−1, and other experimental conditions were pH 6 and 150 minutes of contact time. Samples were filtered and analyzed.

In the present study, three commonly used isotherm equations were applied: Langmuir (Equation (6)), Freundlich (Equation (7)) and the hybrid model of Sips (Equation (8)). Experimental values of qe were calculated as qt by Equation (2). 
formula
(6)
 
formula
(7)
 
formula
(8)
where: qe is the ampicillin mass adsorbed onto each gram of GAC at equilibrium (mg g−1); qmax (mg g−1) is the limiting (maximum) adsorption capacity; Ce is the equilibrium concentration of ampicillin (mg L−1); KL (L mg−1) is Langmuir's constant; KF ((mg g−1)(L mg−1)1/n) and n are empiric Freundlich constants, related to the characteristics of the system; KS is the adsorption constant of Sips and γ is the heterogeneity constant. Whenever γ approaches the unit value, the Sips equation reduces to the Langmuir model, and a homogeneous surface may be presumed for the adsorbent.

Thermodynamics parameters

The estimation of the classic thermodynamic parameters ΔH° (molar enthalpy), ΔS° (molar entropy) and ΔG° (molar Gibbs energy) is possible by constructing equilibrium isotherms at different temperatures. These parameters can indicate bonding energies, the feasibility and the spontaneity of the adsorption process, and they were calculated by the well-known thermodynamic equations: 
formula
(9)
 
formula
(10)
where Ke is the equilibrium constant (qe Ce−1); R is the universal gas constant (8,314 J mol−1 K−1) and T is the temperature in Kelvin (K).

RESULTS AND DISCUSSION

Influence of pH

The adsorption removal (%) of ampicillin onto granular activated carbon at different pH values is displayed in Figure 1.

Figure 1

Effect of pH in ampicillin removal (%) (conditions: Co: 20 mg L−1, GAC: 10 g L−1, 30 minutes).

Figure 1

Effect of pH in ampicillin removal (%) (conditions: Co: 20 mg L−1, GAC: 10 g L−1, 30 minutes).

It is possible to observe that the best removal efficiencies were reached at pH 4 and pH 6. In these cases, percentage removal was 33.7 and 34%, respectively. A statistical analysis of variance (ANOVA) comparing these results indicated that there was no significant difference in the results (95% trust interval). However, due to the fact that the pH of the solution was higher than 4 and wastewaters are generally discharged in the neutral pH range, pH 6 was chosen to carry out further experiments.

At pH values less than 4 and above 6, the antibiotic removal decreases. These results can be explained by the influence of pH on AMP ionized species, the charging of the adsorbent surface and the interactions formed between them. Considering ampicillin is an amphoteric molecule that presents two dissociation constants (pKa1 = 2.77; pKa2 = 7.14), under pH 2.77 the AMP specie is positively charged, and above pH 7.14, the specie is charged negatively. In the pH range between its pKa values (2.77–7.14), AMP predominates in the zwitterion form, in which the carboxylic group is deprotonated (–COO) and the amine is protonated (–NH3+), while the molecule, as a whole, is considered to be neutral (Völgyi et al. 2007).

Accordingly to the zeta potential analysis, the GAG used in the experiments is slightly positive in pH values equal or less than 2, agreeing with the predominant form of ampicillin, which is cationic. Consequently, the electrostatic repulsion causes a lower removal. An analog explanation is applicable when higher pH values are concerned. Above AMP pKa2, where AMP is in its anionic form and the GAC surface is negatively charged, the results are weaker interactions and a reduced removal efficiency. On the other hand, in pH range 4–6, GAC is found to be negative and AMP is in zwitterionic equilibrium, and adsorption may occur either from electrostatic attraction, amongst the negative solid surface and the protonated amine groups from AMP, or the formation of chemical bonds.

Nairi et al. (2017) studied the removal of AMP onto an amino-functionalized ordered mesoporous silica at pH 7.4. While the antibiotic has a negative charge, the adsorbent remains positive in this condition and attraction interactions induce adsorption. In contrast to GAC, Peterson et al. (2010) found that a nanometer-size alumina has a positively charged surface through the whole pH range. The authors concluded that AMP removal is favored at pH 8, when AMP is anionic. In all cases, pH represents a major factor in the adsorption process.

Studies involving amoxicillin (AMX) adsorption, another β-lactam antibiotic chemically very similar to ampicillin, and therefore owning close pKa values (pKa1 = 2.68; pKa2 = 7.49) (Putra et al. 2009), indicate similar results. Pezoti et al. (2016) reached a better removal at pH 4 for AMX through adsorption onto NaOH-activated carbon produced from guava seeds, while Putra et al. (2009) also encountered a higher adsorption capacity for GAC at pH 4.98, if compared to pH 2.23 and 7.05, confirming the observed behavior in the studied GAC.

Phase contact time and kinetics

Figure 2 exhibits the removal of ampicillin throughout phase contact time.

Figure 2

Influence of time in ampicillin removal (%) (conditions: Co: 20 mg L−1, GAC: 10 g L−1, pH 6).

Figure 2

Influence of time in ampicillin removal (%) (conditions: Co: 20 mg L−1, GAC: 10 g L−1, pH 6).

According to Figure 2, the AMP adsorption rate increases until 90 minutes. Also, the rate slowly decreases until equilibrium is reached. By analyzing the driving force of the process, which is the concentration gradient of AMP between the phases, it can be observed that at initial periods of time all active sites in the GAC surface are available, resulting in a fast mass transfer. As the available sites decrease, the system approaches an equilibrium state. The percentage removal reaches close values from 120 minutes on ward, increasing from 71% to 73% at 150 minutes. A statistical analysis (ANOVA, trust interval 95%), confirmed that there was no significant difference in the removal between 120 and 150 minutes. Thus, the best phase contact time established for the process was set at 120 min. At this point, the qe value was 2.8 mg g−1. For kinetics, the pseudo-first order (PFO) and pseudo-second order (PSO) were adjusted to experimental data, as shown in Figure 3. Pseudo-first and pseudo-second order models assume that the adsorption kinetics is controlled by external diffusion, but they are not able to identify if intraparticle diffusion is significant. This way, the model of the intraparticle diffusion model was also compared to experimental data, with the aim of checking mass transfer resistance in the process. This model assumes the adsorbed mass is directly proportional to time0.5. The model adjustment and the possible steps of intraparticle diffusion are displayed in Figure 4. All parameters obtained for the three models are shown in Table 2.

Table 2

Estimated parameters for the kinetics models of pseudo-first order, pseudo-second order and intraparticle diffusion

Experimental Pseudo-first order Pseudo-second order Intraparticle diffusion 
qe = 3,1 mg.g−1 qe (mg g−13.1 qe (mg g−14.0 Kid (mg g–1 min–0,50.24 
k1 (min−10.020 k2 (g mg−1 min−10.005 C (mg g−1−0.01 
R2 0.990 R2 0.985 R2 0.93 
Experimental Pseudo-first order Pseudo-second order Intraparticle diffusion 
qe = 3,1 mg.g−1 qe (mg g−13.1 qe (mg g−14.0 Kid (mg g–1 min–0,50.24 
k1 (min−10.020 k2 (g mg−1 min−10.005 C (mg g−1−0.01 
R2 0.990 R2 0.985 R2 0.93 
Figure 3

Non-linear pseudo-first order and pseudo-second order kinetics models adjustment.

Figure 3

Non-linear pseudo-first order and pseudo-second order kinetics models adjustment.

Figure 4

Intraparticle diffusion model fitted to experimental data.

Figure 4

Intraparticle diffusion model fitted to experimental data.

In Figure 3, visually the PFO model fitted better to experimental data than the PSO model, especially in the plateau observed in the last three points. Analyzing Table 2, the determination coefficients (R2) obtained for pseudo-first and pseudo-second order are, respectively, 0.990 and 0.985, close values. Based on R2 values, it is not possible to ensure which model describes more accurately the system. In that case, looking closely at the estimated values of qe, for PFO and PSO, this parameter was 3.1 mg g−1 (minimum error) and 4.0 mg g−1 (error of 22.5%) respectively. As the experimental value obtained for qe was also 3.1 mg g−1, it was assumed that the pseudo-first order models the system kinetics with more accuracy. This result indicates that the adsorption velocity as a function of time is directly proportional to the difference of adsorbed amounts of AMP at a certain time and at equilibrium. Ahmed et al. (2017) and Liao et al. (2013) also encountered a better adjustment for the PFO model in the adsorption of tetracycline antibiotics onto activated carbons. However, some other studies with antibiotics adsorbed onto activated carbons found the PSO model adjusted best to the experimental data (Pezoti et al. 2016; de Franco et al. 2017), but generally with close determination coefficient values for both models.

The analysis of the intraparticle diffusion model (ID) gives an overview of the adsorption process. By the graphical representation in Figure 4, it is possible to visualize three different linear regions that indicate different steps of mass transfer (Lladó et al. 2015). The first region (I) can be assigned as the external diffusion through the liquid boundary layer. The second region (II) is related to intraparticle diffusion itself, and the third region (III) can be interpreted as the adsorption equilibrium, in which the adsorbed amount is constant (Ruiz et al. 2010).

Intraparticle diffusion actually limits the process if the resulting graphic presents only one linear phase that crosses the origin. As can be seen in Figure 4, this behavior is not observed, which is confirmed by the determination coefficient (R2 = 0.93, also displayed in Table 2), being lower than in the previous nonlinear models. According to Ahmed & Theydan (2012), the deviation from origin may relate to the differences in the mass transfer at the initial and final stages. These results lead to the conclusion that the ID model does not well represent the system, and this stage is not limiting to the process. Previous studies also found that intraparticle diffusion is not the limiting stage in the adsorption of drugs (Lladó et al. 2015).

Equilibrium – adsorption isotherms

Figure 5 presents the experimental adsorption isotherm of AMP onto GAC at 288 K, 298 K and 308 K. The estimated parameters of Langmuir, Freundlich and SIPS isotherms, as well as their determination coefficients, are shown in Table 3.

Table 3

Isotherms parameters at 288 K, 298 K and 308 K

Model Parameter 288 K 298 K 308 K 
Langmuir qmax (mg g−19.7 12.7 12.4 
KL (L mg−10.012 0.015 0.018 
R2 0.998 0.991 0.981 
χ2 0.004 0.013 0.025 
Freundlich KF (mg.g−1) (L.mg−1)1/n 0.7 0.9 1.2 
2.5 2.4 2.6 
R2 0.968 0.920 0.966 
χ2 0.059 0.234 0.082 
Sips qmax (mg g−110.1 11.4 15.3 
KS (L mg−10.015 0.006 0.034 
Γ 0.9 1.1 0.7 
R2 0.998 0.996 0.990 
χ2 0.004 0.013 0.025 
Model Parameter 288 K 298 K 308 K 
Langmuir qmax (mg g−19.7 12.7 12.4 
KL (L mg−10.012 0.015 0.018 
R2 0.998 0.991 0.981 
χ2 0.004 0.013 0.025 
Freundlich KF (mg.g−1) (L.mg−1)1/n 0.7 0.9 1.2 
2.5 2.4 2.6 
R2 0.968 0.920 0.966 
χ2 0.059 0.234 0.082 
Sips qmax (mg g−110.1 11.4 15.3 
KS (L mg−10.015 0.006 0.034 
Γ 0.9 1.1 0.7 
R2 0.998 0.996 0.990 
χ2 0.004 0.013 0.025 
Figure 5

Adsorption isotherms of ampicillin onto granular activated carbon at 288 K, 298 K and 308 K. (conditions: Co: 0–750 mg L−1, GAC: 10 g L−1, pH 6, 240 minutes).

Figure 5

Adsorption isotherms of ampicillin onto granular activated carbon at 288 K, 298 K and 308 K. (conditions: Co: 0–750 mg L−1, GAC: 10 g L−1, pH 6, 240 minutes).

According to Figure 5 and Table 3, it is possible to verify that adsorption at all studied temperatures is favorable, reaching a higher adsorbed amount of ampicillin while increasing temperature. This is characteristic of an endothermic process, which is confirmed with further thermodynamics analysis. Higher temperatures may also facilitate the diffusion of adsorbate molecules through the liquid limit boundary and inside the pores of GAC, and this is confirmed by the enhancement of the adsorption capacity of the system (Ahmad & Kumar 2010).

The favorability of the process is indicated by the obtained values in Table 3. Giles et al. (1960) predicted that when the Freundlich empiric parameter n is higher than 1, isotherms present a concave shape in relation to the abscissa, and adsorption is feasible. The obtained values for the n parameter were 2.5, 2.4 and 2.6 at 288 K, 298 K and 308 K, respectively.

Moreover, considering the obtained values for R2 and χ2 (chi-square error), the Freundlich equation did not adjust that well at all analyzed temperatures. Looking closely at the adjustments of Langmuir and Sips, both present very close determination coefficient values at all temperatures, so it was necessary to take in account the estimated parameters of these models.

The γ parameter from the Sips model characterizes solid surface heterogeneity. When γ is close to unity, the equation reduces to the Langmuir model, which suggests a more homogeneous adsorbent surface. This behavior can be confirmed in Table 3, where the γ value is close to 1 at 288 K and 298 K (0.9 and 1.1, respectively). As a conclusion, the Langmuir model can properly represent the system at 288 K and 298 K, and the adsorption behavior is not altered in this temperature transition. The statistic parameters of R2 and χ2 were 0.998 and 0.004 at 288 K, and 0.991 and 0.028 at 298 K, respectively. The satisfactory adjustment of the Langmuir equation leads to the conclusion that, at these temperatures, adsorption predominates as monolayer, typical of chemisorption mechanisms.

When it comes to 308 K, the γ parameter is equal to 0.7 and the previous simplification is not applicable. At this higher temperature, the Sips model adjusted better to the experimental data (R2 = 0.990), showing the adsorption behavior can change with temperature increase. At low AMP concentrations, adsorption mechanisms approach the Freundlich model with weaker electrostatic interactions, and at higher concentrations, the system behaves with greater similarity to Langmuir (chemical interactions).

Returning to the Langmuir model, it is important to highlight the KL parameter values, which indicate the affinity between adsorbent and adsorbate. According to Langmuir (1918), KL depends on several things: adsorbate molecular diameter, phase contact time, temperature and the amount of active sites. Previous studies in the literature indicate the values for KL in the adsorption of antibiotics are lower than 0.05 (Putra et al. 2009; Chayid & Ahmed 2015), which coincides with the values obtained in this work (0.012, 0.015 and 0.018 at 288 K, 298 K and 308 K, respectively).

Fazelirad et al. (2015) found a better adjustment (R2 = 0.994) to the Langmuir model at 300 K, for the adsorption of AMP onto magnetic multi-walled carbon nanotubes, and Nairi et al. (2017), studying the adsorption of AMP onto ordered mesoporous silica at 298 K, also concluded that Langmuir better represents the adsorption process (R2 = 0.991 as best result). Other studies involving β-lactam antibiotics reached a higher compatibility with the Sips model, as can be seen in the papers of Chayid & Ahmed (2015) and de Franco et al. (2017).

Thermodynamics

With the aim of investigating further the adsorption of ampicillin onto GAC, a thermodynamics study was set. Molar Gibbs free energy (ΔG°), enthalpy (ΔH°) and entropy (ΔS°) were calculated using Equations (9) and (10). Considering that the Langmuir isotherm presented a good adjustment (R2 > 0.965) at all temperatures, KL was used in the equations as the adsorption equilibrium constant. The values obtained for the parameters are displayed in Table 4.

Table 4

Thermodynamic parameters of ampicillin adsorption onto GAC, at 288 K, 298 K and 308 K

T(K) ΔG° (J mol−1ΔH° (J mol−1ΔS° (J mol−1 K−1R2 
288 −6,000    
298 −6,700 14,500 71.0 0,99 
308 −7,500    
T(K) ΔG° (J mol−1ΔH° (J mol−1ΔS° (J mol−1 K−1R2 
288 −6,000    
298 −6,700 14,500 71.0 0,99 
308 −7,500    

Table 4, shows negative values for ΔG° at all studied temperatures, leading to the conclusion that the adsorption process is favorable and spontaneous. Also, ampicillin is likely to have chemical affinity with the solid surface. Another interesting point is that the Gibbs energy decreases with the increment of temperature, so the process is more facilitated at higher temperatures. Solid pores may dilate at higher temperatures, and then intraparticle diffusion enhances. This behavior was also observed in other adsorption studies (Dai et al. 2018; de Franco et al. 2018). Zimnitsky et al. (2006), studying the adsorption of AMP onto oxidized cellulose encountered values for Gibbs energy (298 K) from −8,600 to 10,900 J mol−1, close (same magnitude) to that obtained in this work, −6,700 J mol−1.

The endothermic nature of the process was proven from the positive calculated value for ΔH°, coinciding with higher adsorbed amounts of AMP (qe) at higher temperatures. Also, endothermic adsorption is typical of chemical bonds, in agreement with the best fitting of the Langmuir and Sips models to the experimental isotherms (Rathod et al. 2015; Pezoti et al. 2016). The positive value for ΔS° indicates the increase of heat dispersion and disorder in the system, overall, and also a meaningful affinity between ampicillin and activated carbon (Chayid & Ahmed 2015).

CONCLUSIONS

According to the results in this research, pH 6 was proved to be the best suited for the process, by enhancing chemical and electrostatic interactions. The suggested phase contact time, in the applied conditions, was 120 minutes. Best removal was 73% (tantamount residual concentration of AMP: 5.4 mg L−1) at this point. During the kinetics study, the pseudo-first order model fitted more appropriately to the experimental data (R2 = 0.99), predicting a qe value of 3.1 mg g−1, coincident with the experimental value. From adsorption isotherms, the Langmuir model suited the system well at 288 K and 298 K, suggesting chemical adsorption and a maximum adsorption capacity, while the Sips model was more accurate to model the system at 308 K, leading to the conclusion that AMP adsorption is affected by temperature at a certain point. Thermodynamics indicated the process is endothermic and entropic, as expected.

The overall results point to the viability of the process and the potential of applying GAC in the treatment of water containing ampicillin, such as hospital wastewater and pharmaceutical industrial wastewater.

ACKNOWLEDGEMENT

The authors would like to thank the National Counsel of Technological and Scientific Development (CNPq) and the Coordination for the Improvement of Higher Education Personnel (CAPES) of the Brazilian Government for providing financial support to develop this research.

REFERENCES

REFERENCES
Bhadra
B. N.
,
Seo
P. W.
&
Jhung
S. H.
2016
Adsorption of diclofenac sodium from water using oxidized activated carbon
.
Chem. Eng. J.
301
,
27
34
.
Deblonde
T.
,
Cossu-Leguille
C.
&
Hartemann
P.
2011
Emerging pollutants in wastewater: a review of the literature
.
Int. J. Hyg. Environ. Health
214
,
442
448
.
de Franco
M. A. E.
,
de Carvalho
C. B.
,
Bonetto
M. M.
,
de Pelegrini Soares
R.
&
Féris
L. A.
2018
Diclofenac removal from water by adsorption using activated carbon in batch mode and fixed-bed column: isotherms, thermodynamic study and breakthrough curves modeling
.
J. Clean. Prod.
181
,
145
154
.
Haro
N. K.
,
Del Vecchio
P.
,
Marcilio
N. R.
&
Féris
L. A.
2017
Removal of atenolol by adsorption – study of kinetics and equilibrium
.
J. Clean. Prod.
154
,
214
219
.
Lladó
J.
,
Lao-Luque
C.
,
Ruiz
B.
,
Fuente
E.
,
Solé-Sardans
M.
&
Dorado
A. D.
2015
Role of activated carbon properties in atrazine and paracetamol adsorption equilibrium and kinetics
.
Process Saf. Environ. Prot.
95
,
51
59
.
Nairi
V.
,
Medda
L.
,
Monduzzi
M.
&
Salis
A.
2017
Adsorption and release of ampicillin antibiotic from ordered mesoporous silica
.
J. Colloid Interface Sci.
497
,
217
225
.
Peterson
J. W.
,
Burkhart
R. S.
,
Shaw
D. C.
,
Schuiling
A. B.
,
Haserodt
M. J.
&
Seymour
M. D.
2010
Experimental determination of ampicillin adsorption to nanometer-size Al2O3 in water
.
Chemosphere
80
,
1268
1273
.
Pezoti
O.
,
Cazetta
A. L.
,
Bedin
K. C.
,
Souza
L. S.
,
Martins
A. C.
,
Silva
T. L.
,
Santos Júnior
O. O.
,
Visentainer
J. V.
&
Almeida
V. C.
2016
NaOH-activated carbon of high surface area produced from guava seeds as a high-efficiency adsorbent for amoxicillin removal: kinetic, isotherm and thermodynamic studies
.
Chem. Eng. J.
288
,
778
788
.
Rahardjo
A. K.
,
Susanto
M. J. J.
,
Kurniawan
A.
,
Indraswati
N.
&
Ismadji
S.
2011
Modified Ponorogo bentonite for the removal of ampicillin from wastewater
.
J. Hazard. Mater.
190
,
1001
1008
.
Ruiz
B.
,
Cabrita
I.
,
Mestre
A. S.
,
Parra
J. B.
,
Pires
J.
,
Carvalho
A. P.
&
Ania
C. O.
2010
Surface heterogeneity effects of activated carbons on the kinetics of paracetamol removal from aqueous solution
.
Appl. Surf. Sci.
256
,
5171
5175
.
Zimnitsky
D. S.
,
Yurkshtovich
T. L.
&
Bychkovsky
P. M.
2006
Adsorption of zwitterionic drugs onto oxidized cellulose
.
J. Colloid Interface Sci.
295
,
33
40
.