1 Introduction

Addressing the contamination of waters by substances that cannot be removed by conventional wastewater treatment processes is a challenging task gaining increased interest. In this regard, adsorption onto activated carbon is a technique for the removal of persistent xenobiotics from aquatic environments that offers high efficiency and easy disposal. The cost of the treatment can be significantly reduced when a waste biomass is used as a precursor in the activated carbon synthesis. The efficiency of the adsorption process on activated carbon depends on factors such as pore size, pore area, and surface chemistry [1]. Activated carbons also allow for the modification of their surfaces with heteroatoms, such as nitrogen or oxygen, which can further facilitate interactions with organic molecules [2]. Currently, various agricultural wastes or invasive herbal biomasses have been tested as alternative precursors in activated carbon preparation [3, 4] to replace conventional precursor materials (i.e. wood, coal, coconut shells, etc.) in order to reveal more accessible and low cost activated carbons with a possible higher adsorption capacity/efficiency.

Of the targets for the treatment by activated carbon, drug residues are major substances of concern, as they can exhibit strong physiological effects at low concentrations and many of them are only partially removed by wastewater treatment plants [5]. The most important drug-based pollutants are antibiotics, used in human and veterinary medicine as a tool for fighting bacterial-based infections. In addition to medical use, antibiotics are widely used as growth promoters in animal farming. Thus, effluents from hospitals, household, and farms, have led to the serious presence of antibiotics in surface waters and wastewaters [6] which must be reduced. Antibiotics in waters have negative effects on some water organisms and moreover, the repeated exposure to contaminated water can even lead to health problems in humans [7]. Some studies suggest that water contamination contributes to the development of antibiotic resistance in bacteria [8,9,10].

The present study is focused on fluoroquinolone antibiotics (FQ), synthetic antimicrobials derived from nalidixic acid. FQ derivatives can be tailored to have a neutral, cationic, anionic, or zwitterionic character on physiological conditions, which is the reason for their efficiency in curing a broad range of infections. Most of the FQs are excreted, not metabolized by the treated organisms, and therefore they can subsequently accumulate in the soil and waters. The most common paths of environment contamination with FQs are, among others, farmland, antibacterial treatment in fish farms or the presence of FQs in manure, which is being used as fertilizer [11,12,13]. Even though FQs exhibit a lowered risk of bacterial resistance development [11, 14], the widespread pollution by these compounds has already led to the rise of antibiotic resistance of bacteria to these antimicrobial agents [11, 13, 15, 16]. Norfloxacin and ofloxacin are the most widely used FQs.

Norfloxacin, commonly used in urinary tract infection treatment, occurs in waters at ppb levels [17, 18]. It is a relatively poorly soluble compound (280 mg·L−1) which exists in aqueous solutions mainly in its zwitterionic form owing to the acid/base interactions between the basic nitrogen of the piperazine and the carboxylic acid group [19]. Dissociation constant values for norfloxacin are pKa1 = 6.34 (carboxylic acid) and pKa2 = 8.75 (piperazine ring). Ofloxacin is a carboxylic acid with a broad spectrum of uses, including, e.g. urinary and respiratory tract infections, otitis, or conjunctivitis [20]. Compared to norfloxacin, it has approximately 100 × times higher solubility in water (28.3 g·L−1). The dissociation constant values are 5.97 (pKa1) and 9.28 (pKa2). Both norfloxacin and ofloxacin are particularly resistant to conventional wastewater treatment and can disrupt the reproduction and evolution of aquatic organisms [21].

Several methods for the removal of fluoroquinolones from waters have been studied, e.g. gamma-ray irradiation [18], nanofiltration [22], advanced oxidation processes [23, 24], or photocatalysis [25, 26]. Several studies have also been devoted to removal of norfloxacin and ofloxacin by adsorption. Adsorption onto chitosan-based nanosheets was studied by [27]. The adsorption isotherm followed the Langmuir model with maximum adsorption capacity 1 000 mg·g−1. Liu et al. [28] studied the adsorption of norfloxacin onto lotus stalk-based activated carbon, which synthesis included activation with H3PO4 and carbonization at 450 °C for 1 h [29]. The maximum sorption capacity corresponded to 295 mg∙g−1 and was obtained at pH 5.5. The adsorption isotherm was consistent with the Langmuir model while the adsorption kinetics followed a pseudo-second-order model. Xie et al. [30] examined an activated carbon developed from Trapa natans husk for norfloxacin sorption. The husk was activated with H3PO4 and subsequently carbonized at 470 °C for 90 min. Again, the adsorption kinetics followed the pseudo-second-order model, but the adsorption isotherm corresponded to the Freundlich and Temkin model. Liu et al. [31], on the other hand, worked with granular activated carbon prepared via superheated steam physical activation at 540 °C. The isotherm agreed with the Langmuir model, the maximum sorption capacity was 113 mg∙g−1. Adsorption kinetics matched the pseudo-second-order model. Wang et al. [32] used bamboo biochar, prepared using a pyrolysis at 500 °C, for the adsorption of ofloxacin and enrofloxacin. The adsorption isotherms corresponded to the Freundlich model. Removal of ofloxacin with activated carbon prepared from luffa sponge was reported by Kong et al. [33]. The luffa sponge was activated with 85% H3PO4 and carbonized at 450 °C for 2 h. The adsorption kinetic data fitted the pseudo-second-order model and the adsorption isotherms corresponded to the Freundlich model with a maximum adsorbed amount of 132 mg∙g−1. Li et al. [34] used potato stems and leaf-based biochar (pyrolysed at 500 °C for 5 h) for an easy separation of fluoroquinolones; its adsorption isotherm for norfloxacin corresponded better to the Langmuir model and its adsorption capacity was 8.37 mg∙g−1. Wang et al. [35] prepared luffa-based activated carbon by carbonization at 800 °C for 6 h. The product of the carbonization was subsequently mixed with an agarose solution for 6 h, stored in a refrigerator, and freeze-dried under a vacuum condition. A prepared lightweight and compressible biosorbent showed very high levels of maximum uptake for both norfloxacin (434.8 mg·g−1) and ofloxacin (581.4 mg·g−1). Several works were also focused on the adsorptive removal of other fluoroquinolone antibiotics, for instance ciprofloxacin [27, 34, 36, 37], enrofloxacin [32, 34, 38], or levofloxacin [39]. The adsorption studies overview reveals that the adsorption kinetics of FQ follows a pseudo-second-order model, but the type of adsorption isotherm describing the steady-state FQ adsorption determining the overall adsorption capacity of produced activated carbon (in mg·g−1) differs, depending on the waste material and the method of treatment (i.e. activation, pyrolysis conditions, post-treatment) used for biochar/activated carbon production. The optimum pH value lies mostly in the range 5–8, where dissolved fluoroquinolones are of a cationic or zwitterionic character. Hence, surface character and charge of the adsorbent are affected by the solution pH. This indicates that there still exists a knowledge gap about the most suitable and cheapest waste biomass-based activated carbon for FQ adsorption and the activated carbon features and surface chemistry affecting the adsorption mechanism of individual FQ.

This work focuses on the use of activated carbons produced from different agricultural waste biomasses for the adsorption of fluoroquinolone antibiotics—as model water pollutants. The activated carbons were prepared from several agricultural waste materials of Peruvian origin (i.e. red mombin seed, corn cob, coffee husk, internal and external parts of mango seeds, and ice cream beans), being widely abundant in Latin America [40, 41]. The special properties of selected agrowaste-based activated carbons are their large mesopore surface area, large micropore volume and net pore volume, narrow pore-size distribution within the microporous–mesoporous region, and slightly acidic character (pHPZC = 5.9–6.9) due to the presence of oxygen-containing surface groups [42]. These properties of the prepared materials seemed appropriate for the efficient sorption of antibiotics from water that had not been tested before. Therefore, the adsorption performance of six agricultural waste-based activated carbons for two fluoroquinolone antibiotics (norfloxacin and ofloxacin) in relation to their textural and surface properties were investigated, and the sorption mechanism was described using proper kinetic and adsorption isotherm models. Moreover, the molecular dimensions of norfloxacin and ofloxacin were estimated based on molecule structure optimization via DFT calculations implemented in a Gaussian software package to consider the possible steric shielding of the micropores of the investigated activated carbons.

2 Materials and methods

2.1 Activated carbon preparation and characterization

Activated carbons prepared from various agricultural wastes (red mombin seed (RMS), corn cob (CC), coffee husk (CH), internal (MSIP) and external (MSEP) parts of mango seed, and ice cream bean (GS)), were used as adsorbents. Raw materials were collected in the agricultural areas in Peru, as published in [42]. First of all, raw materials were washed with potable water and at dried at 80 °C until the reaching of constant weight. The dried raw materials were ground and sieved to 0.5–1 mm particles.

All raw materials were firstly mixed with ZnCl2 in a 1:1 mass proportion and subsequently placed in a ceramic reactor into a horizontal oven and thermally treated/pyrolysed at 600 °C (temperature ramp 10 °C/min) in a nitrogen atmosphere (flow rate 150 mL/min) for 2 h. The samples were carbonized without water addition [43]. The mixing step was done strongly in a mechanical stirrer. After the treatment, the material was cooled in an inert atmosphere. The carbonized materials were washed with a 0.15 M HCl solution and the with boiled and room-temperature distilled water.

The detailed physicochemical characterization of the activated carbons is given in the previous work, see [42]. In the present study, all activated carbon-based adsorbents were sieved to particle-size fraction < 0.09 mm and their textural properties were newly evaluated with the aid of nitrogen physisorption at 77 K.

Scanning electron microscopy (SEM) images were purchased with the aid of scanning electron microscope Tescan Vega (TESCAN, Czech Republic) with Tungsten cathode. Micrographs were obtained in secondary electrons and backscattered electrons mode with an acceleration voltage of 30 keV. Samples were gold-sputtered before analysis to ensure adequate electron conductivity.

2.2 Adsorption

Norfloxacin and ofloxacin antibiotics (Table 1, Fig. 1) used as model adsorbates were obtained from Sigma Aldrich (St. Louis, MO, USA). Their stock solutions (150 mg·L−1 for norfloxacin and 250 mg·L−1 for ofloxacin) were prepared in ultrapure water. The concentration of the norfloxacin solution was reduced due to the significantly lower solubility of pure antibiotic in water. In adsorption experiments, the concentration of norfloxacin and ofloxacin was monitored with the aid of an ultraviolet spectrophotometer (Specord 250 Plus, Analytik Jena AG, Germany) in a quartz cuvette (light path 10 mm) at wavelengths of 275 nm (norfloxacin) and 290 nm (ofloxacin). The amount adsorbed onto 1 g of activated carbon (qe) was calculated according to Eq. (1):

$${q}_{e}=\frac{\left({c}_{0}-{c}_{e}\right).V}{{m}_{AC}}$$
(1)

where c0 represents the initial concentration of solute, ce the concentration in equilibrium/after adsorption, V the volume of solution, and mAC the weight of the adsorbent.

Table 1 General physicochemical properties of adsorbed fluoroquinolones
Full size table
Fig. 1
figure 1

Modelled structure of a) norfloxacin and b) ofloxacin (black atoms—carbon, grey—hydrogen, red—oxygen, blue—nitrogen, light blue—fluorine)

Full size image

2.2.1 Experimental determination of adsorption kinetics

Kinetic measurements were performed to assess the equilibration time and also to determine which kinetic model was appropriate. For the norfloxacin kinetic experiments, a series of sample containers with 20 mL of stock solution and 10 mg of activated carbon were shaken on a laboratory shaker (Standard Analog Shaker, Model 3500, agitation speed 150 rpm, VWR, Radnor, PA, USA) for different time periods ranging from 0 to 420 min. Then the solution was filtered through 80 g∙cm−3 filtration paper (Swinnex, Merck, Darmstadt, Germany). For ofloxacin adsorption kinetics measurements, 10 mL of stock solution, 10 mg of activated carbon, and time intervals between 0 and 420 min were used.

2.2.2 Adsorption isotherms construction

The adsorption isotherms were constructed from the data determined at 23 °C, the pH of initial solutions was 7. Each solution was mixed with 10 mg of activated carbon and shaken for 24 h on a laboratory shaker. The volume of solution varied for each carbon material in order to obtain representative parts of adsorption isotherms, and then the suspension was filtered off (for details, see 2.2.).

3 Theory and calculations

3.1 Modelling of adsorption kinetics

To find the best quantitative description of the adsorption kinetics, the data were fitted by the pseudo-first-order, the pseudo-second-order, and the Elovich models.

The pseudo-first-order model (Eq. (2)) was first introduced by Lagergren [44]:

$$q\left(t\right)={\mathrm{q}}_{e}\left[1-exp\left(-{\mathrm{k}}_{1}t\right)\right]$$
(2)

where q (mg·g−1) represents the quantity of the adsorbed solute, qe (mg·g−1) adsorbed amount at equilibrium, t (min) time, and k1 (L·min−1) the pseudo-first-order rate constant.

The pseudo-second-order model of adsorption process (Eq. (3)) was proposed by Blanchard et al. [45]

$$q\left(t\right)={\mathrm{q}}_{\mathrm{e}}\frac{{\mathrm{k}}_{2}{\mathrm{q}}_{\mathrm{e}}t}{1+{\mathrm{k}}_{2}{\mathrm{q}}_{\mathrm{e}}t}$$
(3)

where k2 (g·mg−1·min−1) is a pseudo-second-order rate constant.

The Elovich model (Eq. (4)) was first published by Roginsky and Zeldovich [46] in the form of an empirical equation, for which the theoretical framework was later developed by Elovich and Zhabrova [47]. It is given as:

$$q\left(t\right)=\frac{1}{\upbeta }\mathrm{ln}\left(1+\mathrm{\alpha \beta }t\right)$$
(4)

where α is the initial sorption rate (mg·g−1·min−1), β is parameter related to the extent of surface coverage and the activation energy for chemisorption (mg·g−1).

3.2 Modelling of adsorption isotherms

Two models were tested for fitting the experimental data. The Langmuir adsorption isotherm [48], assuming monolayer formation as the adsorption limit (Eq. (5)):

$${q}_{e}=\frac{{q}_{max}{\mathrm{k}}_{\mathrm{L}}{c}_{e}}{1+{\mathrm{k}}_{\mathrm{L}}{c}_{e}}$$
(5)

where qmax (mg·g−1) is the maximum monolayer coverage capacity, ce is the concentration of solute at equilibrium (mg·L−1), and kL Langmuir constant (dm3·g−1).

The Freundlich model (Eq. (6)) is used to fit the adsorption isotherm of a heterogeneous system and assumes the possibility of a multilayer formation [49].

$${q}_{e}={\mathrm{k}}_{\mathrm{F}}{c}_{e}^{1/n}.$$
(6)

ce (mg·L−1) is the adsorbate concentration at equilibrium, kF (mg·g−1) is the Freundlich isotherm constant, and n adsorption intensity.

Redlich–Peterson isotherm [50] is an adsorption model that incorporates elements both from the Langmuir and Freundlich model [51]. It is given by the formula (Eq. (7))

$${q}_{e}=\frac{\mathrm{A}{c}_{e}}{1+\mathrm{B}{c}_{e}^{g}}$$
(7)

where A is a Redlich–Peterson isotherm constant (L·g−1), B is constant (L·mg−1) and g is an exponent that lies between 0 and 1.

3.3 Molecular modelling

Molecular structure calculations (i.e. the estimation of norfloxacin and ofloxacin molecular dimensions), reported in this work, were performed with the Gaussian 03 software package [52]. The equilibrium geometries were optimized with the hybrid density functional B3LYP [53, 54] in combination with the 6–31 + G(d,p) basis set. The molecular dimensions of optimized FQ molecules were evaluated by using the Avogadro advanced molecular editor and visualizer (Table 1).

4 Results and discussion

4.1 Activated carbons porous structure properties

The textural properties of the studied activated carbons are presented in Table 2. RMS-based activated carbon shows the highest specific surface as well as the highest micropore volume, followed by CC and MSEP. RMS-based activated carbon shows overall the highest porosity and surface area. CC possesses the highest microporosity (80%), followed by RMS (71%) and CH (70%). The highest surface of mesopores was observed in the case of both mango seed-based adsorbents MSEP and MSIP, followed by RMS and CC. In contrast to that, GS showed the lowest mesopore surface area as well as the lowest Vmicro/Vnet.

Table 2 Textural properties of activated carbons
Full size table

As shown in Fig. 2, RMS-, CC-, and MSEP-based activated carbons show the biggest pores. The pores present in the structures of the other three adsorbents are smaller and/or less frequent on the carbon surfaces.

Fig. 2
figure 2

SEM images of activated carbons studied in this work; a) RMS, b) MSEP, c) MSIP, d) CC, e) CH, and f) GS. Secondary and backscattered electrons, 5.000 × magnitude

Full size image

4.2 Adsorption kinetics

Figure 3 shows the kinetic curves of the norfloxacin adsorption process on the studied activated carbons. The adsorption rate was relatively rapid in the first 10 min, and then the adsorption process slowed down likely due to the continuous filling of the pores of the adsorbent with adsorbate until the equilibrium was reached. For all activated carbons, the Elovich model fitted best. The RMS-based activated carbon showed the highest α and lowest β parameter values (Table 3), indicating the best performance for norfloxacin sorption. Previous works concerning norfloxacin adsorption onto activated carbons [28, 30,31,32] were comparing pseudo-first and pseudo-second-order models and concluded that the latter fits norfloxacin adsorption kinetics better. The superior agreement of our adsorbents with the Elovich model than with the pseudo-second order indicates that the surfaces of our sorbents were energetically heterogeneous and therefore exhibiting different adsorption energies for chemisorption [55,56,57].

Fig. 3
figure 3

Kinetics of the adsorption of norfloxacin by activated carbons (solid line—pseudo-first-order model, dotted line—pseudo-second-order model, dashed line—Elovich model). c0 = 125 mg·L−1, cAC = 0.5 g·L−1, T = 23 °C, agitation = 150 rpm

Full size image
Table 3 Kinetic parameters of norfloxacin and ofloxacin adsorption
Full size table

The kinetic curves of ofloxacin adsorption are shown in Fig. 4. Most of the curves correspond best to the Elovich model except for the curve pertaining to RMS which fits fitted slightly better to the pseudo-second-order model. The data presented in Table 3 indicate that the pseudo-second-order model fits better to ofloxacin adsorption than the pseudo-first model, in accordance with the finding by Kong et al. [33].

Fig. 4
figure 4

Kinetics of the adsorption of ofloxacin by activated carbons (solid line—pseudo-first-order model, dotted line—pseudo-second-order model, dashed line—Elovich model). c0 = 250 mg·L−1, cAC = 1 g·L−1, T = 23 °C, agitation = 150 rpm

Full size image

A comparison between the kinetic parameters of norfloxacin and ofloxacin shows that norfloxacin has significantly higher Elovich α and β parameters. This reflects a higher rate of adsorption process at the beginning, but on the other hand also its steeper decrease with an increasing time of adsorption.

4.3 Adsorption isotherms

The amount of adsorbed solute increased sharply with an increasing equilibrium concentration until the adsorption sites on the activated carbon surface became gradually saturated with the adsorbed fluoroquinolone (Fig. 5). In the case of norfloxacin removal, the order of adsorption performance was RMS >  > CC ~ MSEP ≥ CH ≥ MSIP >  > GS. The isotherms of all six studied activated carbons belong to class H or high-affinity isotherms. This isotherm indicates the combined chemisorption and adsorption by electrostatic forces [58]. The isotherms fit best to the Redlich–Peterson model (RMS, MSEP, CC) and Freundlich model (CH, MSIP, GS) (Table 4). It is important to highlight that the Redlich–Peterson equation reduces to the Freundlich equations at higher values of equilibrium concentration [59]. Therefore, the adsorbents with better adsorbent properties followed rather the Redlich–Peterson model, whereas the adsorbents with lower adsorption performance corresponded better to the Freundlich model. Generally, we can state that for all six curves the Freundlich model fitted better than the Langmuir one. This means that a multilayer adsorption probably occurs on the surfaces of examined activated carbons. This agrees with Xie et al. [30], but not with Liu et al. [28] whose data fitted better to the Langmuir isotherm. Undoubtedly, the course of adsorption isotherms depends not only on the adsorbate, but also on the textural and surface properties of the adsorbent. Wang et al. [60] obtained isotherms corresponding slightly better to the Langmuir model. Peng et al. [36] worked with amine-functionalized magnetic bamboo-based activated carbon and found that the isotherms fitted better to the Langmuir model.

Fig. 5
figure 5

Adsorption isotherms of norfloxacin on activated carbons (solid line—Freundlich model, dotted line—Redlich–Peterson model, dashed line—Langmuir model). c0 = 15–150 mg·L−1, T = 23 °C, agitation = 150 rpm

Full size image
Table 4 Parameters of norfloxacin and ofloxacin adsorption isotherms
Full size table

In the case of ofloxacin removal (Fig. 6), the adsorption performances of activated carbons follow the order: RMS >  > CC ~ MSEP > CH ≥ MSIP >  > GS. The isotherms also follow the H type; therefore, one can state that chemisorption and attraction by electrostatic forces occurred during ofloxacin adsorption [58]. As shown in Table 4, the Redlich–Peterson model fitted best for five of six samples, excluding the one with lowest adsorption performance—GS. This observation is similar to the one obtained by Wang et al. [32] and Kong et al. [33], whose data also corresponded better to the Freundlich model than to the Langmuir one. In contrast, Wang et al. [35] reported that the adsorption of ofloxacin and norfloxacin onto the loofah activated carbon with hierarchical structures followed the Langmuir model.

Fig. 6
figure 6

Adsorption isotherms of ofloxacin on activated carbons (solid line—Freundlich model, dotted line—Redlich–Peterson model, dashed line—Langmuir model). c0 = 25–250 mg·L−1, T = 23 °C, agitation = 150 rpm

Full size image

RMS-based activated carbon showed remarkable adsorption properties. The experimental qmax values were the highest of the studied carbons. In contrast to this, GS-based activated carbon had the lowest qmax. This agrees with the textural properties of activated carbons listed in Table 2. RMS shows the highest net pore volume (0.778 cm3liq·g−1), as well as micropore volume (0.555 cm3liq·g−1) and high mesopore surface area, similar to MSEP- and CC-based activated carbons which belong among the best adsorbents for both fluoroquinolones. Since the microporosity (Vmicro/Vnet) of an adsorbent plays a role, representing the generally large surface area available for the adsorption of the fluoroquinolones, it should be pointed out that RMS, MSEP, and CC show the high microporosity of 71, 68, and 80%, respectively, furthermore showing the highest total pore volumes. With respect to the similar median of the micropore width/size of all investigated activated carbons, moving in the very small range of 0.54–0.59 nm, it is evident that the micropore volume and net pore volume will be determined through the textural properties for adsorption.

The effects of individual textural parameters on the adsorption properties of the studied activated carbons are visualized in Fig. 7. The volume of micropores (Vmicro) was the most determining parameter for the adsorption of norfloxacin and ofloxacin. Contrary to that, adsorption properties did not correlate with the surface of mesopores (Smeso) well. Specific surface area, SBET, is not considered in Fig. 7 because BET specific surface area measurement does not fit well for microporous materials [61]. This parameter is a close correlation between qmax and Vmicro and was found in both norfloxacin and ofloxacin cases (Fig. 8). The linear regressions did not cross the origin; hence we can consider a slight adsorption on the external surface or on the surfaces of the mesopores.

Fig. 7
figure 7

Comparison of the effects of individual textural parameters on adsorption of norfloxacin and ofloxacin. Relative values (0–1), related to the maximum value for a given parameter among sorbents, are plotted on the axes. The proximity of the points on the individual axes indicates a positive correlation between the parameters

Full size image
Fig. 8
figure 8

Correlation between Vmicro and qmax of norfloxacin and ofloxacin

Full size image

The adsorption capacities obtained in this work with that reported in the literature are presented in Table 5.

Table 5 Adsorption capacities and kinetic and equilibrium models reported in the literature and obtained in this work
Full size table

4.4 Evaluation of the influence of surface chemistry

Figure 9 shows the correlation between adsorption properties and the surface chemistry of individual activated carbons and Fig. 10 presents the pH values of fluoroquinolones solutions in relation to pKa values of both antibiotics and to pHPZC of individual activated carbons. The parameters used for the construction of both figures were published by Cruz et al. [42]. Adsorption of fluoroquinolones onto RMS took place under equilibrium pH on a deprotonated surface. Low coverage of RMS surface with oxygen-containing functional groups reduced the negative effect of equilibrium pH. Favourable adsorption properties indicate that π electrons of the surface brought a positive effect on the adsorption capacity. Norfloxacin in anionic form showed a higher tendency (according to R2 of the isotherms modelling) to multilayer adsorption than ofloxacin in zwitterionic form with lowest R2 value for the Freundlich model and highest g value for the Redlich–Peterson one.

Fig. 9
figure 9

Correlation between the share of various functional groups and qmax of norfloxacin and ofloxacin

Full size image
Fig. 10
figure 10

Scheme of pH values of fluoroquinolones solutions (solid circles) in relation with pKa values of both antibiotics (solid lines) and with pHPZC of individual activated carbons (dotted lines). For each antibiotic, an ionic form of molecule (cationic, anionic a zwitterionic) in different pH regions defined by pKa values is shown

Full size image

The surface of CC-based activated carbon had a low amount of oxygen-containing groups, which means a lower negative influence of pH and surface charge. The adsorption of ofloxacin took place at pH almost corresponding to the pHPZC value, i.e. the surface charge was minimal. In this case, ofloxacin was present in the solution in zwitterionic form. The reduction of the effect of the interaction of the surface charge with dipoles of the ion may have favoured the interactions of π electrons (R2 value for the Freundlich model was lower in comparison with other sorbents).

MSEP- and CH-based activated carbons had a similar share of oxygen-containing groups and C = C bonds; both adsorbents showed a similar R2 value for the Langmuir model fit and a similar value of the parameter of non-homogeneity g in the case of ofloxacin removal. For norfloxacin adsorption, MSEP had highest Langmuir R2 value and highest Redlich–Peterson g (i.e. highest homogeneity) of all adsorbents. Equilibrium pH of the adsorption of norfloxacin on CH was higher than pHPZC, i.e. the adsorbent surface was slightly deprotonated with a predominantly negative charge, whereas the norfloxacin molecule was present as a zwitterion. This combination led to the most distinct deviation from the Langmuir model of all adsorbents. In contrast to this, during the adsorption of ofloxacin the pH stabilized on a value very close to pHPZC and the surface was close to the state of minimal coverage of charge. Ofloxacin was present in zwitterionic form. Unlike the case of CH-norfloxacin interaction, the Langmuir model fit was closer for CH than for other sorbents.

The GS-based adsorbent had a lot of oxygen-containing groups on its surface, so we can expect a strong influence of the negative surface charge when pH > pHPZC.

Considering the influence of surface homogeneity, represented by a Redlich–Peterson parameter g, we can state that a more distinctive negative effect of oxygen-containing groups and more distinctive positive effect of π bonds on homogeneity was observed in the case of ofloxacin adsorption compared to the removal of norfloxacin.

4.5 Molecular modelling

Considering the molecular dimensions of norfloxacin and ofloxacin estimated from theoretical computations (Table 1, Fig. 1), it can be said that their x- and y-axis dimensions are very similar. With respect to z-axis, norfloxacin is larger than ofloxacin, but still both molecules may enter into the smallest micropores (pore width ~ 0.4 nm). Thus, based on the estimated molecular dimensions of both fluoroquinolones, it can be assumed that there should not exist any favourable steric shielding of micropores for the entering of norfloxacin or ofloxacin, and the adsorption of both molecules should proceed very similarly and comparably, also with respect to the functional groups present (e.g. –COOH) [42]. Compared to the majority of carbonaceous materials studied in previous works focused on the adsorption of norfloxacin or ofloxacin [28, 30,31,32,33], RMS-based activated carbon studied in this paper belongs to the most efficient adsorbents. In particular, this adsorbent seems to be very promising for the removal of pharmaceuticals from waters. This fact makes RMS-based activated carbon a material of potential interest for wastewater treatment for, e.g. hospitals, the pharmaceutical industry, or sewage treatment plants. Both fluoroquinolones were adsorbed comparatively efficiently on individual investigated activated carbons; this fact corresponds to the very similar molecule dimensions of both fluoroquinolones.

5 Conclusions

Three activated carbons from the six investigated, prepared from six different agricultural waste biomasses, proved to be very suitable for the removal of norfloxacin and ofloxacin from water. Red mombin seed-based activated carbon showed the best adsorption properties for both pharmaceuticals.

In most cases, the Elovich kinetic model fitted best, which indicates that the prepared adsorbents have heterogeneous adsorbing surface sites. Adsorption isotherms corresponded to the Freundlich model and to the Redlich–Peterson model, which shows that the mechanism of adsorption onto the examined activated carbons accords rather to the multilayer formation. High micropore volume is of the cause of the enhanced adsorption performance of RMS-, MSEP-, and CC-based activated carbons.