Introduction

Quinolones were widely used to treat human and animal diseases (Lombardo-Agüí et al. 2015; Alcaráz et al. 2015). However, a large quantity of quinolones were emitted into the environment through human and animals’ excrement, causing environmental pollution because of their bacterial resistance, allergy, and toxicity (Junza et al. 2014; Hou et al. 2014). Therefore, extensive use of quinolones will pose harmful effect on ecological environment and human health (Moreno-González et al. 2014; Douros et al. 2015). Among quinolone antibiotics, fleroxacin (FLRX), enoxacin (EN), norfloxacin (NOR), ciprofloxacin (CIP), enrofloxacin (ENRO), and lomefloxacin hydrochloride (LOME) were often detected in environment waters (Lang et al. 2018; Zhang and Cheng, 2019). Currently, high performance liquid chromatography (HPLC) and gas chromatography mass-spectrometry (GC-MS) were frequently applied for simultaneous determination of these quinolones (Xie et al. 2015; Rocha et al. 2015). However, these methods faced a drawback that a rigorous pretreatment process was needed before analysis, which required great labor and time. Additionally, little work had been done to carry out simultaneous determination of multiple quinolones using a relatively simple and sensitive pretreatment method. In our previous study, a hyphenated technique using magnetic molecular imprinting polymers based solid phase extraction (SPE) coupled with dispersive liquid-liquid microextraction (DLLME) were frequently applied in the sample pretreatment procedures, which presented simplicity, selectivity, and high enrichment efficiency (Xie et al. 2019; Guo et al. 2019; Li et al. 2018).

Molecularly imprinted polymer (MIP) with template-created cavities has been widely used as molecular recognition material in various fields, such as solid phase extraction, drug delivery, biosensors, and protein crystallization (Li et al. 2015; Huang et al., 2015a, b, 2017; Liu et al. 2016, 2018, 2019; Wang et al. 2010; Lai et al. 2017). This could be attributed to their high selectivity, strong anti-interference ability, and good stability (Chang et al. 2012; Gu et al. 2015; Han et al. 2015; Pourjavadi et al. 2015). However, traditional synthesis methods of MIP such as precipitation polymerization, bulky organic co-polymerization, and so on have a disadvantage that plenty of the recognition sites in the particles are embedded in the interior area (Wang et al. 2010). On the contrary, surface molecular imprinting technology can overcome the above drawback and generate more affinity sites on the surface of MIP. In addition, to achieve high molecular recognition ability, an effective carrier of MIP is very important. Graphene oxide (GO) has high surface area and rich functional groups, and mesoporous silica (mSiO2) owns tunable pore sizes and large surface area; both of them have many applications in adsorption (Dreyer et al. 2014; Wang et al. 2019; Zhang and Cheng, 2019) Li et al. 2018). In our previous researches, mSiO2 supported on the surface of magnetic GO was used as the carrier of MIPs, and the MIPs showed excellent adsorption capacity, rapid kinetics, and high selectivity for analysis of organic pollutants in water (Xie et al. 2019; Guo et al. 2019; Li et al. 2018).

In this study, a new multi-templates MIP, with FLRX, EN, NOR, CIP, ENRO, and LOME as the templates, was successfully synthesized for simultaneous preconcentration and determination of quinolones in water. In the synthesis process, mesoporous silica modified magnetic graphene oxide was utilized as the carrier. The adsorption behaviors of MIP for these quinolones were investigated. Besides, a coupling of the MIP-based SPE with dispersive liquid-liquid microextraction (DLLME), another efficient sample pretreatment technique (Gao and Ma 2011; Ebrahim et al. 2015; Zhang and Cheng, 2019), was developed, and the conditions for SPE and DLLME were optimized. Finally, the SPE-DLLME-HPLC method was used to separate and detect trace amount of FLRX, EN, NOR, CIP, ENRO, and LOME in real water samples.

Materials and methods

Chemicals

Natural flake graphite (China National Medicine Corporation Ltd), fleroxacin (FLRX, 98 wt%), enoxacin (EN, 98 wt%), norfloxacin (NOR, 99 wt%), ciprofloxacin (CIP, 99 wt%), enrofloxacin (ENRO, 99 wt%), lomefloxacin (LOME, 98 wt%), ofloxacin (98 wt%), and gatifloxacin (99 wt%), 4-vinylbenzoic acid (98 wt%), 2,2′-azobisisobutyronitrile (AIBN, 98 wt%), ethylene glycol dimethacrylate (EGDMA, 98 wt%), tetraethyl orthosilicate (TEOS, 98 wt%), vinyltrimethoxysilance (VTTS, 98 wt%), cetyl trimethylammonium bromide (CTAB, 98 wt%), and HPLC-grade tetrachloroethylene, chlorobenzene, tetrachloroethane, carbon tetrachloride, methanol, and formic acid were bought from Aladdin Reagent (Shanghai, China).

Analytical methods

High performance liquid chromatography (HPLC, PDA, C18 column (250 × 4.6 mm, 5 μm), Shimadzu Corporation, Japan) was applied to detect the quinolone antibiotics. Other analytical condition were as follows: PDA detection at 270 nm; column temperature at 40 °C; injection volume of 10 μL; elution rate at 1 mL min−1 with water/formic acid (9/1, V:V) as eluent A and methanol as eluent B, from 80% B to 70% B in 0–3 min, 70% B to 60% B in 3–8 min, 60% B to 55% B in 8–13 min, 55% B to 60% B in 13–17 min, and 60% B in 17–20 min. Under these test conditions, the retention times of FLRX, EN, NOR, CIP, ENRO, and LOME were 10.9, 11.9, 12.8, 13.4, 13.6, and 14.1 min, respectively, and the chromatogram was showed in Fig. S1.

Preparation of multi-templates MIP

The synthesis procedures of magnetic graphene oxide embellished with mesoporous silica (MGO@mSiO2) and (MGO@mSiO2) modified with vinyl groups (VTTS-MGO@mSiO2) were introduced in detail in our previous work (Xie et al. 2019; Guo et al. 2019; Li et al. 2018). The synthesis steps of MIP were as follows: first, 30 mg FLRX, 26 mg EN, 26 mg NOR, 27 mg CIP, 29 mg ENRO, and 31 mg LOME (1:1:1:1:1:1, molar ratio) blended with 221 mg 4-vinylbenzoic were dispersed in 50-mL toluene and undergo an ultrasound for 30 min, then reacting under air bath shaking table for 4 h at 15 °C. After that, 300 mg VTTS-MGO@mSiO2, 50 mg AIBN, and 471 μL EGDMA were added to the mixture and then the nitrogen was used to deoxygenate under ice-cooling. The reaction was processed for 24 h at 60 °C. To completely remove templates, the obtained MIP was eluted with the mixtures of methanol and acetic acid (9:1, V:V) until no templates were detected using HPLC. Meanwhile, the non-imprinted polymer (NIP) was synthesized using the abovementioned ways in the lack of the target templates.

Batch adsorption experiments

The influences of pH and initial concentrations of quinolones on the adsorption performance of the MIP (or NIP) were investigated. A total of 20 mg MIP or NIP was dispersed in methanol/water (4:1, V: V) solutions of FLRX, EN, NOR, CIP, ENRO, and LOME with pH values from 3 to 9, and shaken for 1 h. The sorption isotherm tests were performed at three constant temperatures (298 K, 308 K, and 318 K). Then, the saturated solutions and MIP (or NIP) were separated by an external magnet. Next, the residual concentrations of FLRX, EN, NOR, CIP, ENRO, and LOME were detected using HPLC system. The adsorption capacity was calculated as

$$ {Q}_e=\left({C}_o-{C}_e\right)\ V/M $$
(1)

where Qe (mg/g) represents the adsorption capacity of the MIP (or NIP), Co and Ce stand for the initial and equilibrium concentration (mg/L), and V (L) and M (mg) represent the solution volume and adsorbent mass, respectively.

Adsorption selectivity and regeneration experiments

Ofloxacin and gatifloxacin (the retention times of ofloxacin and gatifloxacin were 11.9 and 15.8 min, respectively, and the chromatogram was showed in Fig. S2.) were chosen to prove the selectivity of MIP toward FLRX, EN, NOR, CIP, ENRO, and LOME. In total, 20 mg MIP or NIP was dispersed in methanol/water (4:1, V:V) solutions of FLRX, EN, NOR, CIP, ENRO, and LOME and shaken for 1 h to adsorption equilibrium with adsorption. Then the residual concentration of FLRX, EN, NOR, CIP, ENRO, and LOME was detected using HPLC. Meanwhile, the used MIP (or NIP) was washed in methanol/acetic acid to entirely remove the adsorbed quinolones, and the above steps were repeated five times to investigate the reusability. And the static distribution coefficient Kd, selectivity coefficient, and relative selectivity coefficients ko were calculated as

$$ {K}_d={Q}_e/{\mathrm{C}}_{\mathrm{e}} $$
(2)
$$ k={K}_d\left(\mathrm{template}\ \mathrm{molecule}\right)/{K}_d\left(\mathrm{competitive}\ \mathrm{molecule}\right) $$
(3)
$$ {k}_o={k}_{MIP}/{k}_{NIP} $$
(4)

SPE-DLLME procedures

SPE and DLLME were applied as pretreatment ways to enrich the trace amounts of quinolones in real water samples. The procedures were as follows: first, 20 mg MIP as adsorbent was immersed into 50 mL water sample in a round-bottom flask, and then was stirred for 10 min. Next, the MIP was gathered with an external magnet and washed by deionized water. After that, the templates FLRX, EN, NOR, CIP, ENRO, and LOME were eluted by methanol. Finally, the obtained methanol eluent was used as the dispersant and tetrachloroethane as the extractant for the following DLLME. The formed emulsion was centrifuged and the sedimentary phase was gathered and then detected using HPLC.

Results and discussion

Preparation of multi-templates MIP

The amount of cross-linker and 4-vinylbenzoic acid for the preparation of MIP were optimized in the orthogonal design, and the results were presented in Table S1). The results showed that when 4-vinylbenzoic acid was 221 mg and cross-linker was 471 μL, the synthesized MIP had high adsorption capacity.

Adsorption study

Effects of pH on adsorption

The influence of the solution pH in the chemisorption process was tested in this study. The Fig. 1 showed that the adsorption capacity of MIP for FLRX, EN, NOR, CIP, ENRO, and LOME increased with the increased solution pH and reached to the highest at pH 6, which could be ascribed to the hydrogen bond and π-π interaction between quinolones and MIP. Thus, the solution pH 6 was selected in the following experiments.

Fig. 1
figure 1

Effect of pH on the adsorption of MIPs for FLRX, EN, NOR, CIP, ENRO, and LOME

Full size image

Adsorption isotherm analysis

The adsorption isotherms for six quinolones at three constant temperatures (298 K, 308 K, and 318 K) were shown in Fig. 2. The results indicated that the adsorption capacity of MIP raised fast and then maintained flat with the initial concentrations improving, suggesting that the recognition sites for FLRX, EN, NOR, CIP, ENRO, and LOME reached a saturation. Meanwhile, the NIP showed the similar trend but with lower adsorption capacity. The Langmuir, Freundlich, and Tempkin models were chosen for subsequent analysis and the relevant fitting parameters were shown in Table 1 and Figs. 3, 4, and 5. The results showed that the correlation coefficients (R2) for Langmuir model (> 0.98) were significantly higher than those of both the Freundlich (> 0.44) and Tempkin models (> 0.91). Obviously, the Langmuir model could well describe the adsorption system, indicating that the adsorption process acted up to the monolayer sorption mechanism.

Fig. 2
figure 2

Adsorption isothermic curves of FLRX (a), EN (b), NOR (c), CIP (d), ENOR (e), and LOME (f) on MIP and NIP with the fitting to the Langmuir model at different temperatures

Full size image
Table 1 Adsorption isotherm parameters of FLRX, EN, NOR, CIP, ENRO, and LOME onto the MIP under different temperatures
Full size table
Fig. 3
figure 3

The Langmuir model fitting curves of FLRX (a), EN (b), NOR (c), CIP (d), ENRO (e), and LOME (f) on MIP and NIP under different temperatures

Full size image
Fig. 4
figure 4

The Freundlich model fitting curves of FLRX (a), EN (b), NOR (c), CIP (d), ENRO (e), and LOME (f) on MIP and NIP under different temperatures

Full size image
Fig. 5
figure 5

The Tempkin model fitting curves of FLRX (a), EN (b), NOR (c), CIP (d), ENRO (e), and LOME (f) on MIP and NIP under different temperatures

Full size image

The Langmuir, Freundlich, and Tempkin models are shown as follows:

$$ {\mathrm{C}}_{\mathrm{e}}/{Q}_e={\mathrm{C}}_{\mathrm{e}}/{Q}_m+1/{bQ}_m $$
(5)
$$ {LnQ}_e={LnK}_F+ Ln{\mathrm{C}}_{\mathrm{e}}/\mathrm{n} $$
(6)
$$ {Q}_e=\mathrm{a}+\mathrm{b} Ln{\mathrm{C}}_{\mathrm{e}} $$
(7)

where Ce (mg/L), Qe (mg/g), and Qm (mg/g) stand for the equilibrium concentration, equilibrium, and maximum adsorption capacities, respectively; KF (mg/g) represents the Freundlich binding coefficient; and a, b, and n represent the constants.

Adsorption thermodynamics

The temperature influence on the adsorption of MIP for FLRX, EN, NOR, CIP, ENRO, and LOME was shown in Table 2 and Fig. 6. The results showed that the adsorption capacity of MIP was higher at a lower temperature, and the values of △G0 and of △H0 were all negative, implying that the adsorption process was a spontaneous endothermic reaction, which had good agreement with the isotherm experiments.

Table 2 Thermodynamic parameters of FLRX, EN, NOR, CIP, ENRO, and LOME onto the MIP at various temperatures
Full size table
Fig. 6
figure 6

Adsorption thermodynamics fitting curves of FLRX, EN, NOR, CIP, ENRO, and LOME onto MIP and NIP

Full size image

The adsorption process was investigated through the following equations:

$$ Ln{K}_L=-\varDelta H/\mathrm{R}T+\varDelta S/\mathrm{R} $$
(8)
$$ {K}_L=\mathrm{bM} $$
(9)
$$ \Delta G=\Delta H-T\Delta S $$
(10)

where R, T, and b stand for the universal gas constant (8.314 J mol−1 K−1), Kelvin temperature (K), and the affinity constants, respectively; M represents the relative molecular mass of FLRX, EN, NOR, CIP, ENRO, and LOME, respectively; ΔH (KJ mol−1), ΔS (KJ mol−1), and ΔG represent the standard enthalpy changes, the entropy changes, and the Gibbs free energy changes, respectively.

Adsorption selectivity and regeneration

Ofloxacin and gatifloxacin were chosen to investigate the selectivity of MIP toward FLRX, EN, NOR, CIP, ENRO, and LOME. The results shown in Table 3 and Fig. 7 indicated that the adsorption amounts of MIP toward FLRX, EN, NOR, CIP, ENRO, and LOME were evidently higher than those of loxacin and gatifloxacin. On the contrary, the NIP represented similar adsorption capacity to the competitive molecules, suggesting that the specific recognition sites on the surface of MIP were successfully synthesized. Meanwhile, the adsorption capacity of VTTS-MGO@mSiO2 was investigated to demonstrate the imprinting influence of MIP. The results showed that the adsorption capacity of VTTS-MGO@mSiO2 was much lower than MIP.

Table 3 Partition coefficient and selectivity coefficient of FLRX, EN, NOR, CIP, ENRO, and LOME selective adsorption
Full size table
Fig. 7
figure 7

Competitive adsorption for tow analogue of different adsorbent by VTTS-MGO@mSiO2, MIP, and NIP

Full size image

Besides, the regeneration capacity of MIP was also studied. As shown in Fig. 8, after five regeneration cycles, the adsorption capacity of MIP toward FLRX, EN, NOR, CIP, ENRO, and LOME is still maintained at a high level, suggesting that MIP had good reusability.

Fig. 8
figure 8

The reusability analysis of MIP for FLRX, EN, NOR, CIP, ENRO, and LOME

Full size image

Optimization of SPE-DLLME conditions

The conditions of SPE and DLLME were optimized, and the results were shown in Fig. 9 and 10. As can be seen in SPE, 50 mL solution at pH 6 was tested and the adsorption arrived at equilibrium after 12 min. A total of 3-mL methanol was applied to elute the adsorbed quinolones for 6 min. In DLLME, 50-μL tetrachloroethane was used as the extractant and the eluent of SPE as the dispersant. Then, the recovery and enrichment coefficient of FLRX, EN, NOR, CIP, ENRO, and LOME were acquired under the above optimal parameters.

Fig. 9
figure 9

Effect of adsorption time (a), pH (b), extration solution (c), eluent volume (d), desorption time (e), and volume of water samples (f) on the recovery of quinolones

Full size image
Fig. 10
figure 10

Effect of type of extractant (a), volume (b, c), and pH (d, e) on the recovery (RE) and the enrichment factor (EF) of quinolones

Full size image

Environmental water analysis

The developed SPE-DLLME-HPLC method was used to separate and detect trace FLRX, EN, NOR, CIP, ENRO, and LOME in real water samples (a lake water sample, a well water sample, and a Pearl River water sample). The results were shown in Table S2. As can be seen, in the lake water and river water, NOR, CIP, and ENRO were detected with a concentration of 1.21, 1.52, and 1.64 and 0.96, 1.26, and 1.34 μg/L, respectively. And CIP and ENRO were found at a concentration of 1.01 and 0.88 μg/L in well water, respectively. In addition, the recoveries of the quinolones were 89.67–100.5%, the RSDs of the results were 3.59–7.12%, and the limits of detections (LODs) for FLRX, EN, NOR, CIP, ENRO, and LOME were 0.013, 0.012, 0.011, 0.012, 0.015, and 0.015 μg/L, respectively, indicating that the SPE-DLLME-HPLC method represented outstanding applicability for the rapid detection of trace FLRX, EN, NOR, CIP, ENRO, and LOME in real water samples.

Comparison with other methods

The comparison with other reported papers was presented in Table 4 (Tang et al. 2018; Zhu et al. 2019; Lian and Wang, 2016; David et al. 2014; Francisco et al. 2019). The MIP synthesized in this study could rapidly and simultaneously separate and detect trace six quinolones in real water. Meanwhile, as we can see, the method showed a lower linear range and LOD, a better extraction efficiency, and absorbing capacity toward FLRX, EN, NOR, CIP, ENRO, and LOME than those researches, meaning the sensitivity and feasibility of the suggested method possessed pretty good analytical performance.

Table 4 Comparison of the MSPE-DLLME-HPLC method based on MIP with other analytical methods
Full size table

Conclusions

An effective multi-templates molecularly imprinted polymer has been successfully prepared, with mesoporous silica supported onto the surface of magnetic graphene oxide as the carrier, to simultaneously extract and detect quinolones (FLRX, EN, NOR, CIP, ENRO, and LOME) in real water samples. The MIP presented high selectivity and adsorption capacity for the quinolones. Meanwhile, the SPE-DLLME-HPLCMIP method was used to separate and detect trace FLRX, EN, NOR, CIP, ENRO, and LOME in real water samples, and satisfactory results were obtained.