*3.2. Characterization*

The morphology of the obtained FeOx/ZnS NPs and FeOx/ZnS@MIPs was investigated by transmission electron microscopy. As shown in Figure 2a, the particles of FeOx/ZnS NPs were spherical in morphology, and they were about 135.2 ± 16.7 nm. The magnetic nanoparticles were tightly surrounded by ZnS: Mn2<sup>+</sup> QDs@MEA through the EDC/NHS reaction, which created a rough surface on the FeOx/ZnS NPs. After the molecular imprinting process, the size of the resulting FeOx/ZnS@MIPs increased to about 198.6 ± 13.5 nm and displayed a smooth surface (Figure 2b). As can be seen, an MIP layer had been well coated on the surface of FeOx/ZnS NPs, and an interface could be clearly distinguished between the inner FeOx/ZnS NPs core and the imprinting polymer shell.

**Figure 2.** TEM images of (**a**) FeOx/ZnS NPs and (**b**) FeOx/ZnS@MIPs.

The structure of ZnS: Mn2<sup>+</sup> QDs@MEA, FeOx@COOH and FeOx/ZnS NPs was analyzed by FT-IR spectroscopy. As shown in Figure 3, the characteristic Fe–O bands were present in FeOx/ZnS (632 cm<sup>−</sup>1) and FeOx@COOH MNPs (594 cm<sup>−</sup>1), respectively. Peaks at 1598 cm<sup>−</sup><sup>1</sup> (amino groups), 925 cm<sup>−</sup><sup>1</sup> (bending vibration) and the disappeared peaks at 2550–2670 cm<sup>−</sup><sup>1</sup> (S–H thiol group) in the curve b indicated that the MEA was modified on the surface of ZnS: Mn2<sup>+</sup> QDs through covalent bonds formed between thiols and Zn2<sup>+</sup> surface atoms [50]. The Peak at 1413 cm<sup>−</sup><sup>1</sup> corresponded to C–O on the FeOx, and a peak at 1605 cm<sup>−</sup><sup>1</sup> corresponded to the vibration of water molecules adsorbed on Fe3O4 [13]. Compared with the ZnS: Mn2<sup>+</sup> QDs@MEA and FeOx@COOH (curve b and c), FeOx/ZnS NPs (curve a) showed characteristic peaks at 3425 cm<sup>−</sup><sup>1</sup> (N–H stretching vibration), 1625 cm<sup>−</sup>1(N–H bend), and 1400 cm<sup>−</sup>1(C–N stretching vibration), revealing that an amide bond was formed. The results suggested that the ZnS: Mn2<sup>+</sup> QDs@MEA and FeOx@COOH were successfully combined through an EDC/NHS reaction process.

To study the influence of surface modification on the magnetic behavior of FeOx@COOH MNPs, FeOx/ZnS NPs and FeOx/ZnS@MIPs, the VSM magnetization curves of the as-prepared magnetic materials were compared in Figure 4. As shown in Figure 4, all of the magnetic hysteresis loops of the magnetic materials displayed a typical super-paramagnetic characteristic and high magnetization, and the magnetization saturation values of FeOx@COOH MNPs, FeOx/ZnS NPs, and FeOx/ZnS@MIPs were about 57.348, 45.1033, 24.5796 emu g<sup>−</sup>1, respectively. The magnetization saturation value of prepared NPs was lower than that of pure Fe3O4 magnetite (about 80 emu g<sup>−</sup>1), which is due to some mixed Fe2O3 that lowers the magnetization saturation. The as-prepared magnetic NPs still demonstrated a strong magnetic response and FeOx was used to represent the mixture to describe the magnetic behavior in this paper. The magnetization saturation values of FeOx/ZnS@MIPs were lower than that of FeOX@COOH MNPs, which may be attributed to the MIP shell on the surfaces of FeOX/ZnS NPs. Additionally, the as-prepared magnetic nanocomposites also showed a rapid

magnetic response to the applied magnetic field. Once the magnetic field was removed, the magnetic nanocomposites homogeneously and quickly redispersed with a slight shake. This demonstrates that the FeOX/ZnS@MIPs possess rapid magnetic responsivity and good dispersibility, which enables them to be practically used for the rapid separation or enrichment of analyte in complex samples.

**Figure 3.** FT-IR spectra of (**a**) FeOx/ZnS NPs, (**b**) ZnS: Mn2<sup>+</sup> QDs@MEA and (**c**) FeOx@COOH nanoparticles.

**Figure 4.** The magnetic hysteresis loops of (**a**) FeOx@COOH magnetic nanoparticles, (**b**) FeOx/ZnS NPs, and (**c**) FeOx/ZnS@MIPs.

The optical properties of ZnS: Mn2<sup>+</sup> QDs@MEA and FeOx/ZnS NPs were investigated by UV–Vis spectra and fluorescence spectra. As shown in Figure S2a,b, the QDs and FeOx/ZnS NPs exhibit semblable absorption spectrum, indicating the QDs successfully combined with the FeOx MNPs. Compared with ZnS: Mn2<sup>+</sup> QDs@MEA, the absorption peak of FeOx/ZnS NPs is less pronounced, which attributed to the broad and strong absorption of the combined Fe3O4 MNPs [51]. The ZnS: Mn2<sup>+</sup> QDs@MEA and FeOx/ZnS NPs show well-resolved emission spectra, with both of the maximum emission peaks located at 586 nm. The characteristic emission peak of FeOx/ZnS NPs corresponded to the ZnS: Mn2<sup>+</sup> QDs@MEA, implying that the fluorescence property was not significantly affected in the EDC/NHS reaction process. Both samples possess a narrow and symmetrical emission peak and large Stokes shift, which is more suitable for fluorescence labeling in vivo or biosensing target analytes in complex biological samples.

#### *3.3. Fluorescence Response to Time and Adsorption Kinetics*

The effect of reaction time on the fluorescence intensity and the adsorption kinetics curves of BPA onto FeOx/ZnS@MIPs are presented in Figures S3 and S4, respectively. As shown in Figure S3, the fluorescence intensity of FeOx/ZnS@MIPs showed a rapid decrease in the first 2 min and achieved stable fluorescence intensity after being incubated for 5 min, which corresponded closely to the adsorption equilibrium time (Figure S4). The adsorption equilibrium for the FeOx/ZnS@NIPs emerged at 4 min in the experiment; however, the fluorescence quenching and adsorption amount of FeOx/ZnS@NIPs was lower than that of the MIP ones. This is due to the non-specific binding cavities that are formed in the FeOx/ZnS@NIPs' synthesis process, the non-specific adsorption was dominant in the FeOx/ZnS@NIPs, which resulted in a lower binding capacity and lower fluorescence quenching. These results verified that the optimal reaction time of the FeOx/ZnS@MIPs for detecting BPA was 5 min.

## *3.4. Binding Performance*

The adsorption isotherm was investigated through a batch affinity adsorption experiment of FeOx/ZnS@MIPs. A series of BPA standard samples were used to evaluate the adsorption isotherms of BPA on the FeOx/ZnS@MIPs and FeOx/ZnS@NIPs. As shown in Figure 5, the adsorption capacity of FeOx/ZnS@MIPs increased quickly with the increasing concentration of BPA. The FeOx/ZnS@MIPs exhibited a higher binding amount of BPA than that of the FeOx/ZnS@NIPs. Because of the 3D-imprinted cavities in FeOx/ZnS@MIPs, which possessed better chemical and structure matching with the template BPA, a large number of empty recognition cavities were available in the FeOx/ZnS@MIPs, which enabled BPA molecules to easily enter, and resulted in excellent adsorption capacity. To further study the adsorption performance of FeOx/ZnS@MIPs, the experimental data were fitted with Langmuir and Freundlich isotherm models. Fitting results showed that the binding properties were best described by the Langmuir isotherm model (R = 0.9930), which revealed that the adsorption behavior of FeOx/ZnS@MIPs was basically monolayer adsorption onto a surface with a homogeneous system. The Langmuir isotherm equation of FeOx/ZnS@MIPs for BPA was *C*e/*Q* = 0.2302 + 0.01964 *C*e, and the maximum adsorption capacity calculated by the Langmuir isotherm model was 50.92 mg g<sup>−</sup>1.

**Figure 5.** The adsorption isotherm and Langmuir fit of FeOx/ZnS@MIPs. Experimental conditions: citrate buffer solution (0.02 mol <sup>L</sup>−1, pH = 6.4), room temperature.

#### *3.5. Fluorescence Sensing of BPA*

To maintain the fluorescence stability of the FeOx/ZnS@MIPs, all experiments were performed in citrate buffer solution at pH 6.4. The typical fluorescence quenching of imprinted FeOx/ZnS NPs in the concentration of BPA ranged from 1.0 to 80 ng mL−1, as shown in Figure 6. It can be seen that the FeOx/ZnS@MIPs showed noticeable fluorescence emission responses to different concentrations of BPA. As a control, the FeOx/ZnS@NIP was slightly quenched by BPA because of there being no specific recognized site on the surface of the NIPs; thus, fewer BPA molecules were bound by non-specific interactions. In contrast, the fluorescence emission of the FeOx/ZnS@MIPs quenched gradually with the increasing concentration of BPA. The fluorescence intensity of FeOx/ZnS@MIPs in the presence of BPA (80 ng mL−1) was only 32.7% compared to that in the absence of BPA. The experiment results showed that the fluorescence quenching of the FeOx/ZnS@MIPs depended on the specific binding with template BPA. Therefore, FeOx/ZnS@MIPs can be used for the detection of trace BPA. Under optimal conditions, the plot of fluorescence intensity change (*F*0/*F*-1) versus the concentration of BPA (ng mL−1) showed a good linear relationship. The linear relationship between fluorescence intensity and the BPA concentrations was in the range of 0 to 80 ng mL−<sup>1</sup> with a correlation coefficient of 0.9968 (n = 11). The corresponding limit of detection (LOD) following the IUPAC criteria (3σ/*S*) was calculated as 0.3626 ng mL−1. This value of LOD was lower than the permitted maximum residue limits of BPA in drinking water (10 ng mL−1, GB 5749-2006) [40], indicating that the as-prepared FeOx/ZnS@MIPs can be employed for environmental and drinking water safety monitoring. The comparison of this performance with other reported analytical methods for BPA detection is shown in Table 1. The results showed that the strategy presented is more rapid, selective and sensitive than the traditional methods. The proposed magnetic/fluorescence molecularly imprinted polymer coated FeOx/ZnS nanocomposites not only have the merits of convenience and low cost, but also have high selectivity and a comparable or lower limit of detection, which makes them a promising fluorescent sensor for the selective and sensitive detection of target molecules.

**Figure 6.** Fluorescence emission spectra of (**a**) FeOx/ZnS@MIPs and (**b**) FeOx/ZnS@NIPs with increasing concentrations of BPA. Inset graphs: the linear calibration of the fluorescence intensity change (*F*0/*F*) versus BPA concentration. Experimental conditions: BPA (0, 5, 10, 20, 30, 40, 50, 60, 70, 80 ng mL−1), citrate buffer solution (0.02 mol <sup>L</sup>−1, pH = 6.4), room temperature.


**Table 1.** The comparison of the results of different analytical techniques for the detection of BPA.

SERS: surface enhanced Raman scattering; GC-MS: gas chromatography-mass spectrometer; SPE: solid-phase extraction; Au NCs: gold nanoclusters; CDs: carbon dots; LOD: limit of detection.

## *3.6. Rebinding Selectivity*

The specific recognition ability of FeOx/ZnS@MIPs was further investigated. The imprinting factor (IF) and selectivity coefficient (SC) were used to evaluate the selectivity of the FeOx/ZnS@MIPs towards the template BPA and structural analogs (Figure S5). The imprinting factor is the ratio of the *K*MIP and *K*NIP (*K* is the slope of the linear equation), and the selectivity coefficient is the ratio of IF for the template molecule and structural analogs (Table S1).

As shown in Figure 7, the BPA molecule exhibited a significant fluorescence quenching effect on the FeOx/ZnS@MIPs. Table S1 shows that the imprinting factor for BPA was 11.19, which is much larger than that of BPZ, BP, and PTBP (1.47, 1.66, and 1.39, respectively). This was because the template BPA had more access to the recognition cavities of FeOx/ZnS@MIPs, which were formed in the imprinting process. Conversely, there are no complementary binding sites formed between FeOx/ZnS@NIPs and analyte molecules. The BPA and its analogs bound on the FeOx/ZnS@NIPs are mainly bound through non-specific interactions. Therefore, the *K*NIP for the BPA and structural analogs (BPZ, BP and PTBP) was almost the same, and a lower fluorescence quenching effect on the FeOx/ZnS@NIPs was observed (Figure 7). The experimental results showed that the selectivity of the MIPs for template BPA was much higher than that of structural analogs, indicating that a molecular imprinting process can greatly enhance the selectivity of FeOx/ZnS@NPs.

**Figure 7.** The quenching constant (*K*sv) and the imprinting factor (IF) of BPA, 1.1-Bis (4-hydroxyphenyl) cyclohexane (BPZ), 4,4-bisphenol (BP) and 4-tert-butylphenol (PTBP).

#### *3.7. Analysis of Real Samples*

In order to demonstrate the practical applicability of FeOx/ZnS@MIPs in real samples, different water samples (drinking water, tap water, and lake water) were used to evaluate the separation effectiveness and detection accuracy. The detection strategy is shown in Figure S1.

As shown in Table 2, the detected BPA values in the lake sample were 1.52, 7.39 and 6.54 ng mL−1, respectively. BPA was not found in drinking water and tap water at the detection limit level of 0.36 ng mL−1. The recovery rates of the present FeOx/ZnS@MIPs ranged from 90.8% to 103.1%, and the relative standard deviation (RSD) values were between 2.7% and 5.4% (n = 3). The above results proved that the prepared FeOx/ZnS@MIPs could be successfully applied to the magnetic separation and fluorescence detection of the target molecule in practical applications.


**Table 2.** Detection of BPA in real water samples using FeOx/ZnS@MIPs sensor.

aAverage value; RSD: relative standard deviation.

#### *3.8. Recyclability and Stability*

The recyclability test was done by observing the changes in fluorescence intensity with six rebind/elution cycles. As shown in Figure S6a, the fluorescence properties of the FeOx/ZnS@MIPs performance slightly decreased in five regeneration cycles. However, the fluorescence intensity decreased 25.5% and 31.1% in the next two regeneration cycles, respectively. This indicates that too much binding/removing template will a ffect the structure of the FeOx/ZnS@MIPs. The stability of FeOx/ZnS@MIPs was evaluated by measuring the initial fluorescence intensities and those after 1–35 days of storage under dark conditions (Figure S6b). It can be seen that the fluorescence intensity showed no significant change after being stored for a long time. These results sugges<sup>t</sup> that FeOx/ZnS@MIPs have good regeneration capacity and stability.
