All chemicals were of analytical grade reagents. Tetraethylorthosilicate (TEOS) was purchased from Gracia Chemical Technology Co., Ltd. Triethylamine (TEA), 3-methacryloxypropyltrimethoxysilane (KH-570), 3-isocyanatopropyltriethoxy-silane (IPTES) and acrylic acid (MAA) were supplied from Shanghai Aladdin Industrial Corporation. 17β-Estradiol (E2) was purchased from Adamas Reagent, Ltd. Ethylene glycol dimethyl acrylate (EGDMA) and 2,2'-Azobis (2 methylpropionitrile) (AIBN) were obtained from Guangdong J&K Scientific Ltd. Acetonitrile (ACN), tetrahydrofuran (THF), toluene were purchased from Sichuan Xilong Chemical Co., Ltd. Ammonia solution (25% - 28%), ethanol, methanol and acetone were obtained from Sichuan Kelong Chemical Co., Ltd. Ultrapure water (UPW) was used throughout the experimental procedures. Co., Ltd.

SiO₂ NPs were prepared by the Stӧber method [24] . Ethanol (44 ml), water (11 mL), and ammonia solution (1.1 mL) were mixed in a flask and vigorously stirred at 40˚C. Then a mixing solution of TEOS (4.4 mL) and ethanol (20 mL) was quickly added to the above solution and the mixture was stirred for 15 h. The SiO₂ NPs were centrifuged and rinsed with acetone. Finally, the obtained SiO₂ NPs were dried at 120˚C for 24 h.

First, the template-functional monomer complex (EstSi) was prepared: E2 (1 mmol), IPTES (1.3 mmol), and TEA (8 mmol) were dissolved in THF. The mixture was stirred under N₂ at 65˚C for 24 h. The reaction solution was concentrated under reduced pressure, and the residue was purified by precipitation in n-hexane and filtration [25] . EstSi was collected as a white solid (70 mg).

SiO₂-(E2+KH-570) were prepared by covalently attachment of KH-570 and EstSi to the surface of SiO₂ nanoparticle. Firstly, the SiO₂ NPs were dispersed in dry toluene (75 mL per gram of particles) under inert conditions. Then, EstSi (0.5 mmol per gram of particles) and KH-570 (1 mmol per gram of particles) were added to the suspension under constant stirring. The mixture was refluxed at 90˚C for 24 h. The resultant nanoparticles were centrifuged, filtered and washed several times with acetone. The solids were vacuum dried at 60˚C for 24 h to obtain 0.95 g of SiO₂-(E2+KH-570). The SiO₂-(IPTES+KH-570) were prepared by the same conditions, except that IPTES (0.5 mmol per gram of particles) was used instead of EstSi (0.5 mmol per gram of particles).

The above prepared SiO₂-(E2+KH-570) (150 mg) were dispersed into 35 mL of THF under ultrasonication. MAA (0.1 mmol), EGDMA and AIBN (15 mg) were subsequently dissolved into this solution. The mixing solution was purged with nitrogen for 10 min. Polymerization was first done at 50˚C for 5 h, and subsequently at 60˚C for 24 h. The mixing solution was stirred at a rate of 300 rpm/min throughout the experiment. Then particles were collected by filtration and washed with acetone to remove the unreacted monomers and cross-linking agent, and dried under vacuum at The above prepared SiO₂-(E2+KH-570) (150 mg) were dispersed into 35 mL of THF under ultrasonication. MAA (0.1 mmol), EGDMA and AIBN (15 mg) were dissolved into this solution. The mixing solution was purged with nitrogen for 10 min. Polymerization was first done at 50˚C for 5 h, and subsequently at 60˚C for 24 h. The mixing solution was stirred at a rate of 300 rpm/min throughout the experiment. Then particles were collected by filtration and washed with acetone to remove the unreacted monomers and cross-linking agent, and dried under vacuum at SiO₂@MIP1-E2, SiO₂@MIP2-E2, SiO₂@MIP3-E2, respectively.

In order to remove the E2 templates in the imprinted shells, SiO₂@MIP-E2 (220 mg) was refluxed in the solution of DMSO and water (5:1, v/v) at 180˚C for 3 h, and then the particles were obtained by Soxhlet extraction with the mixed methanol and acetic acid (8:2, v/v) solution at room temperature for 24 h. The resultant nanoparticles were centrifuged, filtered and washed several times with methanol. The solids were vacuum dried at 60˚C for 24 h. The samples obtained with different SiO₂@MIP-E2 (SiO₂@MIP1-E2, SiO₂@MIP2-E2, SiO₂@MIP3-E2) were denoted as SiO₂@MIP1, SiO₂@MIP2, SiO₂@MIP3 respectively. Nonimprinted nanoparticles were prepared via the same procedure of SiO₂@MIP2, but the SiO₂-(IPTES+KH-570) (150 mg) was used in place of SiO₂-(E2+KH-570), and assigned as SiO₂@NIP1. Moreover, as a control, SiO₂@MIP4 was also prepared in the same way as SiO₂@MIP2 without adding MAA.

The equilibrium adsorption experiments were performed to evaluate the binding capability of four kinds of SiO₂@MIP (SiO₂@MIP1, SiO₂@MIP2, SiO₂@MIP3, SiO₂@MIP4) and SiO₂@NIP1. Standard E2 solutions were prepared in acetonitrile and the E2 concentrations in the solutions were varied from 2 mg/L to 40 mg/L (2, 5, 10, 15, 20, 30, 40 mg/L). Then the 10 mg of SiO₂@MIP or SiO₂@NIP1 were dispersed into 10 mL of standard E2 solution. The obtained mixture was incubated at room temperature for 24 h, the suspension was filtered by the 0.1 µm micro-filter, and the filtered solution was analyzed by HPLC to obtain the E2 concentration.

In order to investigate the kinetics property of SiO₂@MIP2, the adsorption kinetic experiments were performed as follows. E2 solutions were prepared in acetonitrile and the E2 concentration in the solution was 37 µmol/L. SiO₂@MIP2 (50 mg) or SiO₂@NIP1 (50 mg) were added into 10 mL of E2 solution. The obtained mixtures were conducted at room temperature for different contact time (5, 10, 15, 20, 30, 40, 60 min), respectively. The suspension was quickly filtered by the 0.1 µm micro-filter, and the filtrates were analyzed by HPLC to obtain the E2 concentration.

Selectivity experiments were conducted by using E2, Lupeol (LU), 3’,4’,5'-trimethoxyflavanone (TF) and 2-Hydroxynaphthalene (Np) as the structural analogues of E2. The 10 mg of SiO₂@MIP2, SiO₂@MIP4 and SiO₂@NIP1 were dispersed into 2 ml of acetonitrile stock solution, which contains E2, LU, TF and Np at a concentration of 37 μmol/L, respectively. After incubating for 24 h at room temperature, the suspension was filtered by the 0.1 µm micro-filter, and the filtered solution was analyzed by HPLC to obtain the E2, LU, TF, Np concentration, respectively.

The reusability of SiO₂@MIP2 was measured as below: standard E2 solution was prepared in acetonitrile and the E2 concentrations in the solution were 10 mg/L. The SiO₂@MIP2 (100 mg) was dispersed into 20 mL of standard E2 solution. The mixture was incubated at room temperature for 24 h, the suspension was filtered by the 0.1 µm micro-filter. The filtrate was analyzed by HPLC. Then, the E2 of the recovered nanoparticles was removed with 10 mL of elution of methanol-acetic acid (8:2, v/v) for 24 hours to ensure complete removal of the E2 in the nanoparticles and washed with methanol, then vacuum dried at 60˚C for 24 h and reused for the adsorption of E2. The adsorption regeneration cycle was repeated five times with the same SiO₂@MIP2.

Fourier transform infrared (FTIR) spectra were recorded with Perkin-Elmer system 2000 FT-IR spectrometer. ¹H and ¹³C NMR spectra of EstSi were investigated by Bruker Ascend 400 MHz. Transmission electron microscopy (TEM) images were obtained on a Tecnai-G2-F20 electron microscope operating at 200 kV. Morphology of the materials were determined by scanning electron microscopy (SEM) with a JEOL 6300-F microscope.

The Raman spectra were collected on a HORIBA Lab Raman spectrometer (France) with an excitation wavelength of 532 nm, a resolution of 0.65 cm⁻¹ and a beam diameter of 20 μm. Before Raman measurements, the SiO₂@MIP2 (10 mg) or SiO₂@NIP1 (10 mg) was incubated in the 2 ml of E2 acetonitrile solution (the E2 concentrations in the solution was 40 mg/L) for 24 h, then isolated by filtration, washed 3 times with acetonitrile and vacuum dried. The samples were denoted as SiO₂@MIP2 rebinding E2 and SiO₂@NIP1 rebinding E2, respectively. The Raman spectroscopy were recorded by focusing the 532 nm diode laser on the silica nanoparticle materials with a total exposure time of 300 s per spectrum, all spectrograms of samples were smoothed and the baselines were corrected.

A Shimazu Prominence LC-20 A HPLC (Kyoto, Japan) with an Acclaim 120-C18 column (Dionex, USA, 5 μm, 120 Å, 4.6 × 20 mm²). The UV detection wavelength was 208 nm, the flow rate was 1.0 mL∙min⁻¹ and the injection volume was 25 µL. ACN/H₂O (45/55, v/v) was used as the mobile phase.

The covalently modified estradiol-functional monomer (EstSi) was synthesized according to the Yang et al.’s method (Figure 1) [25] . SiO₂ nanoparticles were synthesized according to reference [24] . The covalently modified templates (EstSi) along with KH-570 were grafted onto the surface of SiO₂ nanoparticles and followed by polymerization to result in a MIPs layer on the surface of SiO₂ nanoparticles [26] . The thickness of the MIPs layer was controlled by the added cross linker and the number of imprinted sites was controlled by EstSi used. The quality of the imprinted sites was controlled by both the covalently fixed template and coupling reagent (KH-570). The ratio of EstSi/KH-570 was optimized as 1:2. Finally, the template was removed by heating the SiO₂@MIP-E2 in DMSO at 180˚C, followed by Sohlex extraction with a mixed solvent of methanol and acetic acid [25] [26] . The preparation process was illustrated in Figure 2.

The size and morphology of SiO₂ nanoparticles were characterization by SEM (Figure 3). The as-synthesized SiO₂ nanoparticles were uniform in both size and

shape, with a mean diameter of about 100 nm and relatively mono-disperse. They were used for further grafting of EstSi and KH-570.

Fourier Transform Infrared (FTIR) spectroscopy was used to identify functional groups of KH-570 grafted on SiO₂-(E2+KH-570) and polymerized core-shell SiO₂@MIP (Figure 4). FT-IR spectra of SiO₂-(E2+KH-570) shows the characteristic band of C=O stretching vibration in the 1700 cm⁻¹ region, which indicates the presence of the coupling reagent (KH-570) [27] . After polymerization with EGDMA and MAA, a new peak at 1733 cm⁻¹ appears, which could be assigned to the C=O stretching (Figure 4, SiO₂@MIP2), which verified the successful introduction of the functional groups onto the surfaces of the modified SiO₂ nanoparticles by copolymerization [26] [27] .

TEM images were obtained for SiO₂, SiO₂@MIP1, SiO₂@MIP2 and SiO₂@MIP3 to identify the core-shell structure (Figure 5). TEM images show that the thickness of the polymer layer of SiO₂@MIP1 (Figure 5(b)), SiO₂@MIP2 (Figure 5(c)), SiO₂@MIP3 (Figure 5(d)) was about 1.0 ± 0.2 nm, 2.0 ± 0.8 nm, 5.0 ± 2.0 nm, respectively. These results indicate the thickness of the polymer shell depends on the amount of EGDMA used.

Raman spectra were used to confirm the incorporation of the template molecules (E2) in the SiO₂@MIP2-E2 and removal of E2 from SiO₂@MIP2. As shown in Figure 6, the SiO₂@MIP2-E2 exhibited strong characteristic Raman bands of SiO₂@MIP2-E2 around 829 cm⁻¹, 930 cm⁻¹, 1450 cm⁻¹ (Figure 6(b)), indicating the successful incorporation of E2 in SiO₂@MIP2-E2 (Figure 6(a) and Figure 6(b)). The disappearance of the typical signals of E2 at 829 cm⁻¹, 930 cm⁻¹, 1450 cm⁻¹ from Figure 6(c), implicating that E2 was indeed removed from the polymer shell of SiO₂@MIP2. The main bands of these Raman spectra and their assignments are shown in Table 1 [28] .

The adsorption capacity was calculated according to the following equation:

Q = (C₀ - C_f) V/M

ν―stretching vibration; δ―internal deformation vibration (bending); the schematic diagram of the E2 molecule in Figure 6.

where C₀ (mmol/L) is the initial con-centration of E2, C_f (mmol/L) is the concentration of E2 in the filtrate, V (mL) is the volume of the adsorption mixture, and M (mg) is the mass of SiO₂@MIP2.

The isotherm of imprinted materials with added acrylic acid with shell thickness of 1 nm, 2 nm and 5 nm (Figure 7(b) SiO₂@MIP1, Figure 7(c) SiO₂@MIP2 and Figure 7(d) SiO₂@MIP3), without acrylic acid of shell thickness of 2 nm (Figure 7(e) SiO₂@MIP4), and non-imprinted material (Figure 7(a) SiO₂@NIP) are presented in Figure 7, which clearly demonstrate that the material with added acrylic acid of imprinted shell thickness of 2 nm gave the highest adsorption of E2 (Figure 7(c)). The one without added acrylic acid (Figure 7(e)) otherwise identical with c showed a slightly lower adsorption than c. The shell thickness thicker or thinner than 2 nm showed poorer adsorption. The non-imprinted material had very low adsorption capacity for E2, due to non-specific adsorption. Thus, the material with added acrylic acid with thickness of 2 nm was used for further investigation in the adsorption kinetics and selectivity.

The core-shell imprinted material showed fast adsorption as demonstrated in Figure 8. The adsorption capacity of SiO₂@MIP2 increased quickly at the first 5 mins and reached equilibrium after 10 mins. For SiO₂@NIP1, the equilibrium time was 15 mins and the binding capacity was much lower than SiO₂@MIP2. This phenomenon can be explained by that comparing with the SiO₂@NIP1, SiO₂@MIP2 have the specific binding sites for E2. The rapidly rebinding kinetics suggested SiO₂@MIP2 has great potential in ultrafast enrichment and separation of E2 molecules.

The detection limit of the as-synthesized material was further tested with the Raman intensity changes after adding different concentrations of E2, as exhibited in Figure 9. The peaks at 829, 930 and 1450 cm⁻¹ were selected for identifying E2. As the E2 concentration increased from 0.01 μmol/L to 1 μmol/L , the peak signals intensity slightly increased, with the lowest detectable concentration of 0.01 μmol/L , which was sufficient to meet the requirements of detection limit of E2 requirements in the environmental water samples.

The imprinting effects were further evaluated by adsorption specificity experiment for the template and its structural analogues, LU, TF and Np whose chemical structural formulae are shown in Figure10(a). The results were summarized in Figure 10(b). The adsoption ability of the non-imprinted material SiO₂@NIP1 for E2 was much lower than the SiO₂@MIPs, and there was no significant difference in

binding capacity for the competitive compounds.

The reusability of the materials is an important factor to improve the economic efficiency and extend practical uses. The adsorption-desorption cycles were repeated five times by regenerating the used imprinted. The reusability was evaluated according to the following equation:

R = Q_n/Q₀ × 100%

where R is regeneration efficiencies, the Q_n (μg/g) is the adsorption capacity of

regenerative experiments, the “n” is repeated times, Q₀ (μg/g) is the adsorption capacity of original material.

As observed in Figure 11, calculations showed that of the prepared SiO₂@MIP2 after using for 5 times, the absorption capacity was reduced by about 10.6%. The SiO₂@MIP2 were repeatedly used and regenerated for 5 times with no significant loss of its original adsorb ability and selectivity. SiO₂@MIP2 as a recyclable economic material has a potential applying prospect in the extraction, separation and concentration and trace analysis of 17β-estradiol. It is because that the appropriate thickness and the high stability of the polymer layer, and monolayer recognition sites are distributed uniformly on the surface of silica nanoparticles all can enhance reusability.