Preparation of magnetic molecularly imprinted polymer for rapid ...

Anal Bioanal Chem (2009) 395:1125–1133 DOI 10.1007/s00216-009-3020-5

ORIGINAL PAPER

Preparation of magnetic molecularly imprinted polymer for rapid determination of bisphenol A in environmental water and milk samples Yongsheng Ji & Juanjuan Yin & Zhigang Xu & Chuande Zhao & Huayu Huang & Haixia Zhang & Chunming Wang

Received: 7 May 2009 / Revised: 16 July 2009 / Accepted: 29 July 2009 / Published online: 19 August 2009 # Springer-Verlag 2009

Abstract A magnetic molecularly imprinted polymer (M-MIP) of bisphenol A (BPA) was prepared by miniemulsion polymerization. The morphological and magnetic characteristics of the M-MIP were characterized by Fouriertransform infrared spectroscopy, transmission electron microscopy, and vibrating sample magnetometry. The adsorption capacities of the M-MIP and the nonimprinted polymer were investigated using static adsorption tests, and were found to be 390 and 270 mg g−1, respectively. Competitive recognition studies of the M-MIP were performed with BPA and the structurally similar compound DES, and the M-MIP displayed high selectivity for BPA. A method based on molecularly imprinted solid-phase extraction assisted by magnetic separation was developed to extract BPA from environmental water and milk samples. Various parameters such as the mass of sorbent, the pH of the sample, the extraction time, and desorption conditions were optimized. Under selected conditions, extraction was completed in 15 min. High-performance liquid chromatography with UV detection was employed to determine BPA after the extraction. For water samples, the developed method exhibited a limit of detection (LOD) of 14 ng L−1, a relative standard deviation of 2.7% (intraday), and spiked recoveries ranging from 89% to 106%. For milk samples, the LOD was 0.16µg L−1, recoveries ranged from 95% to 101%, and BPA was found in four samples at levels of Y. Ji : J. Yin : Z. Xu : C. Zhao : H. Huang : H. Zhang (*) : C. Wang State Key Laboratory of Applied Organic Chemistry, Lanzhou University, Lanzhou 730000, China e-mail: [email protected]

0.45–0.94µg L−1. The proposed method not only provides a rapid and reliable analysis but it also overcomes problems with conventional solid-phase extraction (SPE), such as the packing of the SPE column and the time-consuming nature of the process of loading large-volume samples. Keywords Magnetic molecularly imprinted polymer . Solidphase extraction . Magnetic separation . High-performance liquid chromatography . Bisphenol A

Introduction The molecular imprinting technique is the construction of ligand-selective recognition sites in synthetic polymers, in which a template is employed in order to facilitate recognition site formation during the covalent assembly of the bulk phase by a polymerization or polycondensation process. The subsequent removal of some or all of the template is necessary to permit recognition to occur in the spaces vacated by the templating species. This technique has attracted great interest because of its high selectivity (in terms of size, shape, and functionality) for target molecules [1]. It has been successfully applied in several fields, such as sensing [2], chromatography [3, 4], catalysis [5], and solid-phase extraction (SPE) [6–8]. Molecularly imprinted solid-phase extraction (MISPE) has also been extensively applied to environmental and biological samples [9–12]. Compared to natural receptors, molecularly imprinted polymers (MIP) not only demonstrate comparable molecular selectivities but they are also more robust and reusable, and less expensive to prepare [13]. However, some drawbacks to MISPE have restricted its widespread application.

1126

MISPE is usually used in cartridge mode, which often results in a tedious column packing procedure, high backpressure, and a low flow rate. The question is then: how can we overcome these disadvantages while still retaining the advantages of MISPE? Attempts have been made to develop an on-line MISPE system or to prepare a monolithic column coupled to a chromatographic system [3, 4, 11, 14]. Imparting magnetism to the MIP and then using magnetic separation is another promising alternative. Magnetic particles have been applied widely in biological fields, for example in bioseparation [15], drug delivery [16], and biomolecular sensing [17], and were recently used as SPE sorbents in environmental sample pretreatment [18– 20]. The preparation of a magnetic MIP (M-MIP) has been reported [13, 21–24]. Zhang et al. [23] and Chen et al. [24] applied M-MIPs to solid or semisolid samples to perform trace analyses of triazines and tetracycline antibiotics, respectively. The advantage of an M-MIP is obvious, as described in [23]: “the participation of a magnetic component in the imprinted polymer can build a controllable rebinding process and allow magnetic separation to replace the centrifugation and filtration step in a convenient and economical way.” When MIP particles contain magnetic components, the adsorption can be achieved by dispersing them in solution, and they are then easily separated from the matrix by applying an external magnet. Therefore, MISPE with magnetic separation allows convenient and highly efficient enrichment and avoids the need to pack the SPE column and the time-consuming process of loading a large-volume sample. However, a M-MIP is yet to be applied to a large-volume sample. BPA is a xenoestrogen that can disrupt endocrine function and adversely affect the reproductive systems of wildlife and humans. Its weak estrogenic activity has been confirmed in vitro and in vivo [25]. Various analytical methods of determining BPA have been developed [3, 18]. Most of them are based on the use of high-performance liquid chromatography (HPLC) along with sample pretreatment techniques, including MISPE [4, 14]. In the present study, M-MIP was synthesized using BPA as the template, vinylpyridine (2-VP) as the functional monomer, and ethylene glycol dimethacrylate (EGDMA) as the crosslinker by miniemulsion polymerization. It was then characterized, and the results indicated that the M-MIP microspheres had a uniformly spherical and porous structure, the desired magnetic susceptibility, and a high adsorption capacity and selectivity for BPA. A method based on MISPE with magnetic separation followed by HPLC was successfully developed in order to determine BPA in large-volume environmental water and milk samples. The proposed method was found to provide a rapid and reliable analysis of trace amounts of BPA.

Y. Ji et al.

Experimental Chemicals and materials All reagents were of analytical reagent grade. Ferrous chloride, ferric chloride, trichloracetic acid (TCA), and azobis(isobutyronitrile) (AIBN) were purchased from the Tianjin Chemicals Corporation (Tianjin, China). BPA, diethylstilbestrol (DES), hexadecanol, and sodium dodecyl sulfate (SDS) were supplied by the Shanghai Chemical Reagent Co. (Shanghai, China). 2-VP, EGDMA, and methacryloxypropyltrimethoxysilane (MAPS) were from Alfa Aesar (Ward Hill, MA, USA). Acetonitrile and methanol of HPLC grade were purchased from Dima Technology (Richmond Hill, VA, USA). Purified water obtained from a Milli-Q (Millipore, Billerica, MA, USA) system was used throughout the experiments. Stock solutions (1.0 mg mL−1) of BPA and DES in methanol were stored at 277 K. River water samples were collected from the Yellow River (Lanzhou, China), and wastewater samples were from the Lanzhou Petrochemical Company (which produces resin and plastic). Pure milk samples were purchased from a supermarket (Lanzhou, China). These samples were stored at 277 K. Chromatographic system and conditions The chromatographic system consisted of a Varian 210 high-performance liquid chromatographic pump (Palo Alto, CA, USA), a Varian 325 UV–vis detector, and a Varian Star chromatographic workstation. An analytical reversedphase C18 column (5 μm, 4.6×250 mm, Dima Technology was used. The mobile phase was a mixture of acetonitrile and water (60:40, V/V) with a flow rate of 0.8 mL min−1, and detection was carried out at 225 nm. The target compound was identified by the relative retention time and via diode array detection (model 2996, Waters, Milford, MA, USA). Procedures for the preparation of magnetic BPA-imprinted polymeric microspheres The preparation protocol is shown in Fig. 1. Fe3O4@MAPS nanoparticles were first prepared. Ferrous chloride (10.4 g) and ferric chloride (4.0 g) were dissolved in 100 mL of distilled water and heated to 363 K. Ammonium hydroxide (30 mL) and oleic acid (2 g) were rapidly added in sequence. After 3 h, the sediment was washed with ultrapure water until it was neutral. The magnetic nanoparticles were dried under vacuum at 353 K. Then they were modified with MAPS according to Wang et al. [21]. The obtained materials were dried under vacuum at 353 K.

Preparation of M-MIP for rapid determination of BPA

1127

Fig. 1 Preparation protocol for BPA-imprinted polymer

Then M-MIP microspheres were synthesized as follows [13, 26]. The template (BPA, 1 mmol) and the functional monomer (2-VP, 4.0 mmol) were mixed adequately in 3 mL isopropanol by stirring for 1 h. Then the crosslinking monomer (EGDMA, 20 mmol), hydrophobic agent (hexadecanol, 0.5 mL), and Fe3O4@MAPS (1 g) were added to the mixture. After ultrasonication for 5 min, this mixture was added to a 300 mL solution of 0.01 M SDS with vigorous stirring. The resulting miniemulsion was degassed for 10 min, and the initiator (AIBN, 0.5 mmol) was then added. After this, the polymerization was carried out at 343 K for 15 h with vigorous stirring. Upon completion, the polymeric microspheres were washed with water and ethanol, and then the template molecules were removed by Soxhlet extraction. Finally, the materials were dried for 12 h under vacuum at 333 K. Magnetic nonimprinted polymer (M-NMIP) was also prepared using the same procedure but without adding template. Characterization The obtained products were characterized with a JEM1200EX transmission electron microscope (TEM; JEOL, Tokyo, Japan) and a Nicolet Nexus 670 Fourier-

transform infrared (FT-IR) spectrometer (Ramsey, MN, USA). Magnetic properties were measured using a vibrating sample magnetometer (VSM; model 7304, Lakeshore, Westerville, OH, USA). Nitrogen adsorption–desorption experiments were carried out on an ASAP 2010 accelerated surface area and porosimetry system (Micromeritics, Norcross, GA, USA). The Brunauer–Emmett–Teller (BET) surface area (SBET) was calculated from the linearity of the BET equation. The surface area, volume, and pore diameter were calculated from pore size distribution curves using the Barrett–Joyner–Halenda formula. Adsorption study Static adsorption tests To measure the adsorption capacity, 20 mg M-MIP was added to 10 mL ethanol solution with various concentrations of BPA. The mixture was shaken for 12 h at room temperature to facilitate the adsorption of BPA onto the MMIP sorbent. After the M-MIP sorbent had been isolated using an external magnet, the supernatant was analyzed by HPLC. The same procedure was performed for the MNMIP. All tests were conducted in triplicate.

1128

Selectivity experiments Competitive recognition studies were performed with BPA and the structurally similar compound DES. The M-MIP microspheres were placed in a binary mixture of BPA and DES with initial individual concentrations of 500 mg L−1. The mixture was shaken for 12 h at room temperature and analyzed by HPLC in a similar manner to that described in the previous section. The M-NMIP was also examined with the same procedure. The data were obtained in triplicate. MISPE procedure for the extraction of BPA from water samples First, M-MIP (50 mg) was added to 200 mL of the water sample that had been adjusted to pH6.0. After ultasonicating for 2 min, the mixture was agitated for 5 min with a mechanic stirrer. Subsequently, the magnetic particles were isolated from the suspension with an adscititious magnet (Nd–Fe–B, 60×60×30 mm). The analytes were desorbed from the isolated particles with 3 mL methanol (containing 1% acetic acid, V/V) under sonication for 30 s. Finally, the eluate was evaporated at 323 K, and then the residue was dissolved in 0.5 mL of mobile phase. The solution (50 μL) was injected into the HPLC system for analysis. The cross-selectivity of M-MIP between BPA and DES was evaluated. The sample pretreatment procedure mentioned above was applied to an aqueous solution (200 mL) containing BPA and DES with individual concentrations of 10µg L−1. Determination of BPA in milk samples All milk samples were pretreated before analysis by the following procedure [27]. First, protein was removed from the matrix by adding TCA (2.5%, V/V) to a 50 mL milk sample. After being shaken for 30 s and centrifugated for 3 min, the supernatant was collected, and the precipitated protein was rinsed twice by 2 mL methanol. Then the supernatant and eluate was diluted to 500 mL, with the pH adjusted to ~6.0. Finally, the MISPE procedure mentioned above was carried out. Five samples of pure milk were determined.

Y. Ji et al.

were obtained (see Fig. 2). The characteristic band of Fe–O appeared at about 580 cm−1 in each spectrum. The C=C stretching vibration peak at 1,630 cm−1 and the Si–O–Si vibration peak at around 1,100 cm−1 were also observed, indicating that MAPS was indeed coated onto the surfaces of the Fe3O4 nanoparticles. No obvious differences were found between the spectra of M-MIP and M-NMIP. Intense peaks at 3,400, 1,720, and 1,460 cm−1 proved that C–H from pyridyl, C=O and C–N existed in the polymer. Upon comparing MMIP to Fe3O4@MAPS, the intensity of the C–H adsorption band at 2,930 cm−1 is clearly greater for M-MIP. Such results confirmed that the polymerization was successful. Morphological characteristics of the M-MIP and M-NMIP microspheres The morphological features of the M-MIP and M-NMIP were observed by TEM (see Fig. 3); all of the M-MIP (Fig. 3a) and M-NMIP (Fig. 3b) particles exhibited similar morphological features. Most of them were regular spheres with sizes of 500–1,000 nm, which is 50 times larger than the diameter of nanosized magnetite. These observations also confirmed that the magnetic polymeric microspheres had been successfully prepared. Nitrogen sorption measurements of M-MIP and MNMIP were carried out to further examine their morphological properties. The microspheres of M-MIP and M-NMIP did not display any significant differences in terms of their surface areas, pore volumes, and pore diameters. They all possessed a porous structure according to the results of nitrogen sorption measurements. The M-MIP presented a relative surface area of 105.0 m2 g−1, a total pore volume of 0.1538 cm3 g−1, and a mean pore diameter of 5.857 nm.

Results and discussion Characterization FT-IR spectra FT-IR spectra of Fe3O4, Fe3O4@MAPS, M-MIP, and MNMIP were obtained to verify that the expected products

Fig. 2 FT-IR spectra for Fe3O4, Fe3O4@MAPS, M-MIP, and M-NMIP


1129

Fig. 3a–b TEM images of a M-MIP and b M-NMIP

Magnetic properties Fig. 5 Testing the binding of BPA to the M-MIP and M-NMIP

It is vitally important for magnetic separation that the sorbent is sufficiently magnetic. VSM was employed to characterize the magnetic properties of the obtained magnetic materials, and their VSM magnetization curves are shown in Fig. 4. Based on the VSM data, the supporting core (Fe3O4) exhibits a saturation magnetization value of 60 emu g−1. M-MIP and M-NMIP exhibited enhanced magnetic properties after the formation of the polymeric coating, which decreased their saturation magnetization values (18 emu g−1 for MIP and 17 emu g−1 for NMIP). These magnetic properties enabled magnetic separation with a common magnet. Adsorption study Static adsorption tests The rebinding tests were performed by subjecting the MMIP or M-NMIP to various initial concentrations of BPA (from 0.2 to 2.0 mg mL−1). As shown in Fig. 5, the amount of BPA adsorbed increased with the initial concentration of

the BPA solution. The static adsorption capacities of the MMIP and M-NMIP were calculated based on the following formula: Q ¼ ðCi Cf Þ V =m Here, Q (mg g−1) is the mass of BPA adsorbed per gram of polymer, Ci (mg mL−1) is the initial concentration of BPA, Cf (mg mL−1) is its final concentration, V (mL) is the total volume of the adsorption mixture, and m is the mass of polymer in each rebinding mixture. The M-MIP exhibited significantly higher BPA loading than the M-NMIP, and the saturated adsorption capacities of the M-MIP and M-NMIP were 390 and 270 mg g−1, respectively. In general, the MMIP possesses both specific and nonspecific binding sites, while the M-NMIP only has nonspecific binding sites, which enables the M-MIP to take up more BPA than the MNMIP can. Note that the difference in adsorption capacity between the M-MIP and M-NMIP is equivalent to the contribution from the specific binding sites on the M-MIP, and this contribution increases with the concentration of BPA because more specific recognition sites are activated at higher BPA concentrations [28]. The imprinted sites are incorporated into the polymeric network, while physical adsorption primarily contributes to nonspecific adsorption on the surface of the sorbent. The template molecules are first adsorbed mainly to the nonspecific sites rather than the specific sites. After most of the nonspecific sites have been occupied, the specific sites began to get occupied. This is why the adsorption capacities of the M-MIP and M-NMIP are similar at low concentrations. The results obtained here are consistent with previous results [7, 8, 12, 29]. Selectivity experiments

Fig. 4 VSM magnetization curves for Fe3O4, Fe3O4@MAPS, MMIP, and M-NMIP

The structurally similar compound DES was selected to act as a competitor in competitive assays. The distribution

1130

Y. Ji et al.

coefficient (Kd), the selectivity coefficient (k), and the relative selectivity coefficient (k′) were obtained from these competitive experiments and are listed in Table 1. Kd indicates the adsorbtion ability of a substance, k suggests how selective the sorbent is when it is exposed to two substances, and k′ reveals how selective the sorbent is for a particular substance when compared with a different sorbent [29]. The k value of the M-MIP (1.8) is larger than that of the M-NMIP (0.98), indicating that the M-MIP has a higher selectivity for BPA over DES. The value of k′ is 1.8, confirming that the imprinted sorbent has a higher selectivity than the nonimprinted sorbent. M-MIP exhibits a higher affinity for BPA due to its template-specific sites, which can only be bound to by molecules with a specific structure [28]. Although the molecular structure of DES is very similar to that of BPA, and both have the same active group (–OH), they encountered very different levels of recognition by M-MIP. As expected, the template-specific cavities formed during M-MIP synthesis have multidimensional spatial structures that fit BPA molecules very well. Consistent results were obtained in the cross-selectivity test between BPA and DES. The recoveries of BPA and DES from aqueous solution were 100% and 53% after the MISPE procedure, respectively. Optimization of the MISPE procedure applied to water samples Desorption conditions To optimize the recovery of BPA, an elution solvent of methanol containing acetic acid (0–2.0%, V/V) was evaluated with the assistance of sonication. This elution solvent was tested for use under the same conditions: 50 mg of sorbent and 200 mL of sample (BPA 10 ng mL−1 spiked in purified water). As seen from Fig. 6a, the recovery increased dramatically when acetic acid was added to the elution solvent at concentrations from 0% to 1.0%. This was probably because the hydrogen bonds between BPA

and pyridyl were destroyed upon the addition of acetic acid. The recovery barely changed as the proportion of acetic acid was increased further, from 1.0% to 2.0%. Therefore, 1.0% acetic acid was used in subsequent experiments. To determine the minimum volume of elution solvent that can efficiently elute the adsorbed BPA, different volumes of elution solvent were investigated (from 1.0 to 6.0 mL), and 3.0 mL was found to be the optimum volume. Thus, 3.0 mL of methanol containing 1% acetic acid with sonication for 30 s was selected for the desorption stage. Mass of sorbent During the extraction procedure, the M-MIP sorbent was dispersed in a water sample to rebind analytes. The minimum amount of sorbent required to get efficient recovery was then investigated. Amounts of M-MIP sorbent ranging from 20 to 100 mg were applied to 200 mL samples. It was found that 50 mg of sorbent enabled almost complete recovery of the BPA, and increasing the amount of sorbent beyond this level did not produce any improvement in the recovery (Fig. 6b). Therefore, the amount of sorbent used was fixed at 50 mg. After each extraction, the sorbent was easily recovered by rinsing with methanol. The recycling of the sorbent was then studied, and the results showed that the sorbent can be used at least ten times with the same extraction efficiency. Effect of sample pH The effect of the sample pH was studied over the pHrange 2.0–11.5, and the results are shown in Fig. 6c. The best results were achieved in the pH range 3.5–6.0. A large drop in thee recovery was observed at pH values of 2.0 or 11.5. This is because the primary driving forces for the rebinding process, such as hydrophobic interactions, are strongly related to the sample pH. The state of the BPA in the sample was influenced by the pH of sample. Under mild conditions, most of the BPA was in a molecular state, enhancing their adsorption by the M-MIP sorbent. Hydrogen bonding also contributes to the molecular recognition

Table 1 Competitive loading of BPA and DES by M-MIP and M-NMIP Sorbents

MIP NMIP

Initial solution (mg L−1)

Final solution (mg L−1)

Kd

BPA

BPA

BPA

DES

BPA/DES

233 122

132 125

1.8 0.98

500 500

DES 500 500

341 402

DES 395 400

k

k′

1.8

Kd =(Ci −Cf)×volume of solution/(Cf ×mass of sorbent) (mL g−1 ), where Ci and Cf represent the initial and final concentrations, respectively; k= Kd(BPA)/Kd(DES); k′=kimprinted/knonimprinted


1131

Fig. 6a–d Effects of a the amount of HAc in the eluent, b the mass of sorbent, c the pH of the sample, and d the adsorption time on the extraction efficiency. Tests were carried out with 200 mL of an artificial water sample spiked with 10µg L−1 BPA

process, and such bonding is suppressed in strongly acidic or basic solutions due to the ionic state of BPA. Thereby, a sample pHof 6.0 was selected for subsequent experiments.

superior to conventional SPE [30], solid-phase microextraction [31], and stir bar sorptive extraction [32]. Analytical performance and the application of the MISPE to real samples

Extraction time The extraction procedure includes three steps: adsorption, isolation, and desorption. The total time required for extraction is a key factor in the efficiency of the assay. As described for the MISPE procedure, the interaction between the BPA and the sorbent is promoted by agitation. The effect of the adsorption time was studied by varying the stirring time (0–15 min). Figure 6d indicates that 5 min is sufficient to achieve complete recovery. After the adsorption stage, the M-MIP in suspension can be isolated in 5 min using an adscititious magnet. The whole extraction procedure can be accomplished within 15 min, which is

A method based on MISPE coupled to HPLC/UV was established. Under the optimal conditions described in the section “MISPE procedure for the extraction of BPA from water samples,” analytical parameters such as the linear range, the correlation coefficient, the limit of detection (LOD), and the precision were studied by extracting BPA from artificial water samples (Table 2). The linearity of the method was investigated over the range 0.1–40µg L−1 with squared coefficients of correlation (R2, 0.9990). The precision, expressed as the relative standard deviation (RSD), was assessed for five replicates with the same sorbent (intraday 2.7%, interday 5.8%). The LOD for BPA

1132

Y. Ji et al.

Table 2 Analytical parameters for the proposed method of determining BPA in water and milk samples, as well as for methods reported in previous works Sample Water Milk

Linear range (µg L−1)

Linearity (R2)

LODa (ng L−1)

RSDb (%)

RSDc (%)

0.1–40 1.0–200

0.9990 0.9973

14.0 160

2.7 6.1

5.8 10.5

LODs obtained in previous worksd (ng L−1) 60.0 [14], 6.85 [4] 150 [27], 110 [33]

a

LOD for the present method, based on the signal being three times as large as the baseline noise (S/N=3)

b

Intraday and n=5

c

Interday and n=5. The data were obtained with a standard solution (10µg L−1 , water) and spiked samples (80µg L−1 , milk)

d

LODs of methods reported in the references listed in the column

was found to be 14 ng L−1, which was compared with other reported results (Table 2). It can been seen that the proposed method has a lower LOD than the on-line MISPE/LC/UV method [14], but better results were obtained using HPLC/UV with MIP packing in the analytical column [4]. The developed method was applied to determine BPA in Yellow River water and wastewater samples. 1.0µg L−1 BPA was found in wastewater, but it was not found in Yellow River water. Recovery tests were performed at three spiked levels (0.2, 2.0, and 10µg L−1) to evaluate the accuracy of the method. The recoveries of BPA from river water and wastewater were in the range 89–106% with RSDs of 2.6–4.7%, indicating that the developed method is reliable for determining BPA in environmental water samples. It was further applied to analyze BPA in pure milk samples. Matrix effects, including protein precipitation and dilution, were eliminated before performing the MISPE procedure. The protein in milk was precipitated by TCA,

Fig. 7 MISPE/HPLC chromatograms for (a) a real milk sample (BPA 0.54µg L−1) and (b) a spiked milk sample (BPA 80µg L−1). Chromatographic conditions: C18 column (5 μm, 4.6×250 mm), mixture of acetonitrile and water (60:40, V/V) with a flow rate of 0.8 mL min−1, and detection at 225 nm. MISPE was performed under the same conditions: 200 mL of sample at pH6.0, 50 mg of sorbent, and 3 mL methanol containing 1% acetic acid as eluent

and the precipitated protein was rinsed with methanol to avoid the adsorption of BPA onto it. The supernatant and the eluate were pooled and further diluted to reduce matrix effects. A calibration curve was obtained for spiked milk samples (no BPA was detected in these samples), ranging from 0.7 to 200µg L−1, and with an R2 of 0.9973 (Table 2). The precision (RSD) was 6.1% (intraday, n=5), and the LOD was 0.16µg L−1 (about 0.14×10−3 mg kg−1, estimated relative to the density of milk, S/N =3). The spiked recoveries at three concentration levels (1.0, 10, and 80 µg L−1) were 101% (RSD 7.5%), 93% (RSD 6.7%), and 95% (RSD 6.5%), respectively. The method exhibited equivalent recoveries and a slightly higher LOD than a method based on solid-phase microextraction coupled to HPLC and fluorescence detection [27], and a method involving HPLC with fluorescence detection [33] (Table 2). Five pure milk samples purchased from the market had their BPA contents determined using the present method. Chromatograms of the real and spiked milk samples are shown in Fig. 7. BPA was found in four samples at levels of 0.45–0.94µg L−1 (about 0.41–0.85×10−3 mg kg−1, estimated relative to the density of milk). The milk was probably

Fig. 8 MISPE/HPLC chromatograms for milk samples spiked with BPA (100µg L−1) that was extracted using M-MIP and M-NMIP sorbents. The test conditions were the same as in Fig. 7


contaminated by its plastic packaging, or the BPA may have found its way into the milk in a variety of ways. However, the BPA concentrations in all of the milk samples studied were lower than the legislated limit placed on BPA content by the European Union (0.6 mg kg−1) [34]. The effect of using the M-MIP sorbent to extract the BPA from the sample was evaluated by performing comparative tests that used M-MIP and M-NMIP as sorbents to extract BPA from the same spiked milk sample. The utilization of the M-MIP resulted in much better extraction, as shown in Fig. 8.

Conclusions In this work, magnetic M-MIP microspheres were synthesized by miniemulsion polymerization. These M-MIP microspheres were regular spheres with a porous structure and the desired level of magnetic susceptibility. A high BPA adsorption capacity and good selectivity for BPA were demonstrated by the M-MIP in static adsorption and competitive tests. A simple MISPE method with magnetic separation was established. This MISPE not only provided a convenient, economical, and highly efficient extraction, but it also overcame problems with conventional SPE, such as the packing of the SPE column and the time-consuming process of loading large-volume samples. The sorbent can also be recycled easily by rinsing with methanol. A method combining MISPE with HPLC/UV was successfully used to determine the BPA contents of large-volume environmental water and milk samples. The proposed method could thus be a promising alternative for assaying complex large-volume samples. Acknowledgments This work was supported by the National Natural Science Foundation of China Fund (no. 20775029), the Program for New Century Excellent Talents in University (NCET-070400), and the Central Teacher Plan of Lanzhou University.

References 1. Caro E, Marcé RM, Borrull F, Cormack PAG, Sherrington DC (2006) Trends Anal Chem 25:143–154 2. Haupt K, Mosbach K (2000) Chem Rev 100:2495–2504 3. Watabe Y, Hosoya K, Tanaka N, Kubo T, Kondo T, Morita M (2005) J Chromatogr A 1073:363–370

1133 4. Jiang M, Zhang JH, Mei SR, Shi Y, Zou LJ, Zhu YX, Dai K, Lu B (2006) J Chromatogr A 1110:27–34 5. Wulff G (2002) Chem Rev 102:1–27 6. Tamayo FG, Turiel E, Martín-Esteban A (2007) J Chromatogr A 1152:32–40 7. Jiang XM, Tian W, Zhao CD, Zhang HX, Liu MC (2007) Talanta 72:119–125 8. Guo LY, Guan M, Zhao CD, Zhang HX (2008) Anal Bioanal Chem 392:1431–1438 9. Bravo JC, Garcinuño RM, Fernández P, Durand JS (2009) Anal Bioanal Chem 393:1763–1768 10. Schirmer C, Meisel H (2008) Anal Bioanal Chem 392:223–229 11. He JX, Wang S, Fang GZ, Zhu HP, Zhang Y (2008) J Agric Food Chem 56:2919–2925 12. Yan HY, Qiao FX, Row KH (2007) Anal Chem 79:8242–8248 13. Tan CJ, Chua HG, Ker KH, Tong YW (2008) Anal Chem 80:683– 692 14. Ou JJ, Hu LH, Hu LG, Li X, Zou HF (2006) Talanta 69:1001– 1006 15. Zou ZQ, Ibisate M, Zhou Y, Aebersold R, Xia YN, Zhang H (2008) Anal Chem 80:1228–1234 16. Lv G, He F, Wang XM, Gao F, Zhang G, Wang T, Jiang H, Wu CH, Guo DD, Li XM, Chen BA, Gu ZZ (2008) Langmuir 24:2151–2156 17. Zhang LH, Liu BF, Dong SJ (2007) J Phys Chem B 111:10448– 10452 18. Zhao XL, Shi YL, Wang T, Cai YQ, Jiang GB (2008) J Chromatogr A 1188:140–147 19. Li JD, Zhao XL, Shi YL, Cai YQ, Jiang GB (2008) J Chromatogr A 1180:24–31 20. Zhao XL, Shi YL, Cai YQ, Mou SF (2008) Environ Sci Technol 42:1201–1206 21. Wang XB, Ding XB, Zheng ZH, Hu XH, Cheng X, Peng YX (2006) Macromol Rapid Commun 27:1180–1184 22. Lu SL, Cheng GX, Zhang HG, Pang XS (2006) J Appl Polym Sci 99:3241–3250 23. Zhang Y, Liu RJ, Hu YL, Li GK (2009) Anal Chem 81:967–976 24. Chen LG, Liu J, Zeng QL, Wang H, Yu AM, Zhang HQ, Ding L (2009) J Chromatogr A 1216:3710–3719 25. Liu XY, Zhang XY, Zhang HX, Liu MC (2008) J Chromatogr Sci 46:596–600 26. Priego-Capote F, Ye L, Shakil S, Shamsi SA, Nilsson S (2008) Anal Chem 80:2881–2887 27. Liu XY, Ji YS, Zhang HX, Liu MC (2008) Food Addit Contam 25:772–778 28. Yang KG, Liu ZB, Mao M, Zhang XH, Zhao CS, Nishi N (2005) Anal Chim Acta 546:30–36 29. Han DM, Fang GZ, Yan XP (2005) J Chromatogr A 1100:131– 136 30. Zhou QX, Xiao JP, Wang WD, Liu GG, Wang QZ (2006) Talanta 68:1309–1315 31. Ouyang GF, Pawliszyn J (2006) Trends Anal Chem 25:692–703 32. David F, Sandra P (2007) J Chromatogr A 1152:54–69 33. Sun Y, Irie M, Kishikawa N, Wada M, Kuroda N, Nakashima K (2004) Biomed Chromatogr 18:501–507 34. Brede C, Fjeldal P, Skjevrak I, Herikstad H (2003) Food Addit Contam 20:684–689