1
J. Sep. Sci. 2014, 00, 1–7
Zerong Long1 Yi Lu1 Mingliang Zhang2 Hongdeng Qiu2 1 Xinjiang
Uygur Autonomous Region Product Quality Supervision and Inspection Institute, Urumqi, P. R. China 2 Key Laboratory of Chemistry of Northwestern Plant Resources and Key Laboratory for Natural Medicine of Gansu Province, Lanzhou Institute of Chemical Physics, Chinese Academy of Science, Lanzhou, P. R. China Received June 20, 2014 Revised July 16, 2014 Accepted July 16, 2014
Research Article
Selective recognition and discrimination of water-soluble azo dyes by a seven-channel molecularly imprinted polymer sensor array A seven-channel molecularly imprinted polymer sensor array was prepared and characterized by scanning electron microscopy, Fourier transform infrared spectroscopy, UV-Vis spectroscopy, and nitrogen physisorption studies. The results revealed that the imprinted polymers have distinct-binding affinities from those of structurally similar azo dyes. Analysis of the UV-Vis spectral response patterns of the seven dye analytes against the imprinted polymer array suggested that the different selectivity patterns of the array were closely connected to the imprinting process. To evaluate the effectiveness of the array format, the binding of a series of analytes was individually measured for each of the seven polymers, made with different templates (including one control polymer synthesized without the use of a template). The response patterns of the array to the selected azo dyes were processed by canonical discriminant analysis. The results showed that the molecularly imprinted array was able to discriminate each analyte with 100% accuracy. Moreover, the azo dyes in two real samples, spiked chrysoidin in smoked bean curd extract and Fanta lime soda (containing tartrazine), were successfully classified by the array. Keywords: Azo dyes / Canonical discriminant analysis / Molecularly imprinted polymers / Sensor arrays DOI 10.1002/jssc.201400684
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction A number of industrial dyes and their N-substituted aromatic biotransformation products are known to be toxic and/or carcinogenic. Thus, the use of these dyes in food products is considered harmful. At present, the number of harmful dyes detected in packaged food is upwards of 20 dye types, including Sudan I–V, para red, rhodamine B, orange II, acid red, Sudan Red, rhodamine B, Sudan Red 7B, metanil yellow, auramine, Congo Red, butter yellow, Solvent Red I, naphthol yellow, malachite green, leucomalachte green, Ponceau 3R, Ponceau MX, and Orange SS [1, 2]. It would be a demanding task to categorize the various types of dyes in packaged food by employing conventional methods [3–5]. Thus, it is highly
Correspondence: Prof. Hongdeng Qiu, Key Laboratory of Chemistry of Northwestern Plant Resources and Key Laboratory for Natural Medicine of Gansu Province, Lanzhou Institute of Chemical Physics, CAS, Lanzhou 730000, P. R. China E-mail: [email protected] Fax: +86-931-8277088
Abbreviations: AIBN, azobisisobutyronitrile; CDA, canonical discriminant analysis; EGDMA, ethylene glycol dimethacrylate; MAA, methacrylic acid; MIP, molecularly imprinted polymer C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
imperative to develop an effective, efficient, and suitable analysis method for identifying toxic dyes. Molecularly imprinted polymers (MIPs) are simple tailormade materials with highly specific recognition abilities for detecting target molecules. MIPs also possess excellent thermal, chemical, and mechanical stability [6–8]. Puoci et al. reported the synthesis of MIPs using Sudan I as a template, and extracted traces of Sudan I from spiked red chili powder by molecularly imprinted solid-phase extraction [9]. Yan and coworkers prepared Malachite Green-templated MIPs by a precipitation method, and found this method to preferentially bind the dye over other closely related compounds [10]. Recently, Piao et al. described the formation of a magnetic MIP column using Sudan IV, and the use of this polymer as a template to separate four Sudan dyes (Sudan I–IV) from food matrices [11]. The LOD was 6.2, 1.6, 4.3, and 4.5 ng/g for Sudan I–IV, respectively, using HPLC detection and molecularly imprinted solid-phase extraction. Multichannel MIP sensor arrays have been proven to be a highly practical and effective format with good discrimination and accuracy [12–14]. Their sensing strategy is similar to that of the nose or the tongue, which are able to differentiate and identify an almost unlimited number of fragrances and flavors by a finite number of different sensing cells. The sensor Colour Online: See the article online to view Figs. 5 and 6 in colour. www.jss-journal.com
2
J. Sep. Sci. 2014, 00, 1–7
Z. Long et al.
array format has proved to be useful for accurately differentiating multiple structurally similar compounds by transforming sensors of poor or modest selectivity into highly selective sensors. Yan et al. developed an imprinted mesoporous silica matrix sensor array that could differentiate not only individual carbohydrates but also different brands of orange juice [15]. Takeuchi et al. prepared an imprinted polymer sensor array using cytochrome C, ribonuclease A, and ␣-lactalbumin as a template, which could identify five proteins [16]. More recent studies have proven that the selectivity of an MIP sensor array can be tailored for a wide range of analytes and can be expressed in many different sensing formats. Thus, a sensor array could be created and categorized in terms of the use of different signaling mechanisms, including fluorescence, UV-Vis, electrical capacitance, and quartz crystal microbalance [17–21]. For example, Liu et al. reported the creation of a 2D imprinted polymer-coated mass-sensitive quartz crystal microbalance sensor array that can detect staphylococcus enterotoxin A and staphylococcus enterotoxin B [22]. Previous work using a colorimetric MIP array or a 2D sensor array orthogonal combination of four MIPs and three dyes have been successful for the discrimination of amines [13, 14]. However, very little attention has been paid to the discrimination of various structurally similar dyes by sensors. In this work, we developed a new seven-channel MIP sensor array for the accurate identification of water-soluble azo dyes. Good selectivity to recognize the azo dyes of the prepared sensor array was achieved.
2 Materials and methods 2.1 Reagents and apparatus All chemicals were purchased from commercial suppliers and used without further purification. All synthetic manipulations were carried out under a nitrogen atmosphere using standard Schlenk and cannula techniques. FTIR spectra were measured with a Vertex70 instrument (Bruker, Germany). The morphology of the MIPs was measured by a LEO1430VP scanning electron microscope (LEO, Germany). UV-Vis spectra were obtained using UV-2450 spectroscopy (Shimadzu, Japan). All nitrogen sorption measurements were performed on an Autosorb-iQ automated gas sorption analyzer (Quantachrome Instruments, USA) using 0.13–0.25 g of the polymer at 77 K. UHPLC analysis was performed on a Waters ACQUITY UPLC system equipped with a photodiode array detector.
2.2 Polymer preparation (P0–P5) Polymers P0–P5 were synthesized using previously reported methods [13]. Azobisisobutyronitrile (AIBN, 52.54 mg, 0.32 mmol), methacrylic acid (MAA, 0.27 mL, 3.20 mmol), ethylene glycol dimethacrylate (EGDMA, 2.50 mL, 12.80 mmol), and 0.33 mmol of template (B1–B5) were in C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
dividually mixed in a 3:1 v/v solution of ethanol/glycerol (40 mL) under N2 . The nonimprinted polymer P0 was prepared using the same mixture, but without the addition of any template molecule. Polymerization was carried out with a heat trigger and vigorously stirred in a 60⬚C constanttemperature water bath for 12 h. The polymers were subjected to Soxhlet extraction using an acetic acid/methanol solution (1:9 v/v) as an eluent for 24 h, and then thoroughly rinsed with acetonitrile to remove any residual acetic acid within the polymers. The polymers were dried under vacuum at 50⬚C for 24 h. 2.3 Preparation of polymer P6 Polymer P6 has been readily prepared using B6 as template by efficient synthetic routes similar to the polymers P0–P5. However, B6 is almost insoluble in any alcoholic solutions but can be easily dissolved in chloroform. So chloroform was used as the solvent for the synthesis of polymer P6. 2.4 Determination of the MIP sensor array (P0–P6) All analytes were used in their free base form. Polymers P0–P6 were individually placed into seven separate 20 mL comparison tubes. The polymers were equilibrated with 10 mL of deionized water solutions containing a single analyte (0.02 mM) for 4 h. The response of the array was measured by the removal of 2 mL aliquots of the supernatant and the subsequent analysis of this aliquot with UV-Vis spectroscopy. This process was repeated for analytes B1–B5 and D1 to generate five replicates of each analysis. Acetonitrile was used as the solvent for B6 because of this analyte’s insolubility in water. The seven analytes were tested against the seven polymers (P0–P6) five times to generate a 7 × 35 data matrix (7 sensors × 7 analytes × 5 replicates). This raw data matrix was processed using a canonical discriminant analysis (CDA) function as implemented in SPSS software. 2.5 Polymer surface area and porosity measurements Prior to nitrogen sorption measurements, the samples were degassed at 393 K for 3 h to remove adsorbed gases and moisture. Brunauer–Emmett–Teller (BET) surface areas for the polymers were calculated from the adsorption data with 0.162 nm2 as the molecular cross-sectional area for the adsorbed nitrogen molecules [23]. The Barrett–Joyner–Halenda (BJH) method was applied to calculate the pore-size distribution from the adsorption and desorption branches of the isotherms. 2.6 Sample preparation and UHPLC–UV analysis Equal volumes of 7.5 g/mL of each azo dye (B1–B5 and D1) were mixed at 4⬚C for 4 h. Then, these aqueous solutions were directly used for UHPLC–UV analysis. Five mg of P5 www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
Sample Preparation
3
3.1 Selectivity characteristics
Scheme 1. Structures of analytes B1–B6 and D1, corresponding to amaranth, tartrazine, sunset yellow, orange II, chrysoidin and para red, and rhodamine B, respectively.
was added to the mixture solution. After shaking for 4 h at room temperature, the samples were stored at 4⬚C, and then filtered through a 0.22 m membrane filter and analyzed by UHPLC–UV analysis. Polymer P5 was washed five times with 2 mL of the solution of methanol and acetic acid (9:1 v/v). The supernatant was obtained by centrifugation at 12 000 rpm for 15 min and was directly applied for UHPLC–UV analysis. UHPLC–UV analysis was performed at room temperature using an ACQUITY UPLC BEH shield RP18 (2.1 × 100 mm, 1.8 m) with acetonitrile (A) and a 10% ammonium acetate solution (B) as the mobile phase. The elution gradient used was A/B at 85:15 v/v for 0–3.5 min, A/B at 0:100 v/v for 3.5–6 min, and A/B at 85:15 v/v for 6–9 min. The flow rate was 0.2 mL/min and the injection volume was 5 L. A photodiode array with quantification at 440 nm was used for detection.
3 Results and discussion The MIPs (P1–P6) were prepared using a MAA/EGDMA matrix in the presence of six different template analytes, B1–B6, respectively. In addition, a blank, nonimprinted polymer (P0) was synthesized without using any template molecule. The polymers were synthesized under a nitrogen atmosphere in a 60⬚C water bath by using a 1:4 molar ratio of MAA/EGDMA with 10 mol% AIBN as the radical initiator. The prepared MIP sensor array was used to discriminate between the seven azo dyes (B1–B6 and D1, Scheme 1) by the analysis of their UVVis spectra. The response patterns were then processed using CDA analysis [24, 25]. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The affinity of each polymer for each of the seven analytes (B1–B6, D1) was tested by UV-Vis spectroscopy by vigorously shaking a constant weight of polymer with 0.02 mM solutions of each analyte, and then measuring the resulting UV-Vis spectra. As a result, the response was measured as absorbance (A) and the ratio of absorbance ((A0 –Ai )/A0 ) at 480 nm of each analyte solution before and after equilibration with the MIP array (Fig. 1). Thus, the fingerprint maps of each analyte against the MIP array are shown in the inset of Fig. 1. Each analyte was tested for five separate times against the MIP sensor array. Polymers P1–P6 made in the presence of template molecules appeared to be imprinted by the template. This was demonstrated by the lower A values or higher ((A0 –Ai )/A0 ) values for imprinted polymers P1 through P6 compared to those for nonimprinted polymer P0. In addition, the response of analyte B5 against the MIP sensor was the strongest at the same solution concentrations, because the usage amount of the individual polymers was only 1/4th of that of the other dyes. However, the response of each of the six dyes (B1–B6) against the MIP sensor array displayed a faint concentration gradient. This result suggested that the individual imprinted polymers showed poor overall selectivity, because they presented higher affinity not only to their template molecules but also to the other structurally similar analytes (especially P1, P3, and P6). The SEM images and the comparison of the considerable differences in the morphology of the chrysoidin (B5)imprinted polymer and its nonimprinted polymer are shown in Fig. 2. The SEM micrograph showed that the nonimprinted polymers had no cavity in contrast to MIPs and exhibited an irregular, amorphous morphology, as shown in the SEM image. The high-magnification SEM images of the B5-MIPS revealed the self-assembled structures of block copolymers toward the out-of-plane direction, fabricated by layer-by-layer accumulation of imprinted polymer particles with an average size of 200–500 nm. All the particles in the B5-MIPs had a sheet structure, with many large pores between the particles, possibly caused by the removal of the template molecule. MIPs with uniform and open structure are obviously favorable for the adsorption of the template molecules such as the used dye B5 [26]. Thus, this result suggested that the final polymer morphology could affect the selectivity of templaterebinding characteristics. Nitrogen-adsorption measurements at 77 K of the powdered samples of the polymers derived from P0–P6 confirmed their microporosity by demonstrating significant adsorption. The apparent BET surface areas and pore volumes of the polymers calculated by the BJH method are shown in Supporting Information Table S1. Polymer P5 was found to have greater microporosity, both in terms of the average pore diameter and the total pore volume, than nonimprinted polymer P0. This finding is consistent with previous observations that an imprinted polymer is more efficient at generating intrinsic microporosity than a nonimprinted polymer [23]. In addition, polymers P2 and P5 had www.jss-journal.com
4
Z. Long et al.
J. Sep. Sci. 2014, 00, 1–7
Figure 1. Absorbance change of seven analytes (B1–B6 and D1) against the MIP sensor array (P0–P6) after equilibration for 4 h: 10 mL of 0.02 mM at 480 nm (A) B1 aqueous solution; (B) B2 aqueous solution; (C) B3 aqueous solution; (D) B4 aqueous solution; (E) B5 aqueous solution; (F) B6 in acetonitrile; (G) D1 aqueous solution. Each polymer (P0–P6) was consumed individually: 20 mg for (A)–(C), (F) and (G); and 5 mg for (E). The inset pictures show special fingerprints of each of the seven analytes (B1–B6 and D1) against the MIP sensor array (average values of five separate replicates of each analyte).
Figure 2. (A) SEM image of nonimprinted polymers P0; (B) SEM image of P5. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
larger average pore diameters than the other imprinted polymers. To further determine the adsorption of B5 on the polymers (P0 and P5), an equimolar amount of analyte B5 was mixed with 5 mg of P0 and P5 in an aqueous solution. After 4 h of shaking, the samples (P0@B5 and P5@B5) were centrifuged and washed several times with 1 mL of deionized water, and then dried under vacuum at 50⬚C. The dried samples were finely crushed, mixed with potassium bromide (1:15 w/w), pressed at 15 000 psi, and analyzed using quantitative IR analysis. IR spectroscopy was also used to characterize the prepared materials (Fig. 3). In the IR spectrum of B5 (Fig. 3A), the bands at 3321 and 3135 cm−1 were assigned to the N– H stretching vibrations. The most intense band found at 1627 cm−1 was likely due to the stretching vibration of the –N=N– group. The N–H bending vibration of arylamine was at 1561 cm−1 , and the characteristic bands at 1512, 1446, and 1415 cm−1 were attributed to the aryl skeleton vibrations. As shown in Fig. 3B and D, the wide bands at 3565 and 3437 cm−1 were assigned to the stretching vibration of the hydroxyl group of P0 and P5, respectively. In comparison with P5, the band of the hydroxyl group in P0 shifted to a higher wavenumber, indicating the formation of intramolecular hydrogen bonds. However, the band of the hydroxyl group in P0 took a bathochromic shift after combination with B5, indicating the collapse of the intramolecular hydrogen bonds. A band also appeared at 1562 cm−1 , which was attributed to the bending vibration of arylamine in B5 (Fig. 3C). In the P5 and P5@B5 spectra, there appeared to be a hypochromatic shift (ca. 6 nm) corresponding to the stretching vibration of the hydroxyl group in P5 and the N–H bending vibration of arylamine in B5 (Fig. 3D and E). This result indicated the formation of intermolecular hydrogen bonds between B5 and P5. In addition, the band intensity of the N–H bending vibration of arylamine in Fig. 3E was stronger than that in Fig. 3C,
Figure 3. FTIR spectra of (A) B5; (B) P0; (C) P0@B5; (D) P5; and (E) P5@B5.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Sample Preparation
5
which showed that a greater amount of B5 was bound to P5 than P0 because of the imprinting process. These results also demonstrated that polymer P5 had a high selectivity for its template molecule, which is in accordance with that of the UV-Vis spectroscopy results. To further verify whether the MIPs synthesized were selective to their template molecules (azo dyes), the rebinding of those structurally related compounds (azo dyes) on polymer P5 was investigated as an example. Five milligrams of P5 was brought into contact with an aqueous solution of the dyes (C0 = 7.5 g/mL of each of dyes B1–B5 and D1, excluding water-insoluble B6). After the initial mixing, the suspension was allowed to rebind/absorb the dyes by shaking for 4 h at room temperature. The resulting suspension was stored at 4⬚C and filtered through a 0.22 m membrane filter, and then directly used for UHPLC–UV analysis. P5 was used for the selective extraction of B5 from the mixture as described in Fig. 4B. Because the peak area of each dye in UHPLC
Figure 4. UHPLC–UV chromatograms of a standard solution mixture (containing B1–B5 and D1) before and after adsorption by P5, measured at 440 nm: (A) the mixture standard solution before and (B) the supernatant after adsorption by P5, and (C) the collected eluent after P5 was separated from the eluent and purified.
www.jss-journal.com
6
J. Sep. Sci. 2014, 00, 1–7
Z. Long et al.
Figure 5. Plot of the response of the seven analytes (0.02 mM aqueous solutions for B1–B5 and D1, and 0.02 mM B6 solution in CH3 CN) tested in five times in separate replicates against the MIP sensor array at 480 nm.
chromatograms is directly proportional to its concentration within a given range [27], the concentrations of B1–B5 and D1 were calculated as 6.8, 7.2, 5.9, 7.0, 6.5, and 0.4 g/mL, respectively. Compared with the concentration of the other dyes, that of B5 was the lowest after being adsorbed by P5 (Fig. 4A and B). Polymer P5 was separated from the suspension and washed five times with 2 mL of methanol and acetic acid (9:1 v/v). Afterwards, any B5 released from P5 was collected and detected by UHPLC–UV analysis. As shown in Fig. 4C, the main species measured in this sample was B5. Small amounts of B4 and D1 were also measured in the eluent. These results demonstrated that polymer P5 was capable of high selectivity and specificity for its template molecules.
3.2 CDA analysis To investigate the recognition accuracy for individual dyes against the MIP sensor array, the analysis of five separate replicates of each analyte was carried out by UV-Vis spectroscopy under identical testing conditions. Seven different
values were measured for each analyte, shown as lines on the plot in Fig. 5. The responses of the individual analytes were found to correspond to each other for each of their five analysis repetitions. This showed excellent data reproducible for all analytes, likely due to the uniform particle size distribution by lower polymerization concentrations. Among the analytes, there was little overlap among B4, B5, and D1. However, there appeared to be much more overlap among B1, B2, B3, B4, and B6. Although there was a little amount of overlap between B1 and B6, their fingerprints were very close. As a result, only a limited portion of the fingerprint maps were distinguishable, because fingerprint recognition relies heavily on the quality of the input fingerprint images. To avoid this disadvantage of fingerprint resemblance and overlap, as well as to filter out random noise, multivariate analysis was applied to the dataset. The response patterns were processed using the CDA analysis, which is related to the principle component analysis and canonical correlation. The CDA is used as a means of distinguishing among a group of samples from potentially different populations. To evaluate the responses of the sensor elements, five replicate datasets were consolidated into a 7 × 35 data matrix data sheet (7 sensors × 7 analytes × 5 replicates). CDA analysis was conducted to reduce the dimensionality of the data. The analysis showed impressive levels of discrimination, as the response patterns for each dye were well differentiated in the CDA plot (Fig. 6). The replicate data points for each analysis were clustered together. These groupings were separated equally from one another. The classification accuracy of the original data was up to 100.0% for both the original grouped cases correctly classified and crossvalidated. Rhodamine B (D1) was also correctly classified, even though it was not one of the original template analytes (B1–B6). These results showed that, once an array has been prepared against one set of analytes, it could facilitate the classification of other analytes. This finding greatly enlarges the range of detectable analytes using an MIP array. 3.3 Real sample analysis Real samples were analyzed and identified according to their canonical scores or their placement in the 2D space. The
Figure 6. (A) The spectral response of azo dye against the MIP sensor array, taking B5 as example for one trail; (B) 2D CDA plot of the seven analytes against the MIP sensor array (five trails each) showing clear clustering of the trails.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
array response to each sample was compared to the classification data, and the sample dyes were identified according to the correspondence between their Mahalanobis distances and those of the classified azo dyes (Supporting Information). Spiked B5 in the smoked bean curd extract and Fanta lime soda (containing B2) against the array were identified with 100.0% accuracy for the original grouped cases; among the cross-validated grouped cases, 94.6% of the cases were correctly classified.
4 Conclusions We have demonstrated the utility of an MIP sensor array for the discrimination of various azo dyes. MIP arrays of different selectivity could be rapidly and inexpensively prepared and adjusted by alternating the initial templates used in the imprinting process. The response patterns of these dyes against the MIP sensor array were analyzed by the CDA with 100% classification accuracy. In addition, the array was found to be capable of classifying D1, even though no receptor was specifically designed for this analyte. Furthermore, the array’s identification ability was found to work successfully when used to analyze real samples. As a result, the MIP sensor arrays developed in this work could have a large number of broad practical applications, as they permit discrimination between species in the closely related azo-structural family. The authors are grateful for financial support from Xinjiang provincial natural science fund project (Grant No. 2012211A107). H. Qiu acknowledges the support of the “Hundred Talents Program” of Chinese Academy of Sciences. The authors have declared no conflict of interest.
5 References [1] Lok, C. M., Son, R., Food Res. Int. 2009, 16, 127–140. [2] Oplatowska, M., Donnelly, R. F., Majithiya, R. J., Glenn Kennedy, D., Elliott, C. T., Food Chem. Toxicol. 2011, 49, 1870–1876. [3] Zou, T., He, P., Yasen, A., Li, Z., Food Chem. 2013, 138, 1742–1748. [4] Taverna, D., Di Donna, L., Mazzotti, F., Policicchio, B., Sindona, G., J. Mass Spectrom. 2013, 48, 544–547. [5] Fang, G., Wu, Y., Dong, X., J. Agric. Food Chem. 2013, 61, 3834–3841. [6] Svenson, J., Nicholls, I. A., Anal. Chim. Acta. 2001, 435, 19–24.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Sample Preparation
7
[7] Bossi, A. M., Sharma, P. S., Montana, L., Zoccatelli, G., Laub, O., Levi, R., Anal. Chem. 2012, 84, 4036– 4041. [8] Cheong, W. J., Yang, S. H., Ali, F., J. Sep. Sci. 2013, 36, 609–628. [9] Puoci, F., Garreffa, C., Iemma, F., Muzzalupo, R., Spizzirri, U. G., Picci, N., Food Chem. 2005, 93, 349–353. [10] Yan, S. L., Gao, Z. X., Fang, Y. J., Cheng, Y. Y., Zhou, H. Y., Wang, H. Y., Dyes Pigments 2007, 74, 572–577. [11] Piao, C., Chen, L., J. Chromatogr. A 2012, 1268, 185–190. [12] Greene, N. T., Morgan, S. L., Shimizu, K. D., Chem. Commun. 2004, 10, 1172–1173. [13] Greene, N. T., Shimizu, K. D., J. Am. Chem. Soc. 2005, 127, 5695–5700. [14] Lee, J.-D., Hong, J. I., Tetrahedron Lett. 2013, 54, 2890– 2893. [15] Tan, J., Wang, H. F., Yan, X. P., Anal. Chem. 2009, 81, 5273–5280. [16] Takeuchi, T., Goto, D., Shinmori, H. Analyst 2007, 132, 101–103. [17] Shimizu, K. D., Stephenson, C. J., Curr. Opin. Chem. Biol. 14, 743–750. [18] Qiu, H., Luo, C., Sun, M., Lu, F., Fan, L., Li, X., Anal. Chim. Acta. 2012, 744, 75–81. [19] Croux, D., Weustenraed, A., Pobedinskas, P., Horemans, ¨ H., Haenen, K., Cleij, T., Wagner, P., Thoelen F., Dilien, R., De Ceuninck, W., Phys. Status Solidi. A 2012, 209, 892–899. [20] Zang, D., Yan, M., Ge, S., Ge, L., Yu, J., Analyst 2013, 138, 2704–2711. [21] Alizodeh, T., Rezaloo, F., Sensor. Actuat. B Chem. 2013, 176, 28–37. [22] Liu, N., Gao, Z., Ma, X., Ou, G., Li, X., Chen, X., CN 201110459803 A, 2012. [23] Rosengren, A. M., Karlsson, B. C. G., Nicholls, I. A., Int. J. Mol. Sci. 2013, 14, 1207–1217. ´ [24] de Mendoza, M. F., Gordillo, C. D. M., Expoxito, J. M., Casas, J. S., Cano, M. M., Vertedor, D. M., Baltasar, M. N. F., Food Chem. 2013, 141, 2575–2581. [25] Dimauro, C., Cellesi, M., Steri, R., Gaspa, G., Sorbolini, S., Stella, A., Macciotta, N. P. P., Anim. Genet. 2013, 44, 377–382. [26] Yoshimatsu, K., Reimhult, K., Krozer, A., Mosbach, K., Sode, K., Ye, L., Anal. Chim. Acta. 2007, 584, 112–121. [27] You, Y., Uboh, C. E., Soma, L. R., Guan, F., Li, X., Rudy, J. A., Liu, Y., Chen, J., Rapid Commun. Mass Sp. 2009, 23, 2035–2044.
www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
Zerong Long1 Yi Lu1 Mingliang Zhang2 Hongdeng Qiu2 1 Xinjiang
Uygur Autonomous Region Product Quality Supervision and Inspection Institute, Urumqi, P. R. China 2 Key Laboratory of Chemistry of Northwestern Plant Resources and Key Laboratory for Natural Medicine of Gansu Province, Lanzhou Institute of Chemical Physics, Chinese Academy of Science, Lanzhou, P. R. China Received June 20, 2014 Revised July 16, 2014 Accepted July 16, 2014
Research Article
Selective recognition and discrimination of water-soluble azo dyes by a seven-channel molecularly imprinted polymer sensor array A seven-channel molecularly imprinted polymer sensor array was prepared and characterized by scanning electron microscopy, Fourier transform infrared spectroscopy, UV-Vis spectroscopy, and nitrogen physisorption studies. The results revealed that the imprinted polymers have distinct-binding affinities from those of structurally similar azo dyes. Analysis of the UV-Vis spectral response patterns of the seven dye analytes against the imprinted polymer array suggested that the different selectivity patterns of the array were closely connected to the imprinting process. To evaluate the effectiveness of the array format, the binding of a series of analytes was individually measured for each of the seven polymers, made with different templates (including one control polymer synthesized without the use of a template). The response patterns of the array to the selected azo dyes were processed by canonical discriminant analysis. The results showed that the molecularly imprinted array was able to discriminate each analyte with 100% accuracy. Moreover, the azo dyes in two real samples, spiked chrysoidin in smoked bean curd extract and Fanta lime soda (containing tartrazine), were successfully classified by the array. Keywords: Azo dyes / Canonical discriminant analysis / Molecularly imprinted polymers / Sensor arrays DOI 10.1002/jssc.201400684
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction A number of industrial dyes and their N-substituted aromatic biotransformation products are known to be toxic and/or carcinogenic. Thus, the use of these dyes in food products is considered harmful. At present, the number of harmful dyes detected in packaged food is upwards of 20 dye types, including Sudan I–V, para red, rhodamine B, orange II, acid red, Sudan Red, rhodamine B, Sudan Red 7B, metanil yellow, auramine, Congo Red, butter yellow, Solvent Red I, naphthol yellow, malachite green, leucomalachte green, Ponceau 3R, Ponceau MX, and Orange SS [1, 2]. It would be a demanding task to categorize the various types of dyes in packaged food by employing conventional methods [3–5]. Thus, it is highly
Correspondence: Prof. Hongdeng Qiu, Key Laboratory of Chemistry of Northwestern Plant Resources and Key Laboratory for Natural Medicine of Gansu Province, Lanzhou Institute of Chemical Physics, CAS, Lanzhou 730000, P. R. China E-mail: [email protected] Fax: +86-931-8277088
Abbreviations: AIBN, azobisisobutyronitrile; CDA, canonical discriminant analysis; EGDMA, ethylene glycol dimethacrylate; MAA, methacrylic acid; MIP, molecularly imprinted polymer C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
imperative to develop an effective, efficient, and suitable analysis method for identifying toxic dyes. Molecularly imprinted polymers (MIPs) are simple tailormade materials with highly specific recognition abilities for detecting target molecules. MIPs also possess excellent thermal, chemical, and mechanical stability [6–8]. Puoci et al. reported the synthesis of MIPs using Sudan I as a template, and extracted traces of Sudan I from spiked red chili powder by molecularly imprinted solid-phase extraction [9]. Yan and coworkers prepared Malachite Green-templated MIPs by a precipitation method, and found this method to preferentially bind the dye over other closely related compounds [10]. Recently, Piao et al. described the formation of a magnetic MIP column using Sudan IV, and the use of this polymer as a template to separate four Sudan dyes (Sudan I–IV) from food matrices [11]. The LOD was 6.2, 1.6, 4.3, and 4.5 ng/g for Sudan I–IV, respectively, using HPLC detection and molecularly imprinted solid-phase extraction. Multichannel MIP sensor arrays have been proven to be a highly practical and effective format with good discrimination and accuracy [12–14]. Their sensing strategy is similar to that of the nose or the tongue, which are able to differentiate and identify an almost unlimited number of fragrances and flavors by a finite number of different sensing cells. The sensor Colour Online: See the article online to view Figs. 5 and 6 in colour. www.jss-journal.com
2
J. Sep. Sci. 2014, 00, 1–7
Z. Long et al.
array format has proved to be useful for accurately differentiating multiple structurally similar compounds by transforming sensors of poor or modest selectivity into highly selective sensors. Yan et al. developed an imprinted mesoporous silica matrix sensor array that could differentiate not only individual carbohydrates but also different brands of orange juice [15]. Takeuchi et al. prepared an imprinted polymer sensor array using cytochrome C, ribonuclease A, and ␣-lactalbumin as a template, which could identify five proteins [16]. More recent studies have proven that the selectivity of an MIP sensor array can be tailored for a wide range of analytes and can be expressed in many different sensing formats. Thus, a sensor array could be created and categorized in terms of the use of different signaling mechanisms, including fluorescence, UV-Vis, electrical capacitance, and quartz crystal microbalance [17–21]. For example, Liu et al. reported the creation of a 2D imprinted polymer-coated mass-sensitive quartz crystal microbalance sensor array that can detect staphylococcus enterotoxin A and staphylococcus enterotoxin B [22]. Previous work using a colorimetric MIP array or a 2D sensor array orthogonal combination of four MIPs and three dyes have been successful for the discrimination of amines [13, 14]. However, very little attention has been paid to the discrimination of various structurally similar dyes by sensors. In this work, we developed a new seven-channel MIP sensor array for the accurate identification of water-soluble azo dyes. Good selectivity to recognize the azo dyes of the prepared sensor array was achieved.
2 Materials and methods 2.1 Reagents and apparatus All chemicals were purchased from commercial suppliers and used without further purification. All synthetic manipulations were carried out under a nitrogen atmosphere using standard Schlenk and cannula techniques. FTIR spectra were measured with a Vertex70 instrument (Bruker, Germany). The morphology of the MIPs was measured by a LEO1430VP scanning electron microscope (LEO, Germany). UV-Vis spectra were obtained using UV-2450 spectroscopy (Shimadzu, Japan). All nitrogen sorption measurements were performed on an Autosorb-iQ automated gas sorption analyzer (Quantachrome Instruments, USA) using 0.13–0.25 g of the polymer at 77 K. UHPLC analysis was performed on a Waters ACQUITY UPLC system equipped with a photodiode array detector.
2.2 Polymer preparation (P0–P5) Polymers P0–P5 were synthesized using previously reported methods [13]. Azobisisobutyronitrile (AIBN, 52.54 mg, 0.32 mmol), methacrylic acid (MAA, 0.27 mL, 3.20 mmol), ethylene glycol dimethacrylate (EGDMA, 2.50 mL, 12.80 mmol), and 0.33 mmol of template (B1–B5) were in C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
dividually mixed in a 3:1 v/v solution of ethanol/glycerol (40 mL) under N2 . The nonimprinted polymer P0 was prepared using the same mixture, but without the addition of any template molecule. Polymerization was carried out with a heat trigger and vigorously stirred in a 60⬚C constanttemperature water bath for 12 h. The polymers were subjected to Soxhlet extraction using an acetic acid/methanol solution (1:9 v/v) as an eluent for 24 h, and then thoroughly rinsed with acetonitrile to remove any residual acetic acid within the polymers. The polymers were dried under vacuum at 50⬚C for 24 h. 2.3 Preparation of polymer P6 Polymer P6 has been readily prepared using B6 as template by efficient synthetic routes similar to the polymers P0–P5. However, B6 is almost insoluble in any alcoholic solutions but can be easily dissolved in chloroform. So chloroform was used as the solvent for the synthesis of polymer P6. 2.4 Determination of the MIP sensor array (P0–P6) All analytes were used in their free base form. Polymers P0–P6 were individually placed into seven separate 20 mL comparison tubes. The polymers were equilibrated with 10 mL of deionized water solutions containing a single analyte (0.02 mM) for 4 h. The response of the array was measured by the removal of 2 mL aliquots of the supernatant and the subsequent analysis of this aliquot with UV-Vis spectroscopy. This process was repeated for analytes B1–B5 and D1 to generate five replicates of each analysis. Acetonitrile was used as the solvent for B6 because of this analyte’s insolubility in water. The seven analytes were tested against the seven polymers (P0–P6) five times to generate a 7 × 35 data matrix (7 sensors × 7 analytes × 5 replicates). This raw data matrix was processed using a canonical discriminant analysis (CDA) function as implemented in SPSS software. 2.5 Polymer surface area and porosity measurements Prior to nitrogen sorption measurements, the samples were degassed at 393 K for 3 h to remove adsorbed gases and moisture. Brunauer–Emmett–Teller (BET) surface areas for the polymers were calculated from the adsorption data with 0.162 nm2 as the molecular cross-sectional area for the adsorbed nitrogen molecules [23]. The Barrett–Joyner–Halenda (BJH) method was applied to calculate the pore-size distribution from the adsorption and desorption branches of the isotherms. 2.6 Sample preparation and UHPLC–UV analysis Equal volumes of 7.5 g/mL of each azo dye (B1–B5 and D1) were mixed at 4⬚C for 4 h. Then, these aqueous solutions were directly used for UHPLC–UV analysis. Five mg of P5 www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
Sample Preparation
3
3.1 Selectivity characteristics
Scheme 1. Structures of analytes B1–B6 and D1, corresponding to amaranth, tartrazine, sunset yellow, orange II, chrysoidin and para red, and rhodamine B, respectively.
was added to the mixture solution. After shaking for 4 h at room temperature, the samples were stored at 4⬚C, and then filtered through a 0.22 m membrane filter and analyzed by UHPLC–UV analysis. Polymer P5 was washed five times with 2 mL of the solution of methanol and acetic acid (9:1 v/v). The supernatant was obtained by centrifugation at 12 000 rpm for 15 min and was directly applied for UHPLC–UV analysis. UHPLC–UV analysis was performed at room temperature using an ACQUITY UPLC BEH shield RP18 (2.1 × 100 mm, 1.8 m) with acetonitrile (A) and a 10% ammonium acetate solution (B) as the mobile phase. The elution gradient used was A/B at 85:15 v/v for 0–3.5 min, A/B at 0:100 v/v for 3.5–6 min, and A/B at 85:15 v/v for 6–9 min. The flow rate was 0.2 mL/min and the injection volume was 5 L. A photodiode array with quantification at 440 nm was used for detection.
3 Results and discussion The MIPs (P1–P6) were prepared using a MAA/EGDMA matrix in the presence of six different template analytes, B1–B6, respectively. In addition, a blank, nonimprinted polymer (P0) was synthesized without using any template molecule. The polymers were synthesized under a nitrogen atmosphere in a 60⬚C water bath by using a 1:4 molar ratio of MAA/EGDMA with 10 mol% AIBN as the radical initiator. The prepared MIP sensor array was used to discriminate between the seven azo dyes (B1–B6 and D1, Scheme 1) by the analysis of their UVVis spectra. The response patterns were then processed using CDA analysis [24, 25]. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The affinity of each polymer for each of the seven analytes (B1–B6, D1) was tested by UV-Vis spectroscopy by vigorously shaking a constant weight of polymer with 0.02 mM solutions of each analyte, and then measuring the resulting UV-Vis spectra. As a result, the response was measured as absorbance (A) and the ratio of absorbance ((A0 –Ai )/A0 ) at 480 nm of each analyte solution before and after equilibration with the MIP array (Fig. 1). Thus, the fingerprint maps of each analyte against the MIP array are shown in the inset of Fig. 1. Each analyte was tested for five separate times against the MIP sensor array. Polymers P1–P6 made in the presence of template molecules appeared to be imprinted by the template. This was demonstrated by the lower A values or higher ((A0 –Ai )/A0 ) values for imprinted polymers P1 through P6 compared to those for nonimprinted polymer P0. In addition, the response of analyte B5 against the MIP sensor was the strongest at the same solution concentrations, because the usage amount of the individual polymers was only 1/4th of that of the other dyes. However, the response of each of the six dyes (B1–B6) against the MIP sensor array displayed a faint concentration gradient. This result suggested that the individual imprinted polymers showed poor overall selectivity, because they presented higher affinity not only to their template molecules but also to the other structurally similar analytes (especially P1, P3, and P6). The SEM images and the comparison of the considerable differences in the morphology of the chrysoidin (B5)imprinted polymer and its nonimprinted polymer are shown in Fig. 2. The SEM micrograph showed that the nonimprinted polymers had no cavity in contrast to MIPs and exhibited an irregular, amorphous morphology, as shown in the SEM image. The high-magnification SEM images of the B5-MIPS revealed the self-assembled structures of block copolymers toward the out-of-plane direction, fabricated by layer-by-layer accumulation of imprinted polymer particles with an average size of 200–500 nm. All the particles in the B5-MIPs had a sheet structure, with many large pores between the particles, possibly caused by the removal of the template molecule. MIPs with uniform and open structure are obviously favorable for the adsorption of the template molecules such as the used dye B5 [26]. Thus, this result suggested that the final polymer morphology could affect the selectivity of templaterebinding characteristics. Nitrogen-adsorption measurements at 77 K of the powdered samples of the polymers derived from P0–P6 confirmed their microporosity by demonstrating significant adsorption. The apparent BET surface areas and pore volumes of the polymers calculated by the BJH method are shown in Supporting Information Table S1. Polymer P5 was found to have greater microporosity, both in terms of the average pore diameter and the total pore volume, than nonimprinted polymer P0. This finding is consistent with previous observations that an imprinted polymer is more efficient at generating intrinsic microporosity than a nonimprinted polymer [23]. In addition, polymers P2 and P5 had www.jss-journal.com
4
Z. Long et al.
J. Sep. Sci. 2014, 00, 1–7
Figure 1. Absorbance change of seven analytes (B1–B6 and D1) against the MIP sensor array (P0–P6) after equilibration for 4 h: 10 mL of 0.02 mM at 480 nm (A) B1 aqueous solution; (B) B2 aqueous solution; (C) B3 aqueous solution; (D) B4 aqueous solution; (E) B5 aqueous solution; (F) B6 in acetonitrile; (G) D1 aqueous solution. Each polymer (P0–P6) was consumed individually: 20 mg for (A)–(C), (F) and (G); and 5 mg for (E). The inset pictures show special fingerprints of each of the seven analytes (B1–B6 and D1) against the MIP sensor array (average values of five separate replicates of each analyte).
Figure 2. (A) SEM image of nonimprinted polymers P0; (B) SEM image of P5. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
larger average pore diameters than the other imprinted polymers. To further determine the adsorption of B5 on the polymers (P0 and P5), an equimolar amount of analyte B5 was mixed with 5 mg of P0 and P5 in an aqueous solution. After 4 h of shaking, the samples (P0@B5 and P5@B5) were centrifuged and washed several times with 1 mL of deionized water, and then dried under vacuum at 50⬚C. The dried samples were finely crushed, mixed with potassium bromide (1:15 w/w), pressed at 15 000 psi, and analyzed using quantitative IR analysis. IR spectroscopy was also used to characterize the prepared materials (Fig. 3). In the IR spectrum of B5 (Fig. 3A), the bands at 3321 and 3135 cm−1 were assigned to the N– H stretching vibrations. The most intense band found at 1627 cm−1 was likely due to the stretching vibration of the –N=N– group. The N–H bending vibration of arylamine was at 1561 cm−1 , and the characteristic bands at 1512, 1446, and 1415 cm−1 were attributed to the aryl skeleton vibrations. As shown in Fig. 3B and D, the wide bands at 3565 and 3437 cm−1 were assigned to the stretching vibration of the hydroxyl group of P0 and P5, respectively. In comparison with P5, the band of the hydroxyl group in P0 shifted to a higher wavenumber, indicating the formation of intramolecular hydrogen bonds. However, the band of the hydroxyl group in P0 took a bathochromic shift after combination with B5, indicating the collapse of the intramolecular hydrogen bonds. A band also appeared at 1562 cm−1 , which was attributed to the bending vibration of arylamine in B5 (Fig. 3C). In the P5 and P5@B5 spectra, there appeared to be a hypochromatic shift (ca. 6 nm) corresponding to the stretching vibration of the hydroxyl group in P5 and the N–H bending vibration of arylamine in B5 (Fig. 3D and E). This result indicated the formation of intermolecular hydrogen bonds between B5 and P5. In addition, the band intensity of the N–H bending vibration of arylamine in Fig. 3E was stronger than that in Fig. 3C,
Figure 3. FTIR spectra of (A) B5; (B) P0; (C) P0@B5; (D) P5; and (E) P5@B5.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Sample Preparation
5
which showed that a greater amount of B5 was bound to P5 than P0 because of the imprinting process. These results also demonstrated that polymer P5 had a high selectivity for its template molecule, which is in accordance with that of the UV-Vis spectroscopy results. To further verify whether the MIPs synthesized were selective to their template molecules (azo dyes), the rebinding of those structurally related compounds (azo dyes) on polymer P5 was investigated as an example. Five milligrams of P5 was brought into contact with an aqueous solution of the dyes (C0 = 7.5 g/mL of each of dyes B1–B5 and D1, excluding water-insoluble B6). After the initial mixing, the suspension was allowed to rebind/absorb the dyes by shaking for 4 h at room temperature. The resulting suspension was stored at 4⬚C and filtered through a 0.22 m membrane filter, and then directly used for UHPLC–UV analysis. P5 was used for the selective extraction of B5 from the mixture as described in Fig. 4B. Because the peak area of each dye in UHPLC
Figure 4. UHPLC–UV chromatograms of a standard solution mixture (containing B1–B5 and D1) before and after adsorption by P5, measured at 440 nm: (A) the mixture standard solution before and (B) the supernatant after adsorption by P5, and (C) the collected eluent after P5 was separated from the eluent and purified.
www.jss-journal.com
6
J. Sep. Sci. 2014, 00, 1–7
Z. Long et al.
Figure 5. Plot of the response of the seven analytes (0.02 mM aqueous solutions for B1–B5 and D1, and 0.02 mM B6 solution in CH3 CN) tested in five times in separate replicates against the MIP sensor array at 480 nm.
chromatograms is directly proportional to its concentration within a given range [27], the concentrations of B1–B5 and D1 were calculated as 6.8, 7.2, 5.9, 7.0, 6.5, and 0.4 g/mL, respectively. Compared with the concentration of the other dyes, that of B5 was the lowest after being adsorbed by P5 (Fig. 4A and B). Polymer P5 was separated from the suspension and washed five times with 2 mL of methanol and acetic acid (9:1 v/v). Afterwards, any B5 released from P5 was collected and detected by UHPLC–UV analysis. As shown in Fig. 4C, the main species measured in this sample was B5. Small amounts of B4 and D1 were also measured in the eluent. These results demonstrated that polymer P5 was capable of high selectivity and specificity for its template molecules.
3.2 CDA analysis To investigate the recognition accuracy for individual dyes against the MIP sensor array, the analysis of five separate replicates of each analyte was carried out by UV-Vis spectroscopy under identical testing conditions. Seven different
values were measured for each analyte, shown as lines on the plot in Fig. 5. The responses of the individual analytes were found to correspond to each other for each of their five analysis repetitions. This showed excellent data reproducible for all analytes, likely due to the uniform particle size distribution by lower polymerization concentrations. Among the analytes, there was little overlap among B4, B5, and D1. However, there appeared to be much more overlap among B1, B2, B3, B4, and B6. Although there was a little amount of overlap between B1 and B6, their fingerprints were very close. As a result, only a limited portion of the fingerprint maps were distinguishable, because fingerprint recognition relies heavily on the quality of the input fingerprint images. To avoid this disadvantage of fingerprint resemblance and overlap, as well as to filter out random noise, multivariate analysis was applied to the dataset. The response patterns were processed using the CDA analysis, which is related to the principle component analysis and canonical correlation. The CDA is used as a means of distinguishing among a group of samples from potentially different populations. To evaluate the responses of the sensor elements, five replicate datasets were consolidated into a 7 × 35 data matrix data sheet (7 sensors × 7 analytes × 5 replicates). CDA analysis was conducted to reduce the dimensionality of the data. The analysis showed impressive levels of discrimination, as the response patterns for each dye were well differentiated in the CDA plot (Fig. 6). The replicate data points for each analysis were clustered together. These groupings were separated equally from one another. The classification accuracy of the original data was up to 100.0% for both the original grouped cases correctly classified and crossvalidated. Rhodamine B (D1) was also correctly classified, even though it was not one of the original template analytes (B1–B6). These results showed that, once an array has been prepared against one set of analytes, it could facilitate the classification of other analytes. This finding greatly enlarges the range of detectable analytes using an MIP array. 3.3 Real sample analysis Real samples were analyzed and identified according to their canonical scores or their placement in the 2D space. The
Figure 6. (A) The spectral response of azo dye against the MIP sensor array, taking B5 as example for one trail; (B) 2D CDA plot of the seven analytes against the MIP sensor array (five trails each) showing clear clustering of the trails.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
J. Sep. Sci. 2014, 00, 1–7
array response to each sample was compared to the classification data, and the sample dyes were identified according to the correspondence between their Mahalanobis distances and those of the classified azo dyes (Supporting Information). Spiked B5 in the smoked bean curd extract and Fanta lime soda (containing B2) against the array were identified with 100.0% accuracy for the original grouped cases; among the cross-validated grouped cases, 94.6% of the cases were correctly classified.
4 Conclusions We have demonstrated the utility of an MIP sensor array for the discrimination of various azo dyes. MIP arrays of different selectivity could be rapidly and inexpensively prepared and adjusted by alternating the initial templates used in the imprinting process. The response patterns of these dyes against the MIP sensor array were analyzed by the CDA with 100% classification accuracy. In addition, the array was found to be capable of classifying D1, even though no receptor was specifically designed for this analyte. Furthermore, the array’s identification ability was found to work successfully when used to analyze real samples. As a result, the MIP sensor arrays developed in this work could have a large number of broad practical applications, as they permit discrimination between species in the closely related azo-structural family. The authors are grateful for financial support from Xinjiang provincial natural science fund project (Grant No. 2012211A107). H. Qiu acknowledges the support of the “Hundred Talents Program” of Chinese Academy of Sciences. The authors have declared no conflict of interest.
5 References [1] Lok, C. M., Son, R., Food Res. Int. 2009, 16, 127–140. [2] Oplatowska, M., Donnelly, R. F., Majithiya, R. J., Glenn Kennedy, D., Elliott, C. T., Food Chem. Toxicol. 2011, 49, 1870–1876. [3] Zou, T., He, P., Yasen, A., Li, Z., Food Chem. 2013, 138, 1742–1748. [4] Taverna, D., Di Donna, L., Mazzotti, F., Policicchio, B., Sindona, G., J. Mass Spectrom. 2013, 48, 544–547. [5] Fang, G., Wu, Y., Dong, X., J. Agric. Food Chem. 2013, 61, 3834–3841. [6] Svenson, J., Nicholls, I. A., Anal. Chim. Acta. 2001, 435, 19–24.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Sample Preparation
7
[7] Bossi, A. M., Sharma, P. S., Montana, L., Zoccatelli, G., Laub, O., Levi, R., Anal. Chem. 2012, 84, 4036– 4041. [8] Cheong, W. J., Yang, S. H., Ali, F., J. Sep. Sci. 2013, 36, 609–628. [9] Puoci, F., Garreffa, C., Iemma, F., Muzzalupo, R., Spizzirri, U. G., Picci, N., Food Chem. 2005, 93, 349–353. [10] Yan, S. L., Gao, Z. X., Fang, Y. J., Cheng, Y. Y., Zhou, H. Y., Wang, H. Y., Dyes Pigments 2007, 74, 572–577. [11] Piao, C., Chen, L., J. Chromatogr. A 2012, 1268, 185–190. [12] Greene, N. T., Morgan, S. L., Shimizu, K. D., Chem. Commun. 2004, 10, 1172–1173. [13] Greene, N. T., Shimizu, K. D., J. Am. Chem. Soc. 2005, 127, 5695–5700. [14] Lee, J.-D., Hong, J. I., Tetrahedron Lett. 2013, 54, 2890– 2893. [15] Tan, J., Wang, H. F., Yan, X. P., Anal. Chem. 2009, 81, 5273–5280. [16] Takeuchi, T., Goto, D., Shinmori, H. Analyst 2007, 132, 101–103. [17] Shimizu, K. D., Stephenson, C. J., Curr. Opin. Chem. Biol. 14, 743–750. [18] Qiu, H., Luo, C., Sun, M., Lu, F., Fan, L., Li, X., Anal. Chim. Acta. 2012, 744, 75–81. [19] Croux, D., Weustenraed, A., Pobedinskas, P., Horemans, ¨ H., Haenen, K., Cleij, T., Wagner, P., Thoelen F., Dilien, R., De Ceuninck, W., Phys. Status Solidi. A 2012, 209, 892–899. [20] Zang, D., Yan, M., Ge, S., Ge, L., Yu, J., Analyst 2013, 138, 2704–2711. [21] Alizodeh, T., Rezaloo, F., Sensor. Actuat. B Chem. 2013, 176, 28–37. [22] Liu, N., Gao, Z., Ma, X., Ou, G., Li, X., Chen, X., CN 201110459803 A, 2012. [23] Rosengren, A. M., Karlsson, B. C. G., Nicholls, I. A., Int. J. Mol. Sci. 2013, 14, 1207–1217. ´ [24] de Mendoza, M. F., Gordillo, C. D. M., Expoxito, J. M., Casas, J. S., Cano, M. M., Vertedor, D. M., Baltasar, M. N. F., Food Chem. 2013, 141, 2575–2581. [25] Dimauro, C., Cellesi, M., Steri, R., Gaspa, G., Sorbolini, S., Stella, A., Macciotta, N. P. P., Anim. Genet. 2013, 44, 377–382. [26] Yoshimatsu, K., Reimhult, K., Krozer, A., Mosbach, K., Sode, K., Ye, L., Anal. Chim. Acta. 2007, 584, 112–121. [27] You, Y., Uboh, C. E., Soma, L. R., Guan, F., Li, X., Rudy, J. A., Liu, Y., Chen, J., Rapid Commun. Mass Sp. 2009, 23, 2035–2044.
www.jss-journal.com
