Electrochemical Molecular Imprinted Sensors ... - Semantic Scholar

ANALYTICAL SCIENCES AUGUST 2015, VOL. 31

793

2015 © The Japan Society for Analytical Chemistry

Electrochemical Molecular Imprinted Sensors Based on Electrospun Nanofiber and Determination of Ascorbic Acid Yunyun ZHAI, Dandan WANG, Haiqing LIU,† Yanbo ZENG, Zhengzhi YIN, and Lei LI† College of Biological, Chemical Sciences and Engineering, Jiaxing University, Jiaxing, Zhejiang Province, 314001, P. R. China

In this study, electrochemical molecularly imprinted sensors were fabricated and used for the determination of ascorbic acid (AA). Nanofiber membranes of cellulose acetate (CA)/multi-walled carbon nanotubes (MWCNTs)/ polyvinylpyrrolidone (PVP) (CA/MWCNTs/PVP) were prepared by electrospinning technique. After being transferred to a glass carbon electrode (GC), the nanofiber interface was further polymerized with pyrrole through electrochemical cyclic voltammetry (CV) technique. Meanwhile, target molecules (such as AA) were embedded into the polypyrrole through the hydrogen bond. The effects of monomer concentration (pyrrole), the number of scan cycles and scan rates of polymerization were optimized. Differential pulse voltammetry (DPV) tests indicated that the oxidation current of AA (the selected target) were higher than that of the structural analogues, which illustrated the selective recognition of AA by molecularly imprinted sensors. Simultaneously, the molecularly imprinted sensors had larger oxidation current of AA than non-imprinted sensors in the processes of rebinding. The electrochemical measurements showed that the molecularly imprinted sensors demonstrated good identification behavior for the detection of AA with a linear range of 10.0 – 1000 μM, a low detection limit down to 3 μM (S/N = 3), and a recovery rate range from 94.0 to 108.8%. Therefore, the electrochemical molecularly imprinted sensors can be used for the recognition and detection of AA without any timeconsuming elution. The method presented here demonstrates the great potential for electrospun nanofibers and MWCNTs to construct electrochemical sensors. Keywords Electrospinning, carbon nanotubes, molecularly imprinted technology, electrochemical sensors (Received January 14, 2015; Accepted April 30, 2015; Published August 10, 2015)

Introduction Molecular imprinting technology (MIT) is a powerful technique of preparing synthetic polymers with predetermined molecular Compared with the traditional recognition properties.1–3 separation detection technology, MIT has many superior characteristics such as high selectivity, pre-determined structure and good recognition. The imprinted polymers are resistant, robust and stable in a wide range of pH. They have good anti-interference ability, thermal stability and recycling use.4–6 Molecularly imprinted polymer (MIP) has been widely used in many fields, such as chromatography separation, reaction catalysts, solid-phase extraction and biological sensors.7 Traditional molecular imprinted materials are usually bulk polymers. Therefore it is difficult for the extraction of original templates located in the interior area due to the highly cross-linking nature. Thus, it is important to develop a new imprinted material that can be easily extracted from its original template. Currently, electrospinning is a versatile method used to prepare continuous fibers with diameters ranging from a few nanometers to several hundred nanometers (usually between 2 To whom correspondence should be addressed. E-mail: [email protected] (H. L.); [email protected] (L. L.) †

and 500 nm).8 The method can be applied to polymers, polymer alloys and blends, and polymers loaded with nanomaterials.7–11 By regulating the process parameters and the feeding polymer composition, the electrospinning process can be adjusted to control the fiber morphology (such as diameter, surface area, porosity and pore size) and the physical and chemical properties of the nanofibers.12 Electrospun nanofibers can be easily constructed to produce well defined three-dimensional mats with well-controlled pore sizes. Because of their high specific surface area, high porosity, and flexibility, electrospun nanofiber materials are considered to be an ideal candidate for industrial applications. It is also highly desirable to introduce the selective molecular recognition sites into nanofiber materials.13 They may provide highly efficient compound separation involving high throughput and good selectivity. In this work we investigated the possibility of preparing electrospun nanofibers that contain molecularly imprinted binding sites. Ascorbic acid (AA) is an important water soluble substance that can be found in fruits, vegetables, medicines, etc. It is used as an antioxidant in some processed foods and a radical scavenger.14 Meanwhile, it plays a very important role for the treatment of cancer and Parkinson’s psychosis.15 Therefore, the accurate detection of AA has an considerable significance for human health and the assessment of food quality.16 Up to now, the common detection methods for AA were liquid chromatography, mass spectrometry, spectrophotometry, gas chromatography, solid phase extraction, and electrochemical

794 methods.17 Of this the electrochemical methods are simple, rapid and offer good sensitivity. While the use of electrochemical methods for the detection of real samples is still a matter of concern because of the poor selectivity of the method and the existence of chemical substances with similar oxidation or reduction potential as the target molecule. For example, dopamine (DA) and uric acid (UA) have approximative oxidation potential for AA. The modification of electrodes with nanomaterials, such as carbon nanotubes, C60, and graphene, can significantly improve the detection sensitivity.18 As such, many scientists devote a great deal of attention to the separation of AA, DA and UA with some novel nanomaterials in electrochemistry, because of this approximative oxidation potential.19–23 In our previous work, the simultaneous detection of AA and UA was developed with nanostructured gold, due to their high electrocatalytic activities toward the oxidation of AA and UA.24 In this work, we investigated the possibility of preparing electrospun nanofibers that contain molecularly imprinted binding sites. The cellulose acetate (CA)/multi-walled carbon nanotubes (MWCNTs)/polyvinylpyrrolidone (PVP)-based nanofiber membranes were fabricated by electrospinning technique. The nanofiber interface was further polymerized with pyrrole through electrochemical cyclic voltammetry (CV). Polypyrrole can act as both a functional monomer and cross-linking agent. The electrochemical molecularly imprinted sensors were used for the determination of AA. Furthermore, the monomer concentration, scan rate and scan cycles were optimized in order to achieve the selective detection and identification of AA.

Experimental Chemicals CA (Hydroxyl content, 8.7%, wt%), PVP (Mw = 1300000 g mol–1), dimethyl formamide (DMF), thionyl chloride (SOCl2), Tetrahydrofuran (THF), AA (AR) and pyrrole (99%) were all supplied by Aladdin Reagent. Lithium perchlorate (LiClO4), histidine (L-His), uric acid, tryptophan (L-Trp) and glucose (Glucose) were purchased from Sinopharm Chemical Reagent Co., Ltd., China. Carboxylated MWCNTs were purchased from Shenzhen Nanotech Port Co., Ltd., China (purity: > 97%, length: < 5 μm, diameter: 20 – 40 nm, the carbon atom bearing OH group: 3.0%, the carbon atom bearing COOH group: 2.3%). Distillated water was used throughout this work and all the reagents were used as received without further purification. Instruments The surface morphologies of the nanofiber membranes were studied using scanning electron microscopy (SEM) (S-4800, Hitachi, Japan). Fourier transform infrared spectral data were collected using NEXUS470 (Thermo Nicolet, USA). A pHmeter (Mettler Toledo, Switzerland) was used to measure the pH values. Electrochemical measurements were performed on a CHI660D electrochemical workstation in a conventional threeelectrode system, using the MIP/GCE (3 mm in diameter) as the working electrode, an Ag/AgCl (saturated KCl) electrode as the reference electrode and a platinum wire as the counter electrode. The electrolyte was KH2PO4–K2HPO4 buffer solution (PBS) (0.05 M, pH 8.5), which was deoxygenated with nitrogen and kept under a nitrogen atmosphere during the electrochemical measurements.

ANALYTICAL SCIENCES AUGUST 2015, VOL. 31 Preparation of CA-MWCNTs Under supersonification, MWCNTs (0.67 g) were dispersed in the mixture of SOCl2 (0.8 mL), DMF (0.15 mL) and THF (8 mL) and refluxed for 24 h at 65° C. Then, the solvent was removed under reduced pressure, and the products were repeatedly dispersed in THF and separated via centrifugation (10000 rpm, 10 min) to fully remove the residual SOCl2. The MWCNTs-COCl was added into a solution of CA (0.6 g), THF (15 mL) and pyridine (0.8 mL) under vigorous stirring. Then the mixture was refluxed for 6 h. The solvent was removed under vacuum and the residual washed with water 6 times to remove the excess pyridine and its salt form. Preparation of the electrospun nanofiber membranes CA and PVP were dissolved in a mixture of acetone/DMAc (v/v = 2:1), and then the CA-MWCNTs were added under constant stirring at room temperature for about 8 h. The detailed percentage of these electrospinning composites are shown in Table S1 (Supporting Information). For electrospinning, the solution (5.0 mL) was loaded in a glass syringe with a stainless steel pin jointed in the solution and connected to a power supply (JG50-1, Shanghai Shenfa Detecting Instrument Factory, China). The glass syringe was docked to a syringe pump (KDS200, KD Scientific Inc., USA) with a constant flow rate of 1.0 mL/h. A grounded counter electrode was connected to a stainless steel collector (S = 15.2 cm2) that was set 20 cm away from the spinneret which was positioned perpendicular to the collector in a coaxial manner. The electrospinning was performed under a voltage of 10 kV at room temperature and humidity between 70 – 80%. Electrospinning time of 15 min was employed unless otherwise stated. The resulting fibrous mats were collected and dried under vacuum at room temperature for 24 h. Preparation of molecularly imprinted electrochemical sensors A bare GC electrode was polished to a mirror-like surface with 0.3 and 0.05 μm alumina slurry, followed by sonication in ethanol and distilled water for 30 s, respectively. The nanofiber membranes containing MWCNTs were punched into circular disks (3 mm diameter) and transferred onto the GC electrode. The nanofiber membranes were immobilized on the GC electrode with the chitosan solution (0.5%). Then the nanofiber electrode was immersed into the solution containing pyrrole (25 mM), AA (10 mM) and LiClO4 (100 mM), followed by the cyclic scan in the potential range from –0.6 to 0.8 V for 7 cycles at 100 mV/s. After the electro-polymerization, molecularly imprinted polymer membrane sensors for AA were obtained. The MIP electrode was immersed into PBS solution with magnetic stirring for 15 min to remove the template (AA). The production process for the non-imprinted polymer (NIP) electrode was the same except without the template molecule (AA) in the electro-polymerization. The MIP electrode was immersed into the PBS solution containing AA. After adsorption for 5 min, differential pulse voltammetry (DPV) measurement was carried out at the pulse amplitude of 50 mV in the potential range from –0.2 to 0.4 V. And the pulse width, pulse period and voltage increase were set as 200 ms, 0.5 s, and 10 mV, respectively.

Results and Discussion Characterization of the MIP In our work, PVP and CA were selected as the supporting fiber matrix (Scheme 1). The continuous nanofibers were obtained by electrospinning from the polymer solution with a


795

Scheme 1 Illustration of the preparation procedure for the MIP and detection process for its electrochemical sensor.

Fig. 1 SEM images of the electrospun CA/MWCNTs/PVP (A) and CA/PVP nanofibers (B). Scale bars: 2 μm. Diameter distributions for CA/MWCNTs/PVP (C) and CA/PVP nanofibers (D).

concentration of 13 wt%. As revealed by the SEM images (Fig. 1(A)), the long CA/MWCNTs/PVP nanofibers were randomly distributed in a fibrous mat with very uniform diameters in the range of 200 ± 20 nm (frequency 82% in Fig. 1(C)). As shown in Fig. 1(B), the CA/PVP polymer solution increased the fiber diameters (300 ± 25 nm) (frequency 80% in Fig. 1(D)). As shown in Fig. S1(A) (Supporting Information), with the

electrochemical polymerization of pyrrole, there were wide oxidation peaks at 0.15 V and wide reduction peaks at 0 V. Furtermore, the new oxidation peaks at 0.45 V (Fig. S1(B), Supporting Information) were ascribed to the presence of the AA in the polypyrrole backbone, which indicated the successful preparation of the MIP sensors. After the electro-polymerization, the polypyrrole backbone exhibited positive charges.25,26 This means the anions in the solution can be embedded in the

796

Fig. 2 Before (a) and after (b) removal of the template molecule at the imprinted electrode.

polypyrrole backbone.27,28 Thus, the molecularly imprinted modified electrode can be prepared by the electrostatic attraction between the negative charges of template molecules and the positive charges of the polypyrrole backbone. The infrared spectra of imprinted and non-imprinted membranes are shown in Fig. S2 (Supporting Information). For non-imprinted membranes, the carbonyl vibration (1743.3 cm–1) and hydroxyl vibration (3467 cm–1) were not clear. In the molecularly imprinted membrane, the carbonyl vibration (1743.3 cm–1) and hydroxyl vibration (3467 cm–1) significantly increased, due to the carbonyl and hydroxyl in AA. The comparative results showed that the AA was successfully electrically polymerized Figure S3 into the molecularly imprinted membranes. (Supporting Information) shows the scanning electron microscope image of imprinted and non-imprinted membranes. It can be seen that the MIP and NIP membranes had uniform fiber diameters from 400 ± 20 nm. They have larger diameters than CA/MWCNTs/PVP. The image illustrates that PPy was polymerized onto the surface of CA/MWCNTs/PVP nanofibers. Electrochemical impedance spectroscopy can accurately reflect the electron transfer performance in the process of sensor electrode preparation. In Fig. S4(A) (Supporting Information), the electron transfer impedance of the CA/MWCNTs/PVP membranes (50 kΩ) was significantly reduced compared to the CA/PVP membranes (150 kΩ), which indicated that the small quantity of doping MWCNTs (5.62% in the CA/MWCNTs/PVP nanofibers) constructed a high-speed electron transfer path between CA/PVP nanofibers and the electrolyte. In Fig. S4(B) (Supporting Information), the MIP shows much larger impedance (800 kΩ) after electro-polymerization because of the absorption of AA. After the extraction of AA, the impedance (135 kΩ) rapidly decreased to approach the value of the CA/ MWCNTs/PVP membranes. When AA was newly adsorbed into the MIP cavity, the impedance rapidly increased again (600 kΩ). This was attributed to the adsorptive AA hindering the redox probe to reach the electrode surface. Optimization of experimental variables The oxidation current of adsorptive AA with adsorption time are shown in Fig. S5 (Supporting Information). The oxidation current rapidly increased with increasing time from 0 to 5 min in an AA solution (100 μM). When the time was longer than


Fig. 3 Contrast of MIPs with the nanofiber (a) and MIPs without the nanofiber (b) for adsorbing in the same AA solution (1 mM).

5 min, the oxidation current was steady. As such, we deemed 5 min was the appropriate adsorption time. In the electropolymerization process, the concentration of pyrrole can affect the thickness of the membranes and the number of AA. The electro-polymerization was carried out in the pyrrole solutions with different concentrations. As shown in Fig. S6 (Supporting Information), the oxidation current of MIP increased with the concentration of pyrrole. When the concentration was higher than 25 mM, the current decreased. Therefore, for this experiment we determined that the optimal concentration of pyrrole was 25 mM. In the electro-polymerization process, the number of scan cycles will affect the sensitivity and linearity of the sensors. As shown in Fig. S7 (supporting information), the oxidation current of AA increased with the increase of scan cycles. Too many scan cycles could lead to thick polymer membranes on the CA/ MWCNTs/PVP nanofibers. This means some of the electrochemical active sites were embedded and lost their recognition activity. When the number of cycles reached seven, the currents for the MIP were steady. So the optimal number of scan cycles was determined to be seven. In the electropolymerization, scan rate was an important experimental factor. It influenced the formation of polymer membranes. Low scan rate of polymerization will make the membrane dense, which would hinder the conduction of the redox probe. Meanwhile, the formation of loose polymer membranes at a high scan rate reduces recognition capacity. As shown in Fig. S8 (Supporting Information), the oxidation current response to AA reached the maximum at the scan rate of 100 mV/s. The adsorption characteristic and electrochemical analysis of MIP Elution of the template molecule was carried out in PBS solution and tracked by DPV method. As shown in Fig. 2, there was a sharp oxidation peak at 0.0 V before the extraction of AA. After extracting for 15 min, the oxidation peak of AA disappeared, indicating complete removal of the template molecule. To evaluate the nanofiber in the MIP, a comparison of sensing characteristics of the MIP film with and without the nanofibers is shown in Fig. 3. Without the nanofibers, the surface of the bare GC electrode was smooth. The surface increased when the CA/MWCNTs/PVP nanofiber was transferred to the GC electrode. So the combination sites in the nanofiber were more than in the bare GC surface. As shown in


Fig. 4 Contrast of MIPs and NIPs for adsorbing in the same AA solution (100 μM).

Fig. 3, the current without the nanofibers was 61% compared to the current with the nanofibers. The adsorption of AA for the MIP and NIP are shown in Fig. 4. The MIP or NIP electrode was immersed into PBS solution containing AA (100 μM) for 5 min. Then the electrode was cleaned with water to remove some AA molecules that were not firmly adsorbed onto the nanofibers. As shown in Fig. 4, the current of AA for the MIP was 3.2 times higher than for the NIP, indicating the formation of the imprinted sites in the MIP nanofibers. To further investigate the selective recognition properties of the MIP, some structural analogues were selected to act as the competitors. The corresponding structures of L-His, UA, L-Trp and glucose are shown in Fig. 5. As shown in Fig. 5, the oxidation current of the MIP for AA was higher than that of the structural analogues. It can be implied that the MIP provided high selectivity to AA, which illuminated that the specific recognition sites for the template molecule were created during the course of imprinting. In order to apply the proposed method in real samples, it was vital to investigate the effect of some interferences, which was used to evaluate the selectivity of the MIP. The DPV determination of the MIP was tested in the presence of spiked interferences. The tolerance limit was defined as the amount of interferences causing a change of ±5% in the peak current intensity. The tolerable limits of interferences are given in Table S2 (Supporting Information). The results indicated that the MIP exhibited a high adsorption and strong affinity toward AA. As can be seen from Fig. 6, the peak current linearly increased with the AA concentration in the range of 10 – 1000 μM, with the linear equation of Ipa (μA) = 0.0606 – 0.0138C (μM) (R2 = 0.9988). The detection limit was 3 μM for the MIP (S/N = 3, defined as the concentration of AA corresponding to three times the standard deviation for replicate detections of the blank solution). Similarly, the fabrication reproducibility was estimated with the DPV method by using eight MIP electrodes. A solution containing AA (100 μM) was determined by these eight electrodes in the same electrochemical cell, with a relative standard deviation of 4.2%, which indicated the good reproducibility of the modified electrode. After use 10 times for the detection of AA (100 μM), a response of 95 ± 4.5% (n = 6) of the initial current was achieved. Moreover, a response of 92 ± 4.5% (n = 6) of the initial current was retained for an MIP electrode after it was stored at room temperature for 2 weeks. This allowed us to conclude that the electrode has good

797

Fig. 5 Current responses of AA and the four structural formula compounds (L-His, UA, L-Trp, Glucose) at the MIP with the same concentration (200 μM) (n = 5).

reproducibility and stability. To demonstrate the feasibility of the MIP, the proposed procedure was applied to determine AA in chewable tablets of vitamin C. The chewable tablets of vitamin C (0.610 g) were grinded and dissolved with ultrapure water (50 mL), followed by filtration and dilution to 250 mL, and then the solution was diluted with PBS (pH 8.5) to a measurable range. The concentration values of AA in the real samples were determined by the proposed method, and the content of AA in the chewable tablet was 37 ± 2.5% (n = 6), which was consistent with the introduction of the packing specification. The spiking method was also adopted to evaluate the AA content in the real samples. The electrochemical results are shown in Table 1. The recoveries were from 94.0 to 108.8%. The results demonstrated that the proposed DPV procedure based on the MIP can be applied for detecting AA in chewable vitamin C samples.

Conclusions In summary, electrochemical molecularly imprinted sensors were prepared by electrospinning and electro-polymerization. Nanofiber membranes of CA/MWCNTs/PVP were prepared by electrospinning technique. Then the nanofiber interface was further electro-polymerized with pyrrole through electrochemical CV. The sensitive and selective electrochemical DPV method was developed for detecting AA based on the MIP electrode. The experimental results demonstrated that the MIP electrode exhibited high adsorption and strong affinity toward AA. This MIP electrode proved to be a suitable sensing tool for the fast, sensitive and selective determination of AA in health care products. This work provided a versatile strategy for the further design and development of electrochemical molecularly imprinted sensors.

Acknowledgements This work was supported by the National Natural Science Foundation of China (Nos. 21177049 and 51103063) and the Zhejiang Provincial Natural Science Foundation (Nos. Y4110545, LQ12B05005, and LQ14B050002).

798


Fig. 6 6(A) DPV curves of AA with different concentrations at the MIP at pH 8.5 after incubation for 5 min. The concentration of AA in the curves was as follows: 20, 50, 120, 170, 330, 500, 670, 830, and 1000 μM, respectively. (B) The linear relationship between the peak current and the concentration of AA (n = 8).

Table 1 Determination results of AA in chewable vitamin C samples (n = 6) Sample Detected, wt% 1 2 3 4 5 6

Added/ Found/ Recovery, % RSD, % c mM–1 c mM–1

39.5 36.9 34.6 35.6 38.8 35.8

1 2 5 1 2 5

0.94 1.91 4.91 1.05 2.18 4.83

94.0 95.3 98.1 105 108.8 96.6

3.3 2.4 3.9 4.5 3.3 2.8

Supporting Information Characterization and optima conditions were shown in Supporting Information. This material is available free of charge on the Web at http://www.jsac.or.jp/analsci/.

References 1. E. L. Holthoff and F. V. Bright, Acc. Chem. Res., 2007, 40, 756. 2. L. Chen, S. Xu, and J. Li, Chem. Soc. Rev., 2011, 40, 2922. 3. S. C. Zimmerman, M. S. Wendland, N. A. Rakow, I. Zharov, and K. S. Suslick, Nature, 2002, 418, 399. 4. D. C. Apodaca, R. B. Pernites, F. R. Del Mundo, and R. C. Advincula, Langmuir, 2011, 27, 6768. 5. D. C. Apodaca, R. B. Pernites, R. Ponnapati, F. R. Del Mundo, and R. C. Advincula, Macromolecules, 2011, 44, 6669. 6. M. Riskin, Y. Ben-Amram, R. Tel-Vered, V. Chegel, J. Almog, and I. Willner, Anal. Chem., 2011, 83, 3082. 7. G. Wulff and J. Liu, Acc. Chem. Res., 2012, 45, 239. 8. A. Greiner and J. H. Wendorff, Angew. Chem., Int. Ed., 2007, 46, 5670.

9. J. A. Lee, Y. S. Nam, G. C. Rutledge, and P. T. Hammond, Adv. Funct. Mater., 2010, 20, 2424. 10. C. Zhu, X. Mu, P. A. van Aken, Y. Yu, and J. Maier, Angew. Chem., Int. Ed., 2014, 53, 2152. 11. S. L. Sampson, L. Saraiva, K. Gustafsson, S. N. Jayasinghe, and B. D. Robertson, Small, 2014, 10, 78. 12. X. Lu, C. Wang, and Y. Wei, Small, 2009, 5, 2349. 13. S. Chigome, G. Darko, and N. Torto, Analyst, 2011, 136, 2879. 14. M. Huang, Q. Xu, and X. Deng, J. Plant. Physiol., 2014, 171, 1205. 15. F. E. Harrison, G. L. Bowman, and M. C. Polidori, Nutrients, 2014, 6, 1752. 16. J. Du, J. J. Cullen, and G. R. Buettner, Biochim. Biophys. Acta, Rev. Cancer, 2012, 1826, 443. 17. A. M. Pisoschi, A. Pop, A. I. Serban, and C. Fafaneata, Electrochim. Acta, 2014, 121, 443. 18. X. Kan, Y. Zhao, Z. Geng, Z. Wang, and J. Zhu, J. Phys. Chem. C, 2008, 112, 4849. 19. Y. Bao, J. Song, Y. Mao, D. Han, F. Yang, L. Niu, and A. Ivaska, Electroanalysis, 2011, 23, 878. 20. L. Kuo-Chiang, T. Tsung-Hsuan, and C. Shen-Ming, Biosens. Bioelectron., 2010, 26, 608. 21. M. Noroozifar, M. Khorasani-Motlagh, R. Akbari, and M. B. Parizi, Biosens. Bioelectron., 2011, 28, 56. 22. Z. Xue, Y. Feng, H. Guo, C. Hu, A. M. I. Mohmed, J. Li, and X. Lu, RSC Adv., 2014, 4, 5849. 23. Y. Zhang, Z. Xia, H. Liu, M. Yang, L. Lin, and Q. Li, Sens. Actuators, B, 2013, 188, 496. 24. H. Liu and Y. Tian, Electroanalysis, 2008, 20, 1227. 25. B. Deore, Z. Chen, and T. Nagaoka, Anal. Chem., 2000, 72, 3989. 26. Z. Chen, Y. Takei, B. Deore, and T. Nagaoka, Analyst, 2000, 125, 2249. 27. N. Ferraz and A. Mihranyan, Nanomedicine, 2014, 9, 1095. 28. Z. Huang, G. Yin, X. Liao, and J. Gu, Front. Mater. Sci. China, 2014, 8, 39.