Electrochemical sensor for melamine based on its copper complex

Abstract

A reliable and highly sensitive electrochemical sensor was first developed for analysis of non-electroactive melamine (Mel) based on its conversion to an electroactive complex by coordination of copper salt to Mel. This provides a simple and easy approach to the detection of Mel in milk products.

---

As an important component of plastic resin, Mel can be found at ppm levels in food and beverages due to migration from Mel-containing resins, but these levels do not pose any danger to humans.¹ Recently, much attention has been focused on the detection of Mel since nitrogen-rich (66% by mass) Mel was intentionally adulterated to milk products to show a false increase in protein concentration. Ingestion of Mel at levels above the safety limit (0.5 ppm in the EU) can induce renal failure and even death, as in the scandal involving Mel tainted infant formula in China in September 2008.² Therefore, the growing concern over tainted milk products requires a simple and effective method for the analysis of Mel in these products.

Various methods including gas chromatography (GC),³ liquid chromatography (LC),⁴ capillary electrophoresis (CE),^5,6 ultraviolet-visible spectroscopy (UV-Vis),⁷ enzyme-linked immunosorbent assay (ELISA),⁸ infrared spectroscopy (IR)⁹ and surface enhanced Raman spectroscopy (SERS),¹⁰ and even colorimetry¹¹ have been developed for the detection of Mel in various matrices. However, there are very few reports regarding the utilization of electrochemical methods except Slisarenko's amperometry¹² and Li's polarography,¹³ owing to poor electroactivity of Mel. So the motivation for this study is derived primarily from the anticipation of improvement of electroactivity of Mel. Our strategy is based on a conversion from non-electroactive Mel to an electroactive Mel complex. Electroactive metal (M) cations become our first choice naturally. It is reasonable that the produced M–Mel complex may find important application in detecting Mel using cheap electrochemical instruments. There were some reports regarding the synthesis and structure of Mel complexes with M–Mel bonding.^14–16 However, all of these synthetic methods require time-consuming treatments including reflux, extraction and filtration etc., because Mel is a surprisingly poor ligand especially in aqueous medium. Obviously they are not suitable for the development of a rapid and simple electrochemical method. It is demonstrated herein for the first time that a Cu–Mel complex can be on-line formed at the surface of a multi-walled carbon nanotubes (MWCNTs) modified electrode in the presence of an excess of copper ions. Furthermore, the Cu–Mel complex shows a sensitive faradaic response that can be used for reliable detection of Mel. The features of this new approach are simplicity, rapidity and sensitivity. Its application to rapid detection of Mel matches the urgent need for the tainted milk product issue.

As a trimer of cyanamide with a 1,3,5-triazine skeleton, Mel has very good stability and relatively poor electroactivity. Even though under strong alkaline conditions, it shows a very weak electrochemical response originating from the electrooxidation of the amino group, such low sensitivity and high potential are not suitable for Mel analysis. The addition of copper ions to this system however may change this situation. Since the pK_a of Mel is about 6, neutral or weak alkaline condition should be suitable for the formation of a Cu–Mel complex by coordination of cuprous cation to neutral Mel. In addition, strong alkaline conditions would be detrimental to the dissolution of copper salts while large amounts are necessary during coordination. As shown in Fig. 1A, in 0.10 M phosphate buffer containing 0.10 M KCl (pH 7.0), no other obvious redox peak appears in the cyclic voltammogram (CV) measured at a MWCNTs modified electrode except broad peaks at about 0 V due to partial structural defects of the carboxylated MWCNTs (curve a).¹⁷ The CV measured in the presence of 40 μM Cu²⁺ demonstrates typical copper oxidation/reduction behavior associated with formation of Cu⁺ and Cu²⁺, which corresponds to P_a1 at −0.024 V and P_c1 at −0.193 V, as well as P_a2 at 0.149 V and P_c2 at −0.074 V, respectively (curve b). If 6.0 μM Mel was added, a pair of new peaks appeared between P₁ and P₂ (P_a at 0.054 V and P_c at −0.079 V, curve c). Though these new redox peaks are related to Mel concentration, they cannot be applied for Mel analysis due to the interference originating from the strong faradaic response of free copper ions. Furthermore, their peak position is another disadvantage for correct quantitative analysis because of the partial overlap with copper peaks.

Fig. 1 (A) Cyclic voltammograms (CVs) measured at the MWCNTs modified electrode in the absence (a) or the presence of 40 μM CuCl₂ (b) or 40 μM CuCl₂ + 6.0 μM Mel (c) in 0.10 M pH 7.0 phosphate buffer containing 0.10 M KCl. (B) CVs in the absence (a) or the presence of 40 μM CuCl₂ (b) or 40 μM CuCl₂ + 2.5 μM Mel (c) in 0.10 M pH 10.0 borate buffer containing 0.10 M KCl. Scan rate: 100 mV s⁻¹.

Consequently, avoiding peak overlapping and decreasing the intensity of Cu peaks become the crux of the matter. Based on our experiments, under weak alkaline condition (0.10 M pH 10.0 borate buffer containing 0.10 M KCl), all of redox peaks in this system shift toward negative potential direction while the change of the Cu–Mel peak potentials is relatively low, with the Cu–Mel oxidation peak appearing at 0.071 V (curve c, Fig. 1B). Furthermore, the peak intensity of the Cu peak decreases dramatically.

However, such a condition still has two problems of stability and sensitivity since copper oxidation/reduction on the surface of modified electrode has a great effect on the electrochemical behavior of the Cu–Mel complex. The inhibition of Cu peaks is necessary for both the formation and the electroactivity of the Cu–Mel complex. Addition of ethanol to the supporting electrolyte solves the above problems perfectly. As shown in Fig. 2, the presence of 20% ethanol (v/v) not only are the Cu peaks inhibited completely, but also the sensitivity of Cu–Mel complex peak is improved greatly. The oxidation peak current was found to be linearly dependent on the Mel concentration ranging from 10.0 to 150 nM (1.2–19 ppb) with a correlation coefficient of 0.998. The detection limit is found to be as low as 2.0 nM (about 0.25 ppb) based on a signal-to-noise ratio of 3. This linear dependency can be used as calibration for high-sensitivity analytical application of Mel. On the other hand, if the reduction peak is chosen for calibration, it should be corrected because this peak overlaps with the reduction peak of the carboxylated MWCNTs.

Fig. 2 CVs in the absence (dashed curve) or the presence (solid curve) of 40 μM CuCl₂ with different concentrations of Mel in 0.10 M pH 10.0 borate buffer containing 0.10 M KCl and 20% ethanol (v/v). The concentration of Mel (nM): (a) 0; (b) 10; (c) 30; (d) 70; (e) 90; (f) 110; (g) 150. Scan rate: 100 mV s⁻¹.

Over the range of 0.005–1 V s⁻¹, the oxidation peak current of Cu–Mel complex increases linearly with the potential scan rate, while the corrected peak current of the reduction peak increases linearly with the square root of the scan rate. According to relevant theories, this means the electrooxidation and electroreduction of Cu–Mel complex are controlled by adsorption and diffusion, respectively. This unusual result show a complicated electrode reaction process. We suppose that the most of Cu(I)–Mel complex are on-site formed on the electrode surface as large amounts of cuprous ions are produced with increasing of potential. A large excess of cuprous ions combined with the MWCNTs microenvironment provide a good platform to efficiently form the Cu(I)–Mel complex. With the potential moving to more positive values, the Cu(I)–Mel complex is oxidized to Cu(II)–Mel followed by its diffusion into bulk solution. In the reverse potential scan, the electrode reaction involving the reduction of Cu(II)–Mel to Cu(I)–Mel is obviously controlled by the diffusion of Cu(II)–Mel towards the electrode surface.

To verify the formula of the complex (1 : 1 or 1 : 2 Cu–Mel) related to the above electrochemical process, a series of CV experiments were performed with various ratios of Cu to Mel. As shown in Fig. 3, when the ratio of Cu to Mel is <1 : 1, a new oxidation peak appears at 0.227 V (P_a′) (curve b) besides the original peak at 0.071 V (P_a, curve c) but when Mel is in excess, P_a′ disappears (curve d). Consequently, it is reasonable to assume that the oxidation peak used to determine Mel (at large excess of Cu) is from the 1 : 1 Cu–Mel complex in this system. Meanwhile, the low electrochemical responses shown in Fig. 3 also reveal that a large excess of copper should facilitate the formation of the 1 : 1 complex.

Fig. 3 CVs for different ratios of CuCl₂ to Mel. (a) 60 μM Mel; (b) 20 μM CuCl₂ + 40 μM Mel; (c) 30 μM CuCl₂ + 30 μM Mel; (d) 40 μM CuCl₂ + 20 μM Mel. Other conditions as in Fig. 2.

In our selected condition of 0.10 M pH 10.0 borate buffer solution containing 0.10 M KCl and 20% ethanol (v/v), there is a weak interaction between copper ion and Mel, as can be observed from UV spectrum that shows only about 3 nm red shift (from 204 to 207 nm) upon addition of Cu salt to Mel solution. There are three principal IR spectrum absorption regions for Mel: 2800–3500, 1440–1650 and 800–1050 cm⁻¹ with the 1440–1650 cm⁻¹ region corresponding to ring distortion and N–H bending. Downward shifts are observed for the above absorption regions indicating interaction of copper ion with Mel.

To obtain the exact composition of Cu–Mel complex, MS/MS spectrum for different ratios of Cu to Mel were recorded. As shown in Fig. 4A, when [Mel] > [Cu] (100 μM CuCl₂–200 μM Mel), the principal composition of complex is CuCl₂(Mel)₂H₂O with m/z 404. In contrast, if the amount of [Cu] > [Mel], as in our selected electrochemical detection condition, CuCl₂Mel(H₂O)₂ dominates with m/z 296 (Fig. 4B). Protonated Mel is observed well above the noise level at m/z 127 from both spectra. The insets show the molecular structures corresponding to the two complexes. The coordination of Mel in both of these complexes occurs through one aromatic nitrogen atom without evidence of chelation to the amine nitrogen atoms.¹⁴

Fig. 4 (A) MS/MS spectrum for [CuCl₂(Mel)₂(H₃O)]⁺. Spray solvent: 100 μM CuCl₂ + 200 μM Mel. (B) MS/MS spectrum for [CuCl₂(Mel)(H₂O)(H₃O)]⁺. Spray solvent: 200 μM CuCl₂ + 8 μM Mel.

In order to verify the performance of this Mel electrochemical sensor, it was applied to Mel analysis in raw milk. Amino acids and most of the lactose in milk could be removed through an extraction procedure. First, 5.0 g milk or Mel spiked milk with 0.75 g NaCl was sonicated for about 5 min, and kept at 100 °C for 10 min. Then, the mixture was centrifuged at 10 000 rpm for 10 min, and the filtrate was filtered through a 0.22-μm polytetrafluoroethylene membrane filter. Finally, 5.0 μL of as-prepared sample was diluted 1 : 1000 times stepwise for the following detection. Table 1 shows experimental results for the determination of Mel in raw milk products. Compared with other previous methods, this approach has higher sensitivity and simplicity.

Table 1 Experimental results for the determination of Mel in raw milk

No.	Mel spiked/ppm	Mel found/ppm	Recovery (%)
1	—	1.6	—
2	2.0	1.9	95
3	4.0	3.6	90

---

Based on an electroactive Cu–Mel complex, a rapid, simple and sensitive electrochemical sensor has been developed for Mel detection. It allows a detection concentration as low as 0.25 ppb to be achieved and provides a new method for analysis of non-electroactive species by using an electrochemical method. Meanwhile, it shows that the utilization of cheap and simple electrochemical instruments in the field of public-health risks of food is quite promising.

This work was supported by the National Natural Science Foundation of China (NSFC, No. 20675005; 20975003 and 20735001). The authors would like to thank Prof. Hai Luo for valuable discussion.

Notes and references

1. L. Zhu, G. Gamez, H. W. Chen, K. Chingin and R. Zenobi, Chem. Commun., 2009, 559–561.
2. E. Chan, S. M. Griffiths and C. W. Chan, Lancet, 2008, 372, 1444–1445.
3. P. C. Bardalaye, W. B. Wheeler and C. W. Meister, J. Assoc. Off. Anal. Chem., 1987, 70, 455–457.
4. L. J. Fremlin and M. Pelzing, LC–GC. N. Am., 2009, 28–29.
5. N. Yan, L. Zhou, Z. F. Zhu and X. G. Chen, J. Agric. Food Chem., 2009, 57, 807–811.
6. H. A. Cook, C. W. Klampfl and W. Buchberger, Electrophoresis, 2005, 26, 1576–1583.
7. R. C. Hirt, F. T. King and R. G. Schmitt, Anal. Chem., 1954, 26, 1273–1274.
8. E. Garber, J. Food Protect., 2008, 71, 590–594.
9. Y. Hase, E. Fukazawa and M. Yamao, Bunseki Kagaku, 1975, 24, 794–797.
10. M. Lin, L. He, J. Awika, L. Yang, D. R. Ledoux, H. Li and A. Mustapha, J. Food Sci., 2008, 73, T129–T134.
11. K. L. Ai, Y. L. Liu and L. H. Lu, J. Am. Chem. Soc., 2009, 131, 9496.
12. V. P. Slisarenko, B. P. Zhantalai and V. T. Chuiko, J. Anal. Chem. USSR, 1977, 32, 1795–1799.
13. P. Li, Y. Huang, S. Xiang and G. Wu, Chin. Chem. Anal., 1993, 29, 270–272.
14. A. B. Wiles, D. Bozzuto, C. L. Cahill and R. D. Pike, Polyhedron, 2006, 25, 776–782.
15. C. Chen, C. W. Yeh and J. D. Chen, Polyhedron, 2006, 25, 1307–1312.
16. D. Goodgame, I. Hussain, A. White and D. J. Williams, J. Chem. Soc., Dalton Trans., 1999, 2899–2900.
17. Y. Zhao, H. Liu, Y. Kou, M. Li, Z. Zhu and Q. Zhuang, Electrochem. Commun., 2007, 9, 2457–2462.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section, UV-Vis and IR spectra. See DOI: 10.1039/b924355k

---