Capillary electrophoresis combined with accelerated solvent extraction as an improved methodology for effective separation and simultaneous determination of malachite green, crystal violet and their leuco-metabolites in aquatic products

Abstract

Malachite green (MG), crystal violet (CV) and their leuco-metabolites (LMG, LCV) could not be determined simultaneously by capillary zone electrophoresis because of their similar structures, low solubility in water, and easy degradation/de-methylation. A novel capillary electrophoresis (CE) method with UV detection for the effective separation and simultaneous determination of the four compounds was developed for the first time. The use of an acetonitrile–sodium dodecyl sulfate (SDS)–ammonium acetate buffer solution (pH 4.0) achieved the effective separation of the four compounds. The effects of CE conditions on separation were investigated in detail. Accelerated solvent extraction (ASE) with acetonitrile as an extractant in the presence of hydroxylamine hydrochloride and p-toluenesulfonic acid was used to extract the four compounds in aquatic products with high extraction efficiency. Under optimized separation conditions, the four analytes could be baseline separated, and there was no interference in analysis of real samples. Calibration curves have good linearity with a correlation coefficient (r) greater than 0.999. The limit of detection for MG, CV, LMG, and LCV were 0.39, 0.29, 0.08, and 0.10 μg mL⁻¹, respectively. The intra- and inter-day variability (RSD) of peak areas were in the range of 0.3–3.4% and 1.0–4.7%, respectively. Average recoveries were 92.6–107% with RSD of 2.1% for shrimp samples, 95.4–108% with RSD of 2.3% for eel samples, and 89.1–105% with RSD of 3.1% for sardine samples. The proposed ASE–CE method has the advantages of high selectivity, good linearity and precision.

---

Introduction

Malachite green (MG) and crystal violet (CV) are triphenylmethane dyes. There have been many reports of the inappropriate use of MG and CV as veterinary drugs. They are readily absorbed into fish tissue from water exposure, and are reduced metabolically by fish to leucomalachite green (LMG) and leucocrystal violet (LCV). MG and CV have been banned for use as fungicides and antiseptics in aquaculture and fisheries because of their carcinogenic and mutagenic properties.¹ Thus, it is necessary to develop a sensitive, rapid, inexpensive, and reliable method for the determination of MG, CV, and their leuco-metabolites in aquatic products.

Spectrophotometry has been used for the determination of MG and CV in water.^2,3 High-performance liquid chromatography (HPLC) has been applied for the determination of MG, CV, LMG, and LCV in water,⁴ and in aquatic products with MS or MS/MS detection.^5–-8 The advantages of capillary electrophoresis (CE) over LC are low solvent consumption and low cost. Capillary zone electrophoresis (CZE) has been primarily carried out using an aqueous electrolyte buffer. CZE and micellar electrokinetic chromatography (MEKC) combined with Raman spectroscopy have been used for the determination of MG in water based on the stacking and sweeping modes.⁹ A CE method combined with conductive detection has been established for the determination of MG in fish with an limit of detection (LOD) of 1.8 μg mL⁻¹.¹⁰ It is difficult to separate MG, LMG, CV, and LCV by CZE because of their similar structures, low solubility in water, and the easy degradation/de-methylation of MG and CV. The use of organic solvents in place of aqueous systems may provide a means of selectivity enhancement. Matysik described the potentials and application of electrochemical detection in conjunction with NACE using an acetonitrile-based buffer system for the separation and detection of MG, CV and rhodamine B.^11,12 To our knowledge there has been no report on the simultaneous determination of MG, CV, LMG, and LCV by CE.

The purpose of the present work was to develop a new CE-UV detection method for the effective separation and simultaneous determination of MG, CV, LMG, and LCV. In this work, the conditions of separation were investigated and optimized. The CE method coupled with accelerated solvent extraction (ASE) has been used for the simultaneous determination of MG, CV, LMG, and LCV in aquatic products with satisfactory results.

Experimental

Instrumentation

All experiments were performed with an Agilent 3D CE system with air-cooling and a diode-array detector (Agilent, Waldbronn, Germany). Data were collected with the Agilent Chemstation version A.10.02 chromatographic data system. A 48.5 cm (40.0 cm to the detector) long, 75 μm i.d. and 365 μm o.d. uncoated fused silica capillary (Yongnian Optical Fabric Factory, Handan, China) was utilized. The extraction equipment was an APLE 2000 automatic accelerated solvent extraction apparatus (Beijing Titan Instruments Co Ltd., China). A centrifuge TGL-16M (Xiangyi centrifuge Co., Hunan, China), RE-2000A rotary evaporator (Shanghai Yarong Biochemistry Instrument Co.,) and PHS-3C pH meter (Shanghai Precision & Scientific Instrument Co., Shanghai, China) were used in sample treatment.

Chemicals and solutions

MD, CV, LMG and LCV (purity: ≥ 99%) were purchased from Dr Ehrenstorfer GmbH (Hamburg, Germany). All chemicals used were of analytical grade.

The stock standard solutions of MG, LMG, CV and LCV, 1000 mg L⁻¹, were prepared by dissolving the solid chemicals in acetonitrile. These stock solutions were stored in the dark at −20 °C and were stable for at least one month. Fresh working standard solutions were prepared daily by diluting the stock solutions with a 1 : 4 (v/v) mixture of acetonitrile and water, and were used for different studies.

The buffers were prepared from ammonium acetate and the pH was adjusted to 4.0 with acetic acid. Sodium dodecyl sulfate (SDS) was used as a surfactant and p-toluenesulfonic acid was used as the organic modifier. Doubly deionized water was used throughout. Buffer solutions were prepared weekly, and filtered through 0.22 μm cellulose acetate filters (Shanghai Xinya Purification Material Factory) prior to injection.

Sample preparation

Shrimp, eel and sardine samples were purchased from a local market, and after being homogenized in a high-speed food blender, they were stored below −20 °C in a freezer until the time of analysis.

Samples were extracted with APLE 2000 automatic accelerated solvent extraction apparatus. Approximately 2.000 g of the blank/spiked sample material was mixed with 2.5 g of diatomite and packed into an 11 mL stainless-steel extraction cell. Each cell was locked with stainless-steel screw caps, and circular glass microfiber filters of 1.98 cm diameter were placed above and below the packing, where diatomite was used as a filter agent to prevent the fine powder breakthrough into the collection bottle. 0.5 mL of 20% (m/v) hydroxylamine hydrochloride, 100 μL of p-toluenesulfonic acid, and 2 mL of acetonitrile were added to each cell. The solvent selected was acetonitrile. Conditions used in the extraction were: oven temperature of 60 °C with 3 min heat-up time under a pressure of 10.3 MPa and two static cycles with a static time of 5 min. The flush volume was 40% of the extraction cell volume. The extract was purged from the sample cell using pressurized nitrogen for 2 min.

Finally, each resulting extract was evaporated to dryness using the rotary evaporator. The residue was rinsed with 5 mL of acetonitrile, and evaporated to dryness under a nitrogen flow at 40 °C. The residue was re-dissolved in 1 mL of buffer solution. The solution was vortexed, and filtered through a 0.22 μm mixed cellulose ester membrane (Shanghai, China) for CE analysis.

Electrophoresis conditions

Prior to its use every day, the capillary was consecutively flushed with 0.1 M NaOH for 10 min, distilled water for 10 min and the running buffer for 10 min. After each analysis run the capillary was flushed for 3 min with the running buffer to maintain the reproducibility of the analysis. Sample introduction was carried out at the anodic capillary side using 50 mbar pressures for 5 s. The high-voltage power supply was set at 23 kV. The 50 mM ammonium acetate solution (pH 4.0) containing 50 mM SDS–acetonitrile (1 : 2, v/v) was used as a running buffer. The capillary temperature was kept at 25 °C. UV detection was employed at a wavelength of 580 nm for MG and CV, and 265 nm for LMG and LCV.

Results and discussion

Optimization of electrophoresis conditions

Effects of electrolyte. The selection of suitable electrolytes is very important for CE. The electrolyte cation also has a clear effect on the capillary electrophoretic selectivity in nonaqueous media. The initial experiments were conducted using trihydroxymethyl aminomethane, sodium tetraborate, potassium biphthalate, and ammonium acetate as electrolyte, respectively (Fig. 1, left). The results showed that ammonium acetate could separate the four compounds. In conclusion, among the electrolyte systems studied, a mixture of acetonitrile–ammonium acetate was the most favourable medium for CE-DAD, taking into consideration its solubilizing power and the reliability of CE performance. All subsequent experiments were done using this electrolyte system. The effects of ammonium acetate concentration from 20 to 80 mM on migration time and peak area were tested (Fig. 1, right).

Fig. 1 Effect of different electrolytes (left) and ammonium acetate concentration (right) on separation: a, trihydroxymethyl aminomethane; b, sodium tetraborate; c, potassium biphthalate, and d, ammonium acetate; 1, MG; 2, CV; 3, LCV, and 4, LMG; each compound, 30 μg mL⁻¹ ; buffer, pH 4.0; buffer–acetonitrile (v/v), 1 : 2; SDS, 50 mM; separation voltage, 23 kV; temperature, 25 °C.

The migration time and peak area of the analytes increased with an increase of electrolyte concentration. This is because the thickness of the electric double layer between the capillary walls and the buffer solution decreased with an increase of buffer concentration. Also, high buffer concentration could increase the coverage of sites on the silica surface, resulting in a decrease of the electroosmotic flow because of the decrease of the silicon hydroxyl dissociation.¹⁸ Synthetically considering migration time and response, a 50 mM ammonium acetate buffer was selected to give high sensitivity and faster analyses.

Effect of running buffer pH. The pK_a of MG and CV were 10.3 and 9.3, respectively. LMG and LCV are alkaline due to the presence of two basic amines, but there was no available information on the pK_a. Using aqueous electrolytes, the separation of these four compounds could not be achieved over a broad pH range. The pK_a values for these compounds in organic solvents can be significantly different from those in water. The pH value of the acetonitrile–buffer solution mainly influenced the resolution and the apparent mobility of the target analytes. The four compounds were protonated easily in acidic medium. The effect of pH values from 3.5 to 4.5 on the separation of the four compounds was investigated, as shown in Fig. 2.

Fig. 2 Effect of running buffer pH on separation: 1, MG; 2, CV; 3, LCV and 4, LMG; each compound, 30 μg mL⁻¹; other conditions were as shown in Fig. 1.

It is shown that the pH of the running buffer affects the resolution of the four compounds. Under higher acidity (pH 3.50 and 3.75), MG, CV, and LCV could not be separated, and had a low response. The resolution and migration time of CV, LCV and LMG increased with an increase in pH. When pH = 4.5, the migration time of LMG was up to 10 min. When pH = 4.0, the separation of the four analytes was achieved within 8 min, with highest sensitivity.

Effect of organic additive on separation. In our initial test, it was proved that CZE with ammonium acetate buffer (pH 4.0) could not be used for the effective separation and simultaneous detection of MG, LMG, CV, and LCV. Under the same conditions, methanol or acetonitrile as an additive in the ammonium acetate buffer was used for comparing its effect on the separation and determination. The result is shown in Fig. 3(left).

Fig. 3 Effect of organic additive on separation buffer: 50 mM, pH 4.0; SDS: 50 mM; separation voltage, 23 kV; temperature, 25 °C; 1, MG; 2, CV; 3, LCV and 4, LMG; each compound, 30 μg mL⁻¹; right: buffer–acetonitrile (v/v), 1.4 to 2 : 1.

The use of an acetonitrile/methanol–ammonium acetate buffer (pH 4.0) is attractive because the four compounds show much better solubility in this medium. The organic additives affect migration of analytes in two basic ways, directly altering physicochemical properties of the separation medium (e.g. viscosity and dielectric constants) and invoking chemical equilibria involving the analytes and a buffer additive.¹⁴ Using methanol as an additive, the separation of the four compounds was achieved in 12.6 min, whereas with acetonitrile, the separation could be achieved within 8 min. This is due to the lower viscosity of acetonitrile than methanol–water mixtures and the higher dielectric constant (ε = 37.5) and autoprotolysis constant (pK_auto ≥33.3) of acetonitrile than those (ε = 32.7, pK_auto = 17.2) of methanol.¹⁵ Using acetonitrile with lower viscosity, higher dielectric constant, and autoprotolysis constant, better separation was achieved compared with that achieved using methanol.

The optimal proportion of acetonitrile in ammonium acetate buffer was investigated further, as shown in Fig. 3 (right). The analysis times decreased drastically with addition of acetonitrile into buffer due to the lowered viscosity. When the relative amount of acetonitrile was increased, effective separation of the four compounds could not be achieved. When the volume ratio of the buffer solution and acetonitrile was selected to be 1 : 2, separation was achieved, with higher response and resolution.

Effect of surfactant SDS. This work employed anionic surfactant SDS in the acetonitrile–ammonium acetate buffer system. The effect of SDS on the separation of the four compounds is shown in Fig. 4 (left).

Fig. 4 Effect of surfactants on separation: a, without SDS, b, with SDS; buffer: 50 mM, pH 4.0; buffer–acetonitrile (v/v), 1 : 2; separation voltage, 23 kV; temperature, 25 °C; 1, MG; 2, CV; 3, LCV and 4, LMG; each compound, 30 μg mL⁻¹.

It is evident that the use of SDS is beneficial for the separation of the analytes based on interactions between negatively charged surfactant monomers and analytes. The presence of SDS allowed the surfactant–analyte complex to migrate towards the anode,¹⁶ so that the analytes were separated in a reasonable time period. In the case of LCV and LMV some change in the peak shape after addition of SDS was observed, this is possibly due to the solubilization of poorly soluble analytes.

Fig. 4 (right) illustrates the relationship between the response of the four analytes and the concentration of the surfactants. When SDS concentrations varied from 20 to 80 mM, the resolution did not change and the migration time and peak area of the analytes increased. This indicates the occurence of stronger interactions between the analytes and the surfactants. Walbroehl and Jorgenson¹⁷ suggested a dynamic equilibrium mechanism between the associated and the unassociated forms of the analytes to account for their migration. This mechanism is also consistent with our results, as indicated by the steadily increasing peak height and peak area in Fig. 4 (right). When SDS concentration was selected to 50 mM, satisfactory separation was achieved within 7 min with higher sensitivity.

Effect of separation voltage and temperature. With an increase of separation voltage the EOF increased and the analytical time decreased. However, the high voltage led to Joule heating and affected the separation of the analytes. The effect of voltage on the separation of the drugs was investigated using acetonitrile–buffer. By increasing the separation voltage from 20 kV to 30 kV, the migration times of the drugs decreased due to the increase of both EOF and mobility of the compounds. However, the higher voltage caused a decrease in the peak areas of the analytes, and resulted in a higher current thereby leading to more Joule heating, which produces a radial temperature gradient and decreases separation efficiency. When the voltage was set at 30 kV, the current was interrupted. A voltage of 23 kV was selected in this work to give higher sensitivity and a shorter analytical time. The effect of temperature was also studied and it was found that when the capillary temperature was decreased from 20 °C to 30 °C the migration time decreased obviously for MG and CV, and for LMG and LCV slightly. The column temperature was set at 25 °C for separation in this work.

Optimization of ASE conditions

Effect of extraction solvent. In the presence of light, heat and oxygen, MG and CV are susceptive to demethylation and oxidative degradation. The degradation of MG and LMG can occur dramatically during the extraction procedure, especially for LMG. And this degradation is most probably due to de-methylation of the dye. In this work, it was demonstrated that addition of hydroxylamine hydrochloride as a reducer to the samples could prevent the oxidative degradation of MG and CV in the extraction process, and addition of p-toluenesulfonic acid (ion pair reagent) could largely reduce MG degradation and increase recovery in all tested tissues. From the results obtained 0.5 mL of 20% (m/v) hydroxylamine hydrochloride and 100 μL of p-toluenesulfonic acid were selected as the optimal dosage.

Methanol and acetonitrile have frequently been used as extractants in ASE.¹³ The system of McIlvaine buffer (pH 3)–acetonitrile has been used for ASE of MG, LMG, GV and LGV prior to LC–MS/MS with recovery of 82–93.5%. In this work, the extraction efficiency with methanol and acetonitrile in the presence of hydroxylamine hydrochloride and p-toluenesulfonic acid was compared. The recoveries of MG, LMG, GV and LGV from the blank shrimp sample spiked at 5 mg kg⁻¹ for each analyte were 93–106% with acetonitrile and 74–110% with methanol (see Fig. 5). Hence, 2 mL acetonitrile–0.5 mL 20% (m/v) hydroxylamine hydrochloride–100 μL p-toluenesulfonic acid was selected for futher work.

Fig. 5 Effects of different solvents and static pressure on extraction efficiency spiked at 5 mg kg⁻¹ for each analyte; acetonitrile/methanol volume, 2 mL; temperature, 60 °C; static time, 5 min; static cycles, 2; 1, MG; 2, CV; 3, LCV and 4, LMG; extraction efficiency data are the average value of three measurements and are less than 3% of RSD.

Effect of temperature and pressure. Temperature and pressure are the most important parameters for ASE. The extraction efficiency with acetonitrile as an extraction solvent for simultaneous extraction of the four compounds from the spiked blank shrimp was compared at different temperatures (from 40 to 80 °C) and pressures (from 5 to 15 MPa). The extraction efficiency of CV, LCV and LMG decreased slightly with an increase in temperature from 60 to 80 °C. At a temperatures of 60 °C, extraction efficiencies of 105, 104, 95 and 100% were acheived for MG, CV, LCV and LMG, respectively. The pressure test showed that at a temperature of 60 °C and a pressure of 10 MPa the highest extraction efficiency was achieved, equal to 105% for MG, 95% for CV, 98% for LCV, and 100% for LMG (see Fig. 5).

Effect of extraction time. The extraction process can be conducted in a static mode. The critical factors are the temperature and time of the extraction. During this process, the analytes are isolated from the sample under stable static conditions. The static process can be repeated several times if low recoveries are obtained in a single stage. In this work, the efficiency of the extraction from the spiked blank shrimp sample was investigated under a pressure of 10 MPa and a temperature of 60 °C using a static time of 3, 5, and 8 min and two cycles. The results showed that significant amounts of the analytes were found with 5 min of static time in the first extract. Two extraction cycles were used with extraction efficiency of 105% for MG, 102% for CV, 95% for LCV, and 98% for LMG. So 5 min static times and 2 static cycles were selected in further work.

Effect of flush volume. Flush volume is the percentage of fresh volume introduced into the cell after the static time to drag the analytes towards the collection vial. This volume ensures that all analytes are eluted and is closely related to the final volume. Different flush volumes were used to extract analytes. To save solvent and time, a flush volume of 40% (cell volume, 11 mL) was enough to push the analytes extracted out of the cell with higher extraction efficiency, being 103% for MG, 98% for CV, 94% for LCV, and 100% for LMG.

Performance of the method

Selectivity. Under the optimized conditions, selectivity was determined for each analyte in the assay. The electropherograms of blank shrimp sample and spiked shrimp sample were obtained, as shown in Fig. 6.

Fig. 6 Electropherograms of blank samples and spiked samples: a, shrimp, b, eel and c, sardine; spiked level: 1, 1.5 mg kg⁻¹ MG; 2, 0.5 mg kg⁻¹ CV; 3, 0.5 mg kg⁻¹ LCV, and 4, 0.5 mg kg⁻¹ LMG.

It was shown that the four analytes could be baseline separated. The unknown peak does not interfere with the separation and determination of MG, CV, LCV, and LMG. There was no interference peak in real samples.

Linearity and detection limit. The linearity for analysis of the four analytes was evaluated with concentrations of calibration standards (5 concentration points) against measured peak areas under the optimized conditions. The equations for the calibration curves obtained based on three parallel measurements of standard solution are listed in Table 1. It can be seen that the linearity is satisfactory with a correlation coefficient (r) greater than 0.999.

Table 1 Linear equation, relative coefficient, linear range, instrument LOD and LOQ

Analyte	Linear equation	Correlation coefficient (r)	Linear range/μg mL^−1	LOD/μg mL^−1	LOQ/μg mL^−1
MG	A = 7.5534C ¬ 26.4239	0.9995	1.32–300	0.39	1.32
CV	A = 10.1398C + 1.5729	0.9999	0.98–100	0.29	0.98
LMG	A = 7.3162C ¬ 10.9297	0.9997	0.27–100	0.08	0.27
LCV	A = 5.9098C ¬ 8.5336	0.9998	0.34–100	0.10	0.34

---

The LOD was determined as the sample concentration that produces a peak with a height three times the level of the baseline noise, and the limit of quantification (LOQ) was calculated as the level that produced a peak with 10 times the signal-to-noise ratio. Table 1 gives the instrument LODs and LOQs. The LOD values were lower than those of the CE method.^9–12 For 2 g of sample and 1 mL of final test solution, the method LOQ values for MG, CV, LCV, and LMG were 0.66, 0.49, 0.14, and 0.17 mg kg⁻¹ in aquatic products, and could be improved by increasing the sample volume.

Repeatability. The precisions of the method were investigated at room temperature (25 °C) by analyzing the four analytes in spiked blank shrimp samples. Under the optimized conditions the intra-day variability (RSD) of peak area for 5 determinations and inter-day RSD for 3 determinations for each day within 3 days were investigated. The results are listed in Table 2.

Table 2 Intra- and inter-day precision as relative standard deviation (RSD)

Analyte	Added/mg kg^−1	Intra-day (n = 5)	Inter-day (n = 15)
		RSD (%)	RSD (%)
MG	1.5	1.0	2.0
	7.5	1.6	4.0
	15.0	3.4	4.7
CV	0.5	1.1	1.4
	2.5	1.9	2.1
	5.0	0.9	3.0
LMG	0.5	0.5	1.0
	2.5	1.6	2.0
	5.0	1.5	2.9
LCV	0.5	0.5	1.2
	2.5	1.0	1.7
	5.0	1.7	2.6

---

The intra- and inter-day RSDs of the peak areas for the four analytes at near LOQ level were in the range of 0.3–3.4% and 1.0–4.7%, respectively. It was shown that the repeatability of the method is satisfactory for the residue determination of the studied compounds in aquatic products.

Sample analysis

Under the optimized conditions of ASE and CE, shrimp, eel and sardine samples were analyzed. The studied four dyes were not detected in real samples. The recovery test of the assays for the analytes was determined by adding known amounts of these dyes to the aquatic product samples. The recoveries are listed in Table 3.

Table 3 Determination of the MG, CV, LMG and LCV in shrimp, eel, and sardine spiked at three concentration levels for each analyte

Sample	Analyte	Spiked/mg kg^−1	Recovery (%)	Mean recovery (%)
Shrimp	MG	1.5, 7.5, 15	100, 94.0, 103	99.0
	CV	0.5, 2.5, 5.0	91.3, 94.0, 97.3	94.2
	LMG	0.5, 2.5, 5.0	110, 109, 102	107
	LCV	0.5, 2.5, 5.0	104, 87.3, 86.4	92.6
Eel	MG	1.5, 7.5, 15	103, 105, 111	108
	CV	0.5, 2.5, 5.0	92.0, 104, 101	99.0
	LMG	0.5, 2.5, 5.0	108, 100, 101	103
	LCV	0.5, 2.5, 5.0	99.2, 94.1, 93.0	95.4
Sardine	MG	1.5, 7.5, 15	97.3, 89.4, 94.1	93.6
	CV	0.5, 2.5, 5.0	85.7, 87.9, 93.8	89.1
	LMG	0.5, 2.5, 5.0	104, 106, 97.0	105
	LCV	0.5, 2.5, 5.0	98.0, 90.3, 92.2	93.5

---

The average recovery was 92.6–107% with an RSD of 2.1% for shrimp samples, 95.4–108% with an RSD of 2.3% for eel samples, and 89.1–105% with an RSD of 3.1% for sardine samples. It is indicated that the proposed method has high recovery, and can be used for the determination of the studied drugs in real samples.

Conclusion

The proposed CE method can effectively separate and simultaneously determine MG, CV, LMG, and LCV, which have similar structures. ASE with acetonitrile as an extraction solvent in the presence of hydroxylamine hydrochloride and p-toluenesulfonic acid is an effective tool for the extraction of the four compounds in aquatic products. The proposed ASE–CE method has the advantages of high selectivity, good linearity and precision.

Acknowledgements

This work was supported by the Natural Science Foundation of Hebei Province (B2008000583) and the Sustainable Plan of Science and Technology of Hebei Province of China (No. 10967126D).

References

1. Council Directive 96/23/EC of 29 April 1996, Brussels, (23 May 1996), Official Journal Europe Communication L125, 10.
2. I. Šafařík and M. Šafaříková, Water Res., 2002, 36, 196–200.
3. L. An, J. Deng, L. Zhou, H. Li, F. Chen, H. Wang and Y. T. Liu, J. Hazard. Mater., 2010, 175, 883–888.
4. Z. Zhang, K. Zhou, Y. Q. Bu, Z. J. Shan, J. F. Liu, X. Y. Wu, L. Q. Yang and Z. L. Chen, Anal. Methods, 2012, 4, 429–433.
5. Y. F. Tao, D. M. Chen, X. Q. Chao, H. Yu, Y. H. Pan, Z. L. Liu, L. L. Huang, Y. L. Wang and Z. H. Yuan, Food Control, 2011, 22, 1246–1252.
6. M. Dowling, R. Duffy, D. G. Smyth, P. P. Mulder, C. Duffy, L. Regan and M. R. Smyth, Anal. Chim. Acta, 2007, 58, 411–419.
7. X. Wu, G. Zhang, Y. Wu, X. Hou and Z. Yuan, J. Chromatogr., A, 2007, 1172, 121–126.
8. G. Chen and S. Miao, J. Agric. Food Chem., 2010, 58, 7109–7114.
9. C. H. Tsai, J. D. Lin and C. H. Lin, Talanta, 2007, 72, 368–372.
10. B. M. Huang, C. W. Yao and H. Cheng, Chin. J. Appl. Chem., 2007, 24, 327–330.
11. F. M. Matysik, J. Chromatogr., A, 1998, 802, 349–354.
12. F. M. Matysik, Electrochim. Acta, 1998, 43, 3475–3482.
13. H. W. Sun, X. S. Ge, Y. K. Lv and A. B. Wang, J. Chromatogr., A, 2012, 1237, 1–23.
14. Q. Yang, L. M. Benson, K. L. Johnson and S. Naylor, J. Biochem. Biophys. Methods, 1999, 38, 103–121.
15. M. L. Riekkola, M. Jussila, S. P. Porras and I. E. Valko, J. Chromatogr., A, 2000, 892, 155–170.
16. J. Li and J. S. Fritz, Electrophoresis, 1999, 20, 84–91.
17. Y. Walbroehl and J. W. Jorgenson, Anal. Chem., 1986, 58, 479–481.
18. N. Chen, L. Wang and Y. K. Zhang, J. Microcolumn Sep., 1995, 7, 193–198.

---