Spectrophotometric determination of melamine in milk by rank annihilation factor analysis based on pH gradual change-UV spectral data

A new spectrophotometric method has been developed in this paper to determine melamine in milk by applying rank annihilation factor analysis (RAFA) based on pH gradual change-UV spectral data (pH-spectra). In the proposed method, the spectra of the sample solutions at different pH data points were recorded and the pH-spectra bilinear data matrix was generated. Based on these data, the RAFA was then applied to calculate the concentration of melamine in milk. The experiments have been conducted and the results were satisfactory. Under the optimised conditions, linearity of the proposed method was in the range of 0.04–4.0 μg mL −1 for calibration samples, and 0.04–3.5 μg mL −1 for the mixed solutions of melamine with the background milk components. The detection limit (DL) was 12 ng mL −1 . The relative predictive error (RPEs) and root mean square error of prediction (RMSEP) of applying RAFA were 0.91% and 0.0151, respectively. Keywords Melamine pH-spectra Rank annihilation factor analysis Milk 1 Introduction Melamine (1,3,5-triazine-2,4,6-triamine, Fig. 1 a) is a nitrogen heterocycle-based organic chemical, and mainly used to produce melamine–formaldehyde resin. The resin is water, heat and arc resistant, and can be used in decorative laminates, amino plastics, adhesive agents and coatings. It can also be used as a paper reinforcing agent, textile auxiliaries and leather retaining agent, etc. ( Rima, Abourida, Xu, Cho, & Kyriacos, 2009; Tsai, Sun, Liao, Lin, & Kuo, 2009; Wang, Jiang, Chu, & Ye, 2009 ). Melamine contains a substantial amount of nitrogen—66.7% by mass. Because of this property, it has been illegally used as the food additive to increase the apparent protein content in food products. Kjeldahl test is the traditional method to measure the protein content. However, Kjeldahl test determines the protein content only by analysing total nitrogen content, without identifying its sources. Melamine has low oral acute toxicity. But the chronic administration of high concentrations can induce renal pathology and even death, especially in babies and children ( Guan, 2009 ). Melamine in combination with cyanuric acid is able to form insoluble melamine cyanurate crystals in kidneys, causing renal failure. Melamine-contaminated pet food ingredients were reported to cause renal disease and deaths of cats and dogs in the USA in 2007. Later, melamine-contaminated infant formula milk powder and eggs emerged in September 2008 and have raised global concerns about melamine ( Chen, 2009 ). Therefore, it is imperative to develop sensitive and reliable analytical methods to determine melamine in human food and animal feed products. The common methods for the determination of melamine in feed and food are liquid chromatography (LC) ( Muñiz-Valencia et al., 2008 ), high performance liquid chromatography (HPLC) ( Sun, Wang, Ai, Liang, & Wu, 2009 ), liquid chromatography/tandem mass spectrometry (LC/MS/MS) ( Wu et al., 2009 ), gas chromatography/mass spectrometry (GC/MS) ( Xu et al., 2009 ), and gas chromatography/tandem mass spectrometry (GC/MS/MS) ( Miao et al., 2009 ). In GB/T 22388 (2008) , melamine in milk and milk products are determined by HPLC–UV or DAD, HPLC–MS/MS and GC–MS or GC–MS/MS with quantification limits of 2, 0.01, and 0.05 mg kg −1 , respectively. The methods reported recently use different approaches, such as enzyme-immunoassay ( Kim, Perkins, & Bushway, 2008 ), electrophoresis ( Klampfl, Andersen, Haunschmidt, Himmelsbach, & Buchberger, 2009; Yan, Zhou, Zhu, & Chen, 2009; Zhai, Qiang, Sheng, Lei, & Ju, 2008 ), mass spectrometry coupled with other techniques ( Yang et al., 2009; Zhu, Gamez, Chen, Chingin, & Zenobi, 2009 ), raman spectroscopy ( Cheng et al., 2008; Lin et al., 2008 ), electrochemical sensor ( Cao et al., 2009; Liang, Zhang, & Qin, 2009; Wang et al., 2009 ), sweeping-micellar electrokinetic chromatography ( Tsai et al., 2009 ), and chemiluminescence (CL) ( Yu, Zhang, Dai, & Ge, 2009 ). All these methods mentioned above are not widely available in ordinary analytical laboratories. Therefore, it is still necessary to develop the approaches that can be more easily managed to determining melamine without pre-separation processes. Spectrophotometry is a convenient and widely used method for quantitative analysis because the common availability of the instrumentation, the simplicity of procedures, and precision and accuracy of the technique make spectrophotometric method still attractive ( Afkhami & Khalafi, 2007 ). Jamil Rima et al. developed a new method for the spectrophotometric determination of melamine with improved specificity and selectivity, which based on the Mannich reaction of melamine with formaldehyde and uranine ( Rima et al., 2009 ). Generally speaking, the spectrophotometric method requires the analyte presented in the given sample to have different absorption spectrum from other unknown coexistent substances with low spectral overlapping. If the method is used to determine melamine in milk directly, the problem seems more intractable. Numerous nutrition components in milk-foodstuffs and buffer solutions, etc. could interfere the quantitative determination of melamine. Thus the detection of melamine in milk-foodstuffs is a grey mixture analysis problem (a multicomponent system can be defined as a grey system or grey mixture when the interested analytes for analysis coexist with the interferents and the determination standards are only applicable for the analytes to be determined) ( Yuan, Liao, Lin, Deng, & He, 2008 ). On the other hand, the introduction of computer and statistical techniques into chemistry helps the chemists to tackle this kind of analytical problems for much more complex samples. Chemometrics, which is the result of the combination of chemistry with computer, enable the chemists to resolve the constituents of a complex system without the pre-separation step ( An et al., 2010 ). Rank annihilation factor analysis (RAFA) is an efficient chemometric technique based on rank analysis of two-way spectral data and can be employed to quantitatively analyse the grey system with unknown background. The idea behind RAFA is to reduce the rank of the calibration sample by subtracting the contribution from the analyte of interest. This is feasible because if the signal from the analyte of interest is subtracted from the sample data, then the rank of this matrix will decrease by one since the contribution of the analyte of interest to the rank is one in the case of such ordinary bilinear rank-one data as spectrophotometric or fluorescence data ( Ho, Christian, & Davidson, 1978 ). RAFA had been used for qualitative and quantitative analysis in many fields ( Abdollahi, Safavi, & Zeinali, 2008; Afkhami & Khalafi, 2007; Bahram & Mabhooti, 2009 ). However, to the best of our knowledge, there is not any report in the literature so far concerning the determination of melamine in grey mixture using the difference of absorption spectra of the analyte at different pH values. In this work, we introduce a simple, sensitive, selective and low cost procedure for direct spectrophotometric determination of melamine in milk samples by applying RAFA to pH gradual change-UV spectral data (pH-spectra). Melamine is a weak base with a pKa of 5.05, and is readily protonated in aqueous solution at pH lower than 5.0 ( Liang et al., 2009; Muñiz-Valencia et al., 2008 ). The absorption spectra of the solution containing the analyte vary with the change of the pH value of the solution. We recorded the changing spectra and obtained the bilinear matrix of pH-spectra. Based on these data, the concentration of melamine in samples could be predicted by RAFA with satisfactory results. 2 Materials and methods 2.1 Instruments A Perkin-Elmer Lambda 45 UV–Vis spectrometer (PerkinElmer, USA) was used for recording absorbance spectra using 1 cm path length quartz cells. A Professional Metre PP-15 pH-metre (Sartorius, Germany) with a combined glass electrode was used for pH measurements. The spectral data were imported to the Matlab environment and all programs were run on personal computers. 2.2 Chemicals and reagents All chemicals used in the experiments were of analytical grade and used as received. All solutions were prepared with doubly distilled water. Melamine, acetonitrile, phenylalanine, tryptophane, methionine, trichloroacetic acid (TCA), lead acetate, methanol, phosphoric acid, disodium hydrogen phosphate and sodium phosphate monobasic were purchased from Shanghai Chemical Third Factory, China. Milk analysed in this work was obtained from local supermarket. 2.3 Standard solutions The stock solutions of melamine (500 μg mL −1 ), phenylalanine (500 μg mL −1 ), tryptophane (500 μg mL −1 ), and methionine (500 μg mL −1 ) were prepared according to the conventional method with 50% (v/v) acetonitrile solvent and stored at 4 °C. Standard working solutions were prepared by appropriate dilution of the stock solution with 50% (v/v) acetonitrile as the diluter. A set of 0.05 mol L −1 phosphate buffer solutions (PBS) with different pH values ranging from 2 to 10 were prepared from H 3 PO 4 (0.05 mol L −1 ), NaH 2 PO 4 (0.05 mol L −1 ) and Na 2 HPO 4 (0.05 mol L −1 ). 2.4 Sample treatment The liquid milk was pretreated according to the following procedure. Five grams of accurately weighed (precision 0.01 mg) milk was placed into a 50 mL polypropylene centrifuge tube to which 40 mL of extraction solvent (acetonitrile:water in a ratio of 1:1) was added. After 10 min shaking, the mixture was centrifuged at 4000 rpm for 10 min, and then the supernatant was filtrated through a 0.22 μm filter membrane. The filtrate was diluted to 50 mL to obtain the samples for detection. Because the melamine contaminated milk cannot be purchased from the market any more, the milk was spiked with appropriate amounts of melamine standard solution directly. 2.5 Analytical method The samples for analysis were prepared by mixing 1 mL of 0.05 mol L −1 PBS at different pH values with an aliquot of different melamine standard solutions or sample solutions. A right amount of solvent was transferred to each sample to reach a final volume of 5 mL. The final concentrations of melamine varied from 0.04 to 4 μg mL −1 . Subsequently, the sample solution was shaken and transferred into a quartz cell for measurements. The absorbance spectrum was measured between 200 and 300 nm with 1 nm interval against a reagent blank. Then the measured absorbance matrix with the size of 13 × 101 was processed by the programs written in MATLAB 7.0. 3 Results and discussion 3.1 Bilinear analysis of absorption spectra of melamine For a certain amount of melamine, when absorption spectra at n pH points were scanned and the absorbance values at p wavelength points in each spectrum were recorded, the spectra of melamine at different pH values can be digitized and arranged in a data matrix A with the size of n × p , where n denotes the number of spectra and p the number of wavelengths at which the absorbance was recorded: (1) A np = ( a 1 t , a 2 t , … , a i t , … , a n t ) t where a i = [ a i1 , a i2 ,…, a ij ,…, a ip ] is the absorption spectrum vector at pH i , a ij is the absorbance value at wavelength point λ j , the superscript t represents the transpose of a matrix or vector. Melamine is a weak base, presenting the base form and the conjugated acid form in the solution (see Fig. 1 ). Both forms only absorb the ultraviolet light. The absorption coefficient of melamine changes with the distribution fraction of the two forms at different pH. Therefore, a bilinear data matrix of pH-spectra can be established by measuring the absorbance of melamine solutions in the ultraviolet region at different pH values. Typical absorption spectra of melamine at varied pH with the concentration of 1.00 μg mL −1 in 50% (v/v) acetonitrile and 0.05 mol L −1 PBS are shown in Fig. 2 (a). Principal component analysis (PCA) was used to analyse matrix A constructed by data from Fig. 2 (a). The results show that the number of principal components of matrix A is two, which means that either the rank of A is two, or melamine exists at different pH with two forms and the absorption spectra of the two species are different. Thus, a spectrum of melamine in PBS at a certain pH can be considered as the result of the linear combination of the spectrum of the two species. Therefore, the two species of melamine can be called acid-type (A-type) and alkali-type (B-type), respectively, and a spectrum of melamine in a buffer solution can be expressed as the linear combination of a spectrum of A-type with a spectrum of B-type. According to the experiment method, the total absorption of melamine will change gradually with the variation of pH. Let c be the initial concentration of melamine, δ Ai and δ Bi be the distribution fractions of A-type and B-type of melamine at a certain pH, respectively. Thus, the spectrum vector of melamine can be expressed as follows: (2) a i = δ Ai c ε A t + δ Bi c ε B t where ε A = [ ε A 1 , ε A 2 , … , ε Ap ] t and ε B = [ ε B 1 , ε B 2 , … , ε Bp ] t are the absorption coefficient vectors of A-type and B-type of melamine, respectively. For the spectra matrix A , Eq. (1) can be transformed to: (3) A np = A A + A B = δ A c ε A t + δ B c ε B t Where δ A = [ δ A 1 , δ A 2 , … , δ An ] t and δ B = [ δ B 1 , δ B 2 , … , δ Bn ] t are distribution fraction vectors of A-type and B-type of melamine, respectively. A is a bilinear data matrix of n row and p column, that is, a bilinear matrix of pH-spectra. It is a spectra vector for each row, and an acidity vector for each column. From Eq. (3) , it can be seen that both A A = δ A c ε A t and A B = δ B c ε B t are bilinear data matrices with rank = 1, and matrix A with rank = 2 can be expressed as the combination of the two bilinear data matrices. Both δ A and δ B are only the function of pH and are independent of the concentration of melamine. δ A increases as δ B decreases, vice versa. For any pH i , δ Ai + δ Bi = 1 . So there are linear dependencies in the second order data, which means the spectral matrices are rank deficient. RAFA can be used to analyse the second order data matrices, no matter whether the matrices are rank deficient or not, and the algorithm of RAFA is relatively simple to implement. Therefore, RAFA has been selected to analyse the matrices, and the experiment results were satisfactory. 3.2 Selection of buffer solution, pH and wavelength range for building spectra matrices The selection of buffer solution is an important step in pH-spectral bilinear analysis. The buffer solution should have a wide pH range, and its absorption spectra should not overlap with the absorption spectra of the analyte, or the overlapping degree is low. The effects of different buffer solutions have been examined on the absorption spectra of 1.00 μg mL −1 melamine standard solution, including Britton-Robinson (BR) buffer and phosphate buffer, both of which have a wide pH range. It was found from Fig. 3 that the absorption spectra of BR buffer overlap seriously with that of melamine, whilst the spectra of PBS have little effects. Therefore PBS was selected for further studies. It can be observed from Fig. 2 (a) that the absorption spectra of melamine vary significantly as pH varies between 2 and 8, but remains almost unchanged in the range of 8–10 (not shown). Thus, the pH range of 2–8 (2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, and 8.0) was selected. It could also be seen from Fig. 2 (a) that the absorption spectra of melamine mainly change with the change of pH in the wavelength range of 200–300 nm. Normally, the more wavelength points are considered, the more information about melamine can be extracted from the system, and consequently, the more accurate results can be generated. Therefore, the second order spectra data matrices in this work were constructed by selecting the wavelength from the range of 200–300 nm with the increments of 1 nm between two consecutive wavelength points. By doing so, it can be guaranteed that the experimental results have sufficient precision. 3.3 Selection of extraction solvents The effect of trichloroacetic acid (TCA)-lead acetate, 50% (v/v) methanol and 50% (v/v) acetonitrile on protein precipitation was investigated. A mixture of acetonitrile and water (1:1, v/v) has been selected as the extraction solvent not only because of its best protein precipitation effect and high recovery, but also because its compositions are the same as the solvent of stork and working solutions. Therefore, the step of removing the extraction solvent from sample matrix can be omitted (the step is necessary in many other methods ( Sun et al., 2009; Xu et al., 2009 )). Moreover, we found that the extraction efficiency would increase significantly if the samples were well dispersed with appropriate amount of water prior to protein precipitation. 3.4 RAFA algorithms A matrix equation can be established as Eq. (4) after the pH-spectral matrices of the melamine validation sample and the mixture sample are obtained through experiments: (4) A x = β A s + R + E where A s is the second order spectra data matrix of a melamine standard solution, which is the validation sample, A x is the second order spectra data matrix of the melamine-contaminated milk sample. β is the regression coefficient which is the ratio of c x to c s , i.e., β = c x / c s (where c x and c s are the concentrations of melamine in the milk sample and the standard solution, respectively), βA s is the bilinear matrix (with rank = 1) of the analyte in the contaminated milk sample, R is the unknown background interference, which is the linear combination of the measurement matrix of multiple unknown interferences, E is the remnant error matrix. The process of RAFA algorithm can be described as follows. Given a certain range of β , the algorithm analyses the primary components in matrix R = A x − βA s , and extracts m primary components to construct the new matrix R ∗. Then, the algorithm calculates the eigenvalues γ of R∗ for different values of β , until the minimal eigenvalues of R∗ , which corresponds to melamine, is reached for a certain β . The calculated results with RAFA were satisfactory. Fig. 2 (b) shows the spectra of a milk sample containing melamine (1.20 μg mL −1 ). Fig. 4 shows the change of eigenvalues γ of the matrix A x − βA s with β . The curve labelled by γ 1 in Fig. 4 is the one with the most significant eigenvalue that corresponds to melamine. The β value corresponding to the lowest point of the curve is the solution. It can be measured from Fig. 4 that the β -value is 0.6. Because the concentration of the analyte in the calibration solution c s is 2.00 μg mL −1 , the concentration of this analyte in the unknown sample set can be calculated as follows: C x = β C s = 0.6 × 2.00 = 1.20 μ g mL - 1 3.5 Detection range, detection limit and selectivity The detection range of melamine has also been tested. When the relative errors are no more than ±5%, the detection range of the melamine calibration solution and the mixed solution of melamine with milk background components are 0.04–4.0 and 0.04–3.5 μg mL −1 , respectively. The detection limit (DL) of melamine was determined through 11 parallel blank tests with extracted milk background components as reagent blank. If the DL is specified as three times of the standard deviation, then the DL of this method is 12 ng mL −1 . The limit of quantification (LOQ) is 40 ng mL −1 calculated from the 10 times of standard deviation. Normally, when using RAFA it is not necessary to determine what other components and how much of them exist in the grey background. The selectivity of the method is often very reliable. In this work, three common ingredients in milk, phenylalanine, tryptophane and methionine, have been examined to verify the selectivity of the proposed method. The results indicate that when the concentrations of these three amino acids are less than 200 times of the concentration of melamine, the melamine detection will not be affected (A component is considered as an interferent when it causes a variation greater than ±5% in the absorbance of the analyte). These results show that the method has very good selectivity. 3.6 Prediction of spiked milk samples In this experiment, the spiked milk samples were considered in the prediction set of the grey mixtures. The amount of melamine in the spiked samples was selected randomly and known precisely. The prediction set was used to determine the accuracy, precision and repeatability of this method. Three parallel tests for each sample were carried out. The results were summarised in Table 1 . From Table 1 , it can be seen that the recoveries ( R ) of melamine for individual samples varied between 98.06% and 100.43%, and the relative standard deviations (RSD) of melamine for individual samples varied between 0.27% and 6.15%. Based on the data in Table 1 , the average recovery (AR) of melamine in mixtures can be calculated as: (5) AR ( % ) = 100 × ∑ i = 1 n ( c ˆ i / c i ) / n where n is the number of samples in the prediction set, c i is the actual concentration in the i th sample, and c ˆ i is its estimated value. The prediction error of melamine in the mixtures can be calculated as the relative predictive error (RPEs) of the predicted concentrations: (6) RPEs ( % ) = 100 × ∑ i = 1 n ( c ˆ i - c i ) 2 ∑ i = 1 n ( c i ) 2 0.5 The root-mean-square error of prediction (RMSEP) can be calculated as (7) RMSEP = ∑ i = 1 n ( c ˆ i - c i ) 2 n 0.5 The merits of RAFA can be demonstrated by examining the AR, RPEs and RMSEP of melamine in the prediction set. The calculated results of AR, RPEs and RMSEP for RAFA are 99.10%, 0.91% and 0.0151, respectively. These results show that RAFA has excellent predictive ability for this system. The experiment results of the method presented in this paper has also been compared with the official methods, including HPLC, LC–MS/MS, GC–MS, and GC–MS/MS ( GB/T 22388, 2008 ). The metrics that have been investigated are the detection range, RSD, LOQ and R . The experiment results are summarised in Table 2 . It can be seen from Table 2 that the proposed method can be used to quantify melamine in a wide detection range with low relative standard deviations, good sensitivity and high recovery rates. 4 Conclusion In this paper, a simple and low-cost approach has been developed to determine melamine in grey mixtures from the pH-spectra matrices by using the RAFA method. This method have the following advantages: (1) the complicated pre-separation operation can be omitted; (2) the approach is simple to producing second order data; (3) the apparatus required in this method is easily obtainable. This proposed method has been successfully used to determine melamine in spiked milk samples with low prediction errors and high recoveries. Although this paper only presents how to use this proposed method to determine melamine in milk, this method is applicable to determine melamine in other food products, such as pet food, wheat gluten, and so on. Acknowledgement Financial support from Hunan Technology Department, China (Project No. 2007FJ4152 and 2010FJ3020 ) is gratefully acknowledged. References Abdollahi et al., 2008 H. Abdollahi A. Safavi S. Zeinali Model-based rank annihilation factor analysis for quantitative analysis of mixtures of monoprotic acids using multivariate spectrophotometric acid–base titrations Chemometrics and Intelligent Laboratory Systems 94 2008 112 117 Afkhami and Khalafi, 2007 Abbas Afkhami Lida Khalafi Spectrophotometric investigation of the effect of β-cyclodextrin on the intramolecular cyclization reaction of catecholamines using rank annihilation factor analysis Analytica Chimica Acta 599 2007 241 248 An et al., 2010 L. An J. Deng L. Zhou H. Li F. Chen H. Wang Simultaneous spectrophotometric determination of trace amount of malachite green and crystal violet in water after cloud point extraction using partial least squares regression Journal of Hazardous Materials 175 2010 883 888 Bahram and Mabhooti, 2009 Morteza. Bahram Mehdi. Mabhooti Rank annihilation factor analysis using mean centering of ratio spectra for kinetic-spectrophotometric analysis of unknown samples Analytica Chimica Acta 639 2009 19 28 Cao et al., 2009 Q. Cao H. Zhao L. Zeng J. Wang R. Wang X. Qiu Electrochemical determination of melamine using oligonucleotides modified gold electrodes Talanta 80 2009 484 488 Chen, 2009 J. Chen A worldwide food safety concern in 2008-melamine-contaminated infant formula in China caused urinary tract stone in 290,000 children in China Chinese Medical Journal 122 3 2009 243 244 Cheng et al., 2008 Cheng, Y., Dong, Y., Wu, J., Yang, X., Bai, H., Zheng, H., et al. (2008). Screening melamine adulterant in milk powder with laser Raman spectrometry. Journal of Food Composition and Analysis, Accepted Manuscript, doi:10.1016/j.jfca.2009.08.006. GB/T 22388, 2008 GB/T 22388-2008, Determination of melamine in raw milk and dairy products, October 7, 2008. Guan, 2009 N. Guan Melamine-contaminated powdered formula and urolithiasis in young children The New England Journal of Medicine 360 11 2009 1067 1074 Ho et al., 1978 C.N. Ho G.D. Christian E.R. Davidson Application of the method of rank annihilation to quantitative analyses of multicomponent fluorescence data from the video fluorometer Analytical Chemistry 50 8 1978 1108 1113 Kim et al., 2008 B. Kim B.L. Perkins J.R. Bushway Determination of melamine in pet food by enzyme immunoassay, high-performance liquid chromatography with diode array detection, and ultra-performance liquid chromatography with tandem mass spectrometry Journal of AOAC International 91 2 2008 408 413 Klampfl et al., 2009 Christian W. Klampfl Liisa Andersen Manuela Haunschmidt Markus Himmelsbach Wolfgang Buchberger Analysis of melamine in milk powder by CZE using UV detection and hyphenation with ESI quadrupole/TOF MS detection Electrophoresis 30 2009 1 4 Liang et al., 2009 R. Liang R. Zhang W. Qin Potentiometric sensor based on molecularly imprinted polymer for determination of melamine in milk Sensors and Actuators B 141 2009 544 550 Lin et al., 2008 M. Lin L. He J. Awika L. Yang D.R. Ledoux H. Li Detection of melamine in gluten, chicken feed, and processed foods using surface enhanced Raman spectroscopy and HPLC Journal of Food Science 73 8 2008 129 134 Miao et al., 2009 H. Miao S. Fan Y. Wu L. Zhang P. Zhou J. Li Simultaneous determination of melamine, ammelide, ammeline, and cyanuric acid in milk and milk products by gas chromatography–tandem mass spectrometry Biomedical and Environmental Sciences 22 2009 87 94 Muñiz-Valencia et al., 2008 R. Muñiz-Valencia S.G. Ceballos-Magana D. Rosales-Martinez R. Gonzalo-Lumbreras A. Santos-Montes A. Cubedo-Fernandez-Trapiella Method development and validation for melamine and its derivatives in rice concentrates by liquid chromatography. Application to animal feed samples Analytical Bioanalytical Chemistry 392 2008 523 531 Rima et al., 2009 Jamil Rima, Maurice Abourida, Ting Xu, Il Kyu Cho, & Soula Kyriacos. (2009). New spectrophotometric method for the quantitative determination of melamine using Mannich reaction. Journal of Food Composition and Analysis. doi:10.1016/j.jfca.02.010 . Sun et al., 2009 Sun, H., Wang, L., Ai, L., Liang, S., & Wu, H. (2009). A sensitive and validated method for determination of melamine residue in liquid milk by reversed phase high-performance liquid chromatography with solid-phase extraction, Food Control, Accepted Manuscript, doi:10.1016/j.foodcont.2009.10.008 . Tsai et al., 2009 I.-Lin. Tsai Shao.-Wen. Sun Hsiao.-Wei. Liao Shu.-Chiao. Lin Ching.-Hua. Kuo Rapid analysis of melamine in infant formula by sweeping-micellar electrokinetic chromatography Journal of Chromatography A 1216 2009 8296 8303 Wang et al., 2009 J. Wang L. Jiang Q. Chu J. Ye Residue analysis of melamine in milk products by micellar electrokinetic capillary chromatography with amperometric detection Food Chemistry, Accepted Manuscript 121 2009 215 219 Wu et al., 2009 Yu-Tse. Wu Chih.-Min. Huang Chia.-Chun. Lin Ho. Wei-An Lie.-Chwen. Lin Ting.-Fang. Chiu Determination of melamine in rat plasma, liver, kidney, spleen, bladder and brain by liquid chromatography-tandem mass spectrometry Journal of Chromatography A 1216 2009 7595 7601 Xu et al., 2009 X. Xu Y. Ren Y. Zhu Z. Cai J. Han B. Huang Direct determination of melamine in dairy products by gas chromatography/mass spectrometry with coupled column separation Analytica Chimica Acta 650 2009 39 43 Yan et al., 2009 N. Yan L. Zhou Z. Zhu X. Chen Determination of melamine in dairy products, fish feed, and fish by capillary zone electrophoresis with diode array detection Journal of Agricultural and Food Chemistry 57 3 2009 807 811 Yang et al., 2009 S. Yang J. Ding J. Zheng B. Hu J. Li H. Chen Detection of melamine in milk products by surface desorption atmospheric pressure chemical ionization mass spectrometry Analytical Chemistry 81 2009 2426 2436 Yu et al., 2009 J. Yu C. Zhang P. Dai S. Ge Highly selective molecular recognition and high throughput detection of melamine based on molecularly imprinted sol-gel film Analytica Chimica Acta 651 2009 209 214 Yuan et al., 2008 J. Yuan L. Liao Y. Lin C. Deng B. He Determination of Sudan I in chilli powder from solvent components gradual change-visible spectra data using second order calibration algorithms Analytica Chimica Acta 607 2008 160 167 Zhai et al., 2008 Zhai, C., Qiang, W., Sheng, J., Lei, J., & Ju, H. (2008). Pretreatment-free fast ultraviolet detection of melamine in milk products with a disposable microfluidic device. Journal of Chromatography A, Accepted Manuscript, doi:10.1016/j.chroma.2009.12.003 . Zhu et al., 2009 L. Zhu Gerardo. Gamez H. Chen Konstantin. Chingin Renato. Zenobi Rapid detection of melamine in untreated milk and wheat gluten by ultrasound-assisted extractive electrospray ionization mass spectrometry (EESI-MS) Chemical Communications 559 2009 561