This paper described the development of a multivariate classification methodology to detect fraud... more This paper described the development of a multivariate classification methodology to detect frauds in bovine meat based on mid-infrared spectroscopy and partial least squares discriminant analysis (PLS-DA). These frauds consisted of adding carrageenan, sodium chloride, and tripolyphosphate, ingredients that increase meat water holding capacity aiming to obtain economic gains. Meat pieces (fresh beef muscle) of the same bovine cut, M. semitendinosus , from different origins were injected with single to ternary mixtures of adulterants, and their purges were analyzed totaling 176 spectra. Multiclass PLS-DA models for specifically detecting each adulterant provided good results (correctly classification rates > 90%) only for tripolyphosphate. Nevertheless, a two-class PLS-DA model discriminating adulterated and non-adulterated meat provided high success rates (≥ 95%). Aiming to verify the model’s ability to detect other (non-trained) adulterant, this last model was combined with outlier detection in a soft version of a discriminant model that was able to correctly detect 100% of a new validation set consisting of 20 meat samples containing maltodextrin.
A fast method of multivariate calibration applied to ESI-MS data for quantification of adulterati... more A fast method of multivariate calibration applied to ESI-MS data for quantification of adulteration of EVOO with cheaper edible oils.
... Iniciação Científica FAFIG/Guaxupé, 2001. Aluna: Thais Maria dos Santos (3° ano) Título: A Co... more ... Iniciação Científica FAFIG/Guaxupé, 2001. Aluna: Thais Maria dos Santos (3° ano) Título: A Conscientização ea aplicação da química do cotidiano para o ensino médio 10. Participação em Projeto de Pesquisa 10.1. ... Coordenadores: RTS Frighetto e PJ Vaiarini xiv Page 10. ...
Nuts and peanuts are foods that are rich in minerals, vitamins, fibre and healthy fats in additio... more Nuts and peanuts are foods that are rich in minerals, vitamins, fibre and healthy fats in addition to antioxidant compounds. However, these food products can be subject to adulterations and fraud mainly due to their cost or contamination as a result of improper handling. Different types and degrees of damage can be caused to consumers due to food fraud, highlighting the serious consequences that can occur when the adulterant is toxic or allergenic. In this paper, portable near-infrared (NIR) spectroscopy combined with multivariate supervised classification was proposed to detect peanuts, Brazil nuts, macadamia nuts and pecan nuts in cashew nut samples, covering a wide concentration range (10.0 to 0.1 % w/w) of adulterants/contaminants. Methods to predict five classes of samples, cashew nuts unadulterated and adulterated with peanuts, Brazil nuts, macadamia nuts and pecan nuts, were developed. Three variable selection strategies were tested: interval partial least squares (iPLS), genetic algorithm (GA) and the combination of iPLS-GA. Partial least squares discriminant analysis (PLS-DA) and soft independent modelling of class analogy (SIMCA) models were compared, and PLS-DA coupled with iPLS-GA provided the best results, with sensitivity between 81 and 93 % and selectivity between 94 and 100 %. Applicability for the rapid and non-destructive detection of fraud and cross-contamination with different types of allergenic nuts with portable equipment was demonstrated.
The development of rapid, reliable, and environmentally friendly analytical methods for food anal... more The development of rapid, reliable, and environmentally friendly analytical methods for food analysis is increasingly required. In this sense, the combination of spectrofluorimetry and second-order calibration is a promising alternative to the determination of naturally fluorescent analytes in food matrices, as compared to the laborious and time-consuming chromatographic methods. This work combined the chemometric second-order method parallel factor analysis (PARAFAC) with spectrofluorimetry aiming to develop a new analytical method to quantify phenylalanine in honey samples. Honey samples of different botanical and geographical origins were diluted with deionized water, and their excitation-emission matrices were recorded between 280–350 nm (excitation) and 360–490 nm (emission). Due to the presence of matrix effect, second-order standard addition was employed. This methodology utilizes the second-order advantage, which allows to quantify an analyte in the presence of uncalibrated/unmodeled interferences. Phenylalanine concentrations were estimated in the range from 5.7 to 32.5 mg kg−1. Method validation was performed by verifying the agreement between the results of the developed method and from an independent methodology based on liquid chromatography. In addition, proper figures of merit were estimated for the proposed method, such as sensitivity, analytical sensitivity, limits of detection, and quantification.
The determination of alcohol content in beers is essential for the quality control of this bevera... more The determination of alcohol content in beers is essential for the quality control of this beverage. This paper proposed and validated a new rapid and direct multivariate method for this aim using a portable near-infrared (NIR) spectrometer and partial least squares (PLS) regression. Reference values were obtained by a gas chromatography with flame ionization detection (GC-FID) method developed and validated for this purpose. Aiming at building a robust model, a great variety of beers, from different styles, brands, and breweries, was incorporated into the model. NIR spectra were recorded between 908 and 1676 nm for 92 beer samples, corresponding to a range from 3.2 to 10.9% (v/v) of alcohol content. PLS model provided accurate results with root-mean-square error of calibration (RMSEC) and prediction (RMSEP) of 0.5% and 0.6%, respectively. The developed method was validated through the estimate of figures of merit, such as linearity, trueness, precision, analytical sensitivity, bias, and residual prediction deviation (RPD). In addition, an elliptical joint confidence region was calculated to verify the linearity, and confidence intervals based on the standard prediction errors were estimated for the validation samples.
Spectrofluorimetry combined with multiway chemometric tools were applied to discriminate pure Aro... more Spectrofluorimetry combined with multiway chemometric tools were applied to discriminate pure Aroeira honey samples from samples adulterated with corn syrup, sugar cane molasses and polyfloral honey. Excitation emission spectra were acquired for 232 honey samples by recording excitation from 250 to 500 nm and emission from 270 to 640 nm. Parallel factor analysis (PARAFAC), partial least squares discriminant analysis (PLS-DA), unfolded PLS-DA (UPLS-DA) and multilinear PLS-DA (NPLS-DA) methods were used to decompose the spectral data and build classification models. PLS-DA models presented poor classification rates, demonstrating the limitation of the traditional two-way methods for this dataset, and leading to the development of three-way classification models. Overall, UPLS-DA provided the best classification results with misclassification rates of 4% and 8% for the training and test sets, respectively. These results showed the potential of the proposed method for routine laboratory analysis as a simple, reliable, and affordable tool.
Abstract The monitoring of biodiesel production processes is currently performed by methods requi... more Abstract The monitoring of biodiesel production processes is currently performed by methods requiring high reagent consumption and long response times. Aiming to overcome these drawbacks, this article developed chemometric models combining principal component analysis (PCA), multivariate curve resolution alternating least squares (MCR-ALS) and mid infrared spectroscopy to determine simultaneously the concentrations of the main components, methyl esters and triglycerides, along transesterification reactions. This methodology also tried to model the intermediates (fatty acids plus monoglycerides, and diglycerides), thus providing semi-quantitative predictions for these components. These reactions were carried out in different experimental conditions according to a factorial design with two factors, the batch of soybean oil (raw material) and the alcohol:oil molar ratio. The concentration values estimated by high performance liquid chromatography (HPLC), as the reference method, were the inputs for the correlation constraint used as a key step in the multivariate calibration MCR-ALS models. The spectral profiles recovered for the methyl esters and for the triglycerides have shown high correlations (greater than 0.985) in comparison to the reference spectra. In addition, the spectra calculated for fatty acids, monoglycerides and diglycerides showed absorption bands characteristic of functional groups present in these molecules, mainly the hydroxyl and carbonyl stretching bands. The root mean square errors of calibration and prediction estimated for the triglycerides and for the methyl esters were within the range from 1.9% to 6.3%.
It was with great regret that we received the incredibly sad news of the untimely death of Profes... more It was with great regret that we received the incredibly sad news of the untimely death of Professor Ronei Jesus Poppi (Figure 1), on the last April 25 in the city of Campinas, S~ao Paulo State, Brazil. Ronei was born on the outskirts of Campinas, son of barber and a housewife, in a lower middle-class family. In 1986, he earned a bachelor's degree in Chemistry at the State University of Campinas (UNICAMP). In 1989, he obtained a master's degree at the same university under the supervision of Professors José Fernando Faigle, already deceased, and Roy Bruns, with a dissertation on the quantification of analytes with overlapped chromatographic peaks by multivariate calibration. This was a pioneering study in multivariate calibration in Brazil and South America, resulting in the first Brazilian/South American article to apply partial least squares (PLS) and principal component regression (PCR), published in the Journal of Chromatography in 1991. It is important to highlight that the start of chemometrics in Brazil had taken place only some years before, in the beginning of the 1980s, when Prof. Bruns invited Bruce R. Kowalski who gave a course on chemometrics during a month stay at UNICAMP, and initiated his landmark studies in this country. Ronei started to work on his master's degree in 1986 and accepted the challenge to do research in multivariate calibration that was a relatively new subject with very few publications at the time. Toluene, isooctane, and ethanol mixtures were analyzed by gas chromatography with a simple thermal conductivity detector. The effects of different degrees of overlap on calibration and prediction errors were investigated using chromatograms obtained at different column temperatures. At that time, peak heights and areas were not registered digitally, and Ronei used a simple ruler to measure 41 chromatographic response values at different retention times for all the calibration and validation samples. The PLS calculations were carried out on an 8-bit DICON microcomputer with an 8-inch 32-kB floppy disk for memory space using the SIMCA3B program. PCR regressions were made with home-made software on a PC-XT that was the successor of IBM's first PC and had 5 and 1/4 inch removable disks and a hard disk with 256 kB of memory. Work was done in a DOS environment before Windows operating systems became available. It was hard to attract analytical chemistry students to work in chemometrics in the 1980s because the mathematical methods were intimidating and computational tools very limited. In fact, chemometrics was not very well regarded by the most classical analytical chemists in Brazil at the time of the beginning of his career. But this did not deter Ronei who did much to help consolidate progress in chemometrics in Brazil. Even at such an early point in his career Ronei expressed his desire to form his own research group and work with students.
This article aims to develop and validate a multivariate model for quantifying Robusta-Arabica co... more This article aims to develop and validate a multivariate model for quantifying Robusta-Arabica coffee blends by combining near infrared spectroscopy (NIRS) and total reflection X-ray fluorescence (TXRF). For this aim, 80 coffee blends (0.0-33.0%) were formulated. NIR spectra were obtained in the wavenumber range 11100-4950 cm-1 and 14 elements were determined by TXRF. Partial least squares models were built using data fusion at low and medium levels. In addition, selection of predictive variables based on their importance indices (SVPII) improved results. The best model reduced the number of variables from 1114 to 75 and root mean square error of prediction from 4.1% to 1.7%. SVPII selected NIR regions correlated with coffee components, and the following elements were chosen: Ti, Mn, Fe, Cu, Zn, Br, Rb, Sr. The model interpretation took advantage of the data fusion between atomic and molecular spectra in order to characterize the differences between these coffee varieties.
This paper described the development of a multivariate classification methodology to detect fraud... more This paper described the development of a multivariate classification methodology to detect frauds in bovine meat based on mid-infrared spectroscopy and partial least squares discriminant analysis (PLS-DA). These frauds consisted of adding carrageenan, sodium chloride, and tripolyphosphate, ingredients that increase meat water holding capacity aiming to obtain economic gains. Meat pieces (fresh beef muscle) of the same bovine cut, M. semitendinosus , from different origins were injected with single to ternary mixtures of adulterants, and their purges were analyzed totaling 176 spectra. Multiclass PLS-DA models for specifically detecting each adulterant provided good results (correctly classification rates > 90%) only for tripolyphosphate. Nevertheless, a two-class PLS-DA model discriminating adulterated and non-adulterated meat provided high success rates (≥ 95%). Aiming to verify the model’s ability to detect other (non-trained) adulterant, this last model was combined with outlier detection in a soft version of a discriminant model that was able to correctly detect 100% of a new validation set consisting of 20 meat samples containing maltodextrin.
A fast method of multivariate calibration applied to ESI-MS data for quantification of adulterati... more A fast method of multivariate calibration applied to ESI-MS data for quantification of adulteration of EVOO with cheaper edible oils.
... Iniciação Científica FAFIG/Guaxupé, 2001. Aluna: Thais Maria dos Santos (3° ano) Título: A Co... more ... Iniciação Científica FAFIG/Guaxupé, 2001. Aluna: Thais Maria dos Santos (3° ano) Título: A Conscientização ea aplicação da química do cotidiano para o ensino médio 10. Participação em Projeto de Pesquisa 10.1. ... Coordenadores: RTS Frighetto e PJ Vaiarini xiv Page 10. ...
Nuts and peanuts are foods that are rich in minerals, vitamins, fibre and healthy fats in additio... more Nuts and peanuts are foods that are rich in minerals, vitamins, fibre and healthy fats in addition to antioxidant compounds. However, these food products can be subject to adulterations and fraud mainly due to their cost or contamination as a result of improper handling. Different types and degrees of damage can be caused to consumers due to food fraud, highlighting the serious consequences that can occur when the adulterant is toxic or allergenic. In this paper, portable near-infrared (NIR) spectroscopy combined with multivariate supervised classification was proposed to detect peanuts, Brazil nuts, macadamia nuts and pecan nuts in cashew nut samples, covering a wide concentration range (10.0 to 0.1 % w/w) of adulterants/contaminants. Methods to predict five classes of samples, cashew nuts unadulterated and adulterated with peanuts, Brazil nuts, macadamia nuts and pecan nuts, were developed. Three variable selection strategies were tested: interval partial least squares (iPLS), genetic algorithm (GA) and the combination of iPLS-GA. Partial least squares discriminant analysis (PLS-DA) and soft independent modelling of class analogy (SIMCA) models were compared, and PLS-DA coupled with iPLS-GA provided the best results, with sensitivity between 81 and 93 % and selectivity between 94 and 100 %. Applicability for the rapid and non-destructive detection of fraud and cross-contamination with different types of allergenic nuts with portable equipment was demonstrated.
The development of rapid, reliable, and environmentally friendly analytical methods for food anal... more The development of rapid, reliable, and environmentally friendly analytical methods for food analysis is increasingly required. In this sense, the combination of spectrofluorimetry and second-order calibration is a promising alternative to the determination of naturally fluorescent analytes in food matrices, as compared to the laborious and time-consuming chromatographic methods. This work combined the chemometric second-order method parallel factor analysis (PARAFAC) with spectrofluorimetry aiming to develop a new analytical method to quantify phenylalanine in honey samples. Honey samples of different botanical and geographical origins were diluted with deionized water, and their excitation-emission matrices were recorded between 280–350 nm (excitation) and 360–490 nm (emission). Due to the presence of matrix effect, second-order standard addition was employed. This methodology utilizes the second-order advantage, which allows to quantify an analyte in the presence of uncalibrated/unmodeled interferences. Phenylalanine concentrations were estimated in the range from 5.7 to 32.5 mg kg−1. Method validation was performed by verifying the agreement between the results of the developed method and from an independent methodology based on liquid chromatography. In addition, proper figures of merit were estimated for the proposed method, such as sensitivity, analytical sensitivity, limits of detection, and quantification.
The determination of alcohol content in beers is essential for the quality control of this bevera... more The determination of alcohol content in beers is essential for the quality control of this beverage. This paper proposed and validated a new rapid and direct multivariate method for this aim using a portable near-infrared (NIR) spectrometer and partial least squares (PLS) regression. Reference values were obtained by a gas chromatography with flame ionization detection (GC-FID) method developed and validated for this purpose. Aiming at building a robust model, a great variety of beers, from different styles, brands, and breweries, was incorporated into the model. NIR spectra were recorded between 908 and 1676 nm for 92 beer samples, corresponding to a range from 3.2 to 10.9% (v/v) of alcohol content. PLS model provided accurate results with root-mean-square error of calibration (RMSEC) and prediction (RMSEP) of 0.5% and 0.6%, respectively. The developed method was validated through the estimate of figures of merit, such as linearity, trueness, precision, analytical sensitivity, bias, and residual prediction deviation (RPD). In addition, an elliptical joint confidence region was calculated to verify the linearity, and confidence intervals based on the standard prediction errors were estimated for the validation samples.
Spectrofluorimetry combined with multiway chemometric tools were applied to discriminate pure Aro... more Spectrofluorimetry combined with multiway chemometric tools were applied to discriminate pure Aroeira honey samples from samples adulterated with corn syrup, sugar cane molasses and polyfloral honey. Excitation emission spectra were acquired for 232 honey samples by recording excitation from 250 to 500 nm and emission from 270 to 640 nm. Parallel factor analysis (PARAFAC), partial least squares discriminant analysis (PLS-DA), unfolded PLS-DA (UPLS-DA) and multilinear PLS-DA (NPLS-DA) methods were used to decompose the spectral data and build classification models. PLS-DA models presented poor classification rates, demonstrating the limitation of the traditional two-way methods for this dataset, and leading to the development of three-way classification models. Overall, UPLS-DA provided the best classification results with misclassification rates of 4% and 8% for the training and test sets, respectively. These results showed the potential of the proposed method for routine laboratory analysis as a simple, reliable, and affordable tool.
Abstract The monitoring of biodiesel production processes is currently performed by methods requi... more Abstract The monitoring of biodiesel production processes is currently performed by methods requiring high reagent consumption and long response times. Aiming to overcome these drawbacks, this article developed chemometric models combining principal component analysis (PCA), multivariate curve resolution alternating least squares (MCR-ALS) and mid infrared spectroscopy to determine simultaneously the concentrations of the main components, methyl esters and triglycerides, along transesterification reactions. This methodology also tried to model the intermediates (fatty acids plus monoglycerides, and diglycerides), thus providing semi-quantitative predictions for these components. These reactions were carried out in different experimental conditions according to a factorial design with two factors, the batch of soybean oil (raw material) and the alcohol:oil molar ratio. The concentration values estimated by high performance liquid chromatography (HPLC), as the reference method, were the inputs for the correlation constraint used as a key step in the multivariate calibration MCR-ALS models. The spectral profiles recovered for the methyl esters and for the triglycerides have shown high correlations (greater than 0.985) in comparison to the reference spectra. In addition, the spectra calculated for fatty acids, monoglycerides and diglycerides showed absorption bands characteristic of functional groups present in these molecules, mainly the hydroxyl and carbonyl stretching bands. The root mean square errors of calibration and prediction estimated for the triglycerides and for the methyl esters were within the range from 1.9% to 6.3%.
It was with great regret that we received the incredibly sad news of the untimely death of Profes... more It was with great regret that we received the incredibly sad news of the untimely death of Professor Ronei Jesus Poppi (Figure 1), on the last April 25 in the city of Campinas, S~ao Paulo State, Brazil. Ronei was born on the outskirts of Campinas, son of barber and a housewife, in a lower middle-class family. In 1986, he earned a bachelor's degree in Chemistry at the State University of Campinas (UNICAMP). In 1989, he obtained a master's degree at the same university under the supervision of Professors José Fernando Faigle, already deceased, and Roy Bruns, with a dissertation on the quantification of analytes with overlapped chromatographic peaks by multivariate calibration. This was a pioneering study in multivariate calibration in Brazil and South America, resulting in the first Brazilian/South American article to apply partial least squares (PLS) and principal component regression (PCR), published in the Journal of Chromatography in 1991. It is important to highlight that the start of chemometrics in Brazil had taken place only some years before, in the beginning of the 1980s, when Prof. Bruns invited Bruce R. Kowalski who gave a course on chemometrics during a month stay at UNICAMP, and initiated his landmark studies in this country. Ronei started to work on his master's degree in 1986 and accepted the challenge to do research in multivariate calibration that was a relatively new subject with very few publications at the time. Toluene, isooctane, and ethanol mixtures were analyzed by gas chromatography with a simple thermal conductivity detector. The effects of different degrees of overlap on calibration and prediction errors were investigated using chromatograms obtained at different column temperatures. At that time, peak heights and areas were not registered digitally, and Ronei used a simple ruler to measure 41 chromatographic response values at different retention times for all the calibration and validation samples. The PLS calculations were carried out on an 8-bit DICON microcomputer with an 8-inch 32-kB floppy disk for memory space using the SIMCA3B program. PCR regressions were made with home-made software on a PC-XT that was the successor of IBM's first PC and had 5 and 1/4 inch removable disks and a hard disk with 256 kB of memory. Work was done in a DOS environment before Windows operating systems became available. It was hard to attract analytical chemistry students to work in chemometrics in the 1980s because the mathematical methods were intimidating and computational tools very limited. In fact, chemometrics was not very well regarded by the most classical analytical chemists in Brazil at the time of the beginning of his career. But this did not deter Ronei who did much to help consolidate progress in chemometrics in Brazil. Even at such an early point in his career Ronei expressed his desire to form his own research group and work with students.
This article aims to develop and validate a multivariate model for quantifying Robusta-Arabica co... more This article aims to develop and validate a multivariate model for quantifying Robusta-Arabica coffee blends by combining near infrared spectroscopy (NIRS) and total reflection X-ray fluorescence (TXRF). For this aim, 80 coffee blends (0.0-33.0%) were formulated. NIR spectra were obtained in the wavenumber range 11100-4950 cm-1 and 14 elements were determined by TXRF. Partial least squares models were built using data fusion at low and medium levels. In addition, selection of predictive variables based on their importance indices (SVPII) improved results. The best model reduced the number of variables from 1114 to 75 and root mean square error of prediction from 4.1% to 1.7%. SVPII selected NIR regions correlated with coffee components, and the following elements were chosen: Ti, Mn, Fe, Cu, Zn, Br, Rb, Sr. The model interpretation took advantage of the data fusion between atomic and molecular spectra in order to characterize the differences between these coffee varieties.
Uploads
Papers by Marcelo Martins Sena