P. Legrand

Classification of edible fats and oils by principal component analysis of Fourier transform infrared spectra

Abstract Principal component analysis of Fourier transform infrared (FTIR) spectra is performed to classify edible fats and oils with regard to their origin. Convenient sampling methods are proposed to record reproducible spectra: horizontal attenuated total reflection for fat samples and mid-IR optical fibre method for oils. The first principal component extracted from the first derivatized spectra of fats allows separation of butters and vegetable margarines on the basis of the sign of the projection. The interpretation of this first principal component establishes that the difference is due to the concentration of unsaturated fatty acids. For the oil samples, the spectral data must be second derivatized and again the first principal component separates sunflower seed oils from the olive and peanut extracts by the different concentrations of linoleic acid. The major interest of these methods using chemometric analysis of spectral data is in their rapidity, since no chemical treatment of samples is required.

Self-Modeling Mixture Analysis Applied to FT-Raman Spectral Data of Hydrogen Peroxide Activation by Nitriles

In the analytical environment, spectral data resulting from analysis of samples often represent mixtures of several components. Extraction of information about pure components of these kinds of mixtures is a major problem, especially when reference spectra are not available or when unstable intermediates are formed. Self-modeling multivariate mixture analysis has been developed for this type of problem. In this paper two examples will be used to show the potential of this technique coupled with FT-Raman spectroscopy to elucidate reaction mechanisms and to follow in situ the kinetics of chemical transformations.

Neural network modelling for very small spectral data sets: reduction of the spectra and hierarchical approach

Abstract For studies on industrial materials, scarcity of samples and incomplete information are everyday situations. Furthermore, the number of points per sample typically reaches several hundreds. Consequently, the sample-to-data ratio does not satisfy the requirements of most of the mathematical treatments. We thus discuss the use of different approaches in order to reduce the number of parameters of the networks in case of data sets with extremely small number of samples. Therefore, more or less new approaches using wavelet or Fourier-transform coefficients for the reduction of spectra have been offered for a few years. Moreover, the necessity emerges to associate these various pre-processing methods with the construction of input–output relationships models. Combinations of different artificial neural networks (ANNs) for non-linear hierarchical modelling are thus examined. In practice, we apply these methods to infrared spectra in three different situations: • qualitative analysis of complex mixtures (identification) • semi-quantitative analysis of a major compound • quantitative and precise analysis of minor compounds. This study demonstrates that, when real data are investigated, a combination of compression methods and multilevel modelling offers accuracy advantages compared with more classical architecture networks.

Classification of Green Coffees by FT-IR Analysis of Dry Extract

Principal component analysis (PCA) of infrared spectra has been used as a classification method for the green beans of coffee from various origin. Before spectral acquisition, sampling methods were tested for 45 samples, and we chose dry extract of water-soluble compounds on SiCaF2 supports. After PCA of the first derivatized spectra, the first four loadings were examined. The scores of the second principal component appear to be directly correlated by their sign to the species arabica or robusta. This result allows an easy classification. In the same way, the pigmentation is well characterized into two groups on the scattergram of the samples with respect to the PC1 and PC3 components. Another feature of this method is that the analysis of the spectral data in terms of residual variance separate components which are correlated with properties. This approach provides assistance in the interpretation of infrared spectra of complex mixtures.

Hydrolysis of haemoglobin surveyed by infrared spectroscopy: I. solvent effect on the secondary structure of haemoglobin

Abstract The hydrolysis of bovine haemoglobin in an acetic acid/sodium acetate buffer enables to produce peptides of major importance in biomedical research. The global objective is to survey this reaction by infrared spectroscopy. This article concerns the first step: the evaluation by spectroscopy of the effect on protein secondary structure of the addition of ethanol in the buffer.Conformational changes are related to solvent–protein interactions as the protein folding is driven by the entropy of removing hydrophobic groups from contact with the solvent. Therefore, the stability of haemoglobin in an ethanol–water mixture results in a competition between the water structure, which is strengthened by the presence of the alcohol, and the solubility of hydrophobic residues. Previous non infrared experiments, based on mass spectrometry for example, have been reported for the investigation of the denaturation of haemoglobin by an organic solvent.The use of vibrational spectroscopy for protein secondary structure determination has proved its efficiency. We focus here on the study of the denaturing of haemoglobin in a water medium by addition of ethanol. As our investigations deal with very low concentrated haemoglobin, we apply a technique that uses films dried from dilute solution. Although it is not generally accepted that the protein conformation is retained when the solvent evaporates off, we validate this method comparing some results to previous infrared study made at upper concentrations with liquid sampling. We observe that Fourier Transform Infrared (FTIR) Spectroscopy, combined with few mathematical treatments, permits to estimate that haemoglobin remains in a native form unless a concentration of more or less 20% of ethanol is reached. For greater values modifications are perceptible on the infrared spectra.

Quantitative analysis of paper coatings using artificial neural networks

Abstract This paper describes a neural network approach to the quantitative analysis of paper coatings. Infrared spectra of samples of coated paper were used as input data for the different types of artificial neural networks. Unsupervised learning was applied to obtain the clustering of samples with respect to similarities in the spectra. The self-organizing artificial neural network of Kohonen type produced a visual representation of the discovered groupings on a two-dimensional plane. Such mapping provided the expert a possibility to analyze the mutual arrangement of samples and to predict the properties of the test samples using their relative position with respect to existing clusters. Supervised learning with a multilayer feedforward network was used to construct the non-linear models that relate the spectral information and concentrations of three basic components of paper coating - styrene, butadiene, and carbonate. These models were used for prediction of concentrations of paper coating components for the test data set. The results of modeling demonstrate that accuracy of classification and prediction is better than those obtained with traditional methods like principal component analysis or partial least squares (from 4% to 2% for different components). According to our experience, the modeling with artificial neural networks is intuitively clear for the expert. This method allows to construct complex multivariable and multiresponse models in unified style. Causal relationships between inputs and outputs can be analyzed and explained.

Standardisation of near-IR spectrometers using artificial neural networks

Abstract In analytical chemistry, the use of chemometrics on near-infrared data undergoes a major problem. Large increase of error prediction is observed when calibration equation developed on a first instrument is directly used on another one. Since many spectral differences between two spectrometers can occur, standardisation procedure has rapidly became a necessary step for a long-dated use of quantitative or qualitative models. An original neural network approach is proposed to correct spectral differences by modelling spectral response of an instrument from second one before using calibration equations. In this way, the consuming time recalibration step of the second spectrometer was avoided and initial error prediction level was retrieved.

Hydrolysis of hemoglobin surveyed by infrared spectroscopy. II.Progress predicted by chemometrics

Abstract The hydrolysis of bovine hemoglobin produces specific peptides which have both hydrophobic and hydrophilic characters and are thus of major importance in biomedical and drug design researches. The global objective is to survey this reaction by infrared spectroscopy, as precisely as possible, compensating the method limitations by powerful chemometric treatments. In a first step of this study, we appreciated the effect on the protein secondary structure of the addition of ethanol in the buffer medium. We now focus on the evaluation, from the infrared spectra, of the degree of hydrolysis which is representative of the advancement of the reaction. Principal Component Analysis (PCA) enables proving that the information about the hydrolysis advancement is contained in the spectra especially in the amide I range since a through time classification of the samples along the first principal axis of the model is revealed. On the other hand, we use two well-known chemometrics approaches for the quantitative prediction of the hydrolysis degree. Artificial Neural Networks (ANNs) and Partial Least Square Regression (PLSR) models have been developed taking care to ensure the robustness of these methods. They both give satisfying results with regard to predictive abilities taking into account the complexity and quality of the spectra but PLSR enables the identification of the most important variables. With the help of these treatments, it is possible to measure the advancement of the reaction from the infrared spectra of the samples taken.

In situ quantitative and kinetic study by Fourier transform Raman spectroscopy of reaction between nitriles and hydroperoxides

Abstract The reaction of nitrile with alkaline hydrogen peroxide has been studied kinetically by means of iodometry several times. We proposed to set up an in situ analytical method to follow the consumption of nitrile. We applied FT-Raman spectroscopy coupled with a partial least-squares quantitative method to the study the reactions of hydrogen peroxide or tert -butyl hydroperoxide on benzonitrile and acetonitrile. The results obtained led us to conclusions on the reactivity of the hydroperoxides and on the rate of consumption of nitrile. Our study made also clear that FT-Raman spectroscopy was a convenient tool for controlling industrial processes.

Vibrational analyses of neoagarose and neocarrabiose oligomers

The structural characterization of five oligomers from the carrageenan family and three oligomers of agarose, has been the aim of the present work. The compounds were chosen so as to study principally the effect of the substitution by a sulphate group on the two main vibrations due to the two glycosidic linkages (the (alpha) 1,3 and the (beta) -1,4). The C-O stretching modes of the two quoted glycosidic linkages have previously been identified in the spectral region between 1120 and 1160 cm-1, whereas the C-O-C bending mode has been demonstrated to occur at about 730 cm-1.