M.C. Ortiz

Capability of detection of an analytical method evaluating false positive and false negative (ISO 11843) with partial least squares

Analytical techniques based on soft multivariate calibrations (as those which provide first and second order analytical signals necessarily are) remain outside the field of application of the ISO norms related to capability of detection. In this work, a complete solution for the problem of applying ISO norm 11843 to soft calibration (for instance, one or multi-way partial least squares (PLS)) is provided. The methodological procedure is applied to different case studies which implies different analytical techniques.

Handling intrinsic non-linearity in near-infrared reflectance spectroscopy

Abstract The relationship between absorption in the near-infrared (NIR) spectral region and the target analytical parameter is frequently of the non-linear type. The origin of the non-linearity can be widely varied and difficult to identify. In some cases, the relationship between absorption and the analytical parameter of interest is intrinsically non-linear owing to the very chemical nature of the sample or analyte concerned. In this work, various multivariate calibration procedures were tested with a view to overcoming intrinsic non-linearity in NIR reflectance. An approach to solving the problem is suggested. Calibration was done, after transformation of spectra, by using linear and non-linear techniques. The linear calibration techniques used are partial least squares (PLS) regression (with and without variable selection), linear PLS with X projection (LP-PLS) and stepwise polynomial principal component (SWP-PC) regression. Non-linear calibration methods included polynomial PLS (PPLS) and artificial neural networks (ANNs). Results were compared on the basis of NIR spectra for ampicillin trihydrate samples, where the simultaneous presence of crystallization water and surface moisture gives rise to intrinsic non-linearity that affects the determination of the total water content in the sample. The best results were obtained by using the non-linear calibration techniques.

Multivariate calibration transfer applied to the routine polarographic determination of copper, lead, cadmium and zinc

Abstract The application of calibration transfer methods has been successful in combination with near-infrared spectroscopy or fluorescence spectroscopy for prediction of chemical composition. One of the methods developed that provides accurate performance is the piecewise direct standardization method (PDS), which in this paper is applied to transfer from one day to another the partial least squares (PLS) models built in the polarographic determination of copper, lead, cadmium and zinc. This is an electrochemical example in which interferences have been found, making necessary the use of soft calibration models because of their ability to model this phenomenon, which implies a large number of standard samples for the calibration. Once the PLS models are built, the calibration transfer is carried out to overcome the instrumental change over time in this routine analysis, allowing one to reduce from 28 to 8 the number of calibration standards necessary for later determinations. Standard errors of prediction (SEP) are found similar to those given by the complete recalibration.

A study of robustness with multivariate calibration. Application to the polarographic determination of benzaldehyde

A procedure to evaluate the robustness of an analytical method when there are changes in some experimental variables, when using multivariate calibration, is proposed. The procedure consists of analysing the root mean square error of prediction (RMSEP) as a response to a Plackett-Burman experimental design, through which the influence of several experimental factors on the prediction capability of the multivariate partial least squares (PLS) models built is studied. Two different ways of analysing the experimental design response are considered: establishing the residual variance with replicates and using Lenths method. The proposed methodology has been applied to estimate the robustness of the polarographic determination of benzaldehyde when PLS calibration is used.

Analysis of protein chromatographic profiles joint to partial least squares to detect adulterations in milk mixtures and cheeses

To prevent possible frauds and give more protection to companies and consumers it is necessary to control that the types of milk used in the elaboration of dairy products correspond to those appearing in their label. Therefore, it is greatly interesting to have efficient, quick and cheap methods of analysis to identify them. In the present work, the multivariate data are the protein chromatographic profiles of cheese and milk extracts, obtained by high-performance liquid chromatography with diode-array detection (HPLC-DAD). These data correspond to pure samples of bovine, ovine and caprine milk, and also to binary and ternary mixtures. The structure of the data is studied through principal component analysis (PCA), whereas the percentage of each kind of milk has been determined by a partial least squares (PLS) calibration model. In cheese elaborated with mixtures of milk, the procedure employed allows one to detect 3.92, 2.81 and 1.47% of ovine, caprine and bovine milk, respectively, when the probability of false non-compliance is fixed at 0.05. These percentages reach 7.72, 5.52 and 2.89%, respectively, when both the probability of false non-compliance and false compliance are fixed at 0.05.

Analysis and comparison of SIMCA models for denominations of origin of wines from de Canary Islands (Spain) builds by means of their trace and ultratrace metals content

Abstract Various models have been constructed and analysed for eight denominations of origin of wines bottled in the Canary Islands, using their content in different metals, with the Soft Independent Modelling Class Analogy (SIMCA) technique. The metals were grouped in three blocks: “rare earths”, “lead isotope ratios” and “other metals”. The model constructed with all the variables was taken as the reference model. This model has adequate sensibility and specificity. The contribution of each element to the model of each denomination of origin (DOs), as well as their capacity to discriminate between the DOs is shown. Cluster analysis, using the Ward method, of the SIMCA distances between the different DOs reveals the similarity of the different denominations of origin. Using only the rare earths or the lead isotope ratios it is not possible to construct adequate models for the different DOs given the low specificity obtained. The models constructed with the other metals, alone or combined with the lead isotope ratios, gave similar results to those obtained using all the variables. Finally, the similarity of the models was analysed by a weighted distance between the sensibilities and specificities of each of them compared with the rest. Cluster analysis using the Ward method shows the models which are similar as to their sensibility and specificity. The innovative aspect of the work lies in the use of cluster analysis to demonstrate the similarity between the SIMCA boxes of a model, and the definition of the distance between models based on the sensibility and specificity for the eight DOs with the five groups of variables considered.

Fluorescence quantification of tetracycline in the presence of quenching matrix effect by means of a four-way model.

In this work, a four-way tensor is used to model the quenching effect in fluorescent measurements. By means of the analysis of excitation-emission matrices obtained in the determination of tetracycline in tea, which acts as quencher, it is shown as the impossibility to use a calibration, or an addition standard based on a three-way model. It is analysed the quencher multiplicative effect made on the tetracycline signal by means of an ANOVA. However, by arranging the experimental data in a four-way tensor, it is viable to perform a calibration based on the parallel factor analysis, PARAFAC, decomposition and a four-way partial least squares, 4-PLS, regression to quantify the tetracycline in the presence of the matrix quencher effect. 4-PLS calibration provides better results. In the range from 40 to 220 microg L(-1) gives an average of relative errors in absolute value equal to 8.02% in prediction (3.40% in calibration). The repeatability as standard deviation in this range is 5.08 microg L(-1) and the method is accurate, slope and intercept being statistically equal to 1 and 0, respectively when a regression calculated versus true concentration is performed. Moreover, it has a decision limit (CCalpha) of 13.87 microg L(-1) for a probability of false positive, alpha, equal to 0.05 and a capability of detection (CCbeta) of 26.63 microg L(-1) (for probabilities of false positive, alpha, false negative, beta, equals to 0.05).

Using continuum regression for quantitative analysis with overlapping signals obtained by differential pulse polarography

Abstract In this work the power continuum regression was used to show its interest and efficiency in the resolution of overlapping polarographic signals. The methodology was applied to an example of calibration with mixtures of T1(I) and Pb(II), using differential pulse polarography (DPP). The errors in absolute value oscillated between 1% and 3.5%, according to whether the analyte determined is that which originates the largest or the smallest signal in the presence of the other component which acts as interferent. The advantages of the power continuum regression in the automatic election of the optimum multivariate regression to be used in each case are also described. Furthermore, the incidence of the autoscaling and of a confidence interval for the PRESS (calculated by the bootstrap method) in the accuracy reached is evaluated.

Optimization of a GC/MS procedure that uses parallel factor analysis for the determination of bisphenols and their diglycidyl ethers after migration from polycarbonate tableware

Bisphenol A (BPA), bisphenol F (BPF) and their corresponding diglycidyl ethers (BADGE and BFDGE) are simultaneously determined using a programmed-temperature vaporizer-gas chromatography/mass spectrometry (PTV-GC/MS) system. BPA is used in the production of polycarbonate (PC), whereas BADGE and BFDGE are for manufacturing epoxy resins. Several food alerts caused by the migration of this kind of substances from contact food materials have led to the harmonization of the European legislation in Commission Regulation (EU) No. 10/2011, in force from 14 January 2011. In consequence, the use of BPA has been prohibited in the manufacture of plastic infant feeding bottles from 1 May 2011 and from 1 June 2011 regarding the placing on the market and importation into the European Union. Recently, the French Parliament has decreed that the presence of BPA in any food containers will be banned. Similarly, the use and/or presence of BFDGE are not allowed. In this work, a GC/MS method has been developed for the simultaneous determination of BPF, BPA, BFDGE and BADGE. For each one of the I samples that are analyzed, the abundance of J characteristic m/z ratios is recorded at K times around the retention time of each peak, so a data tensor of dimension I×J×K is obtained for every analyte. The decomposition of this tensor by means of parallel factor analysis (PARAFAC) enables to: (a) identify unequivocally each analyte according to the maximum permitted tolerances for relative ion intensities, and (b) quantify each analyte, even in the presence of coeluents. This identification, based on the mass spectrum and the retention time, guarantees the specificity of the analysis. This specificity could fail if the total ion chromatogram (TIC) is considered when there is poor resolution between some peaks or whether interferents coelute. With the aim of studying the effect of shortening the time of the analysis on the quality of the determinations while maintaining the specificity of the identifications, two of the heating ramps in the oven temperature program are changed according to a two-level factorial design. Each analyte is identified by means of a PARAFAC decomposition of a data tensor obtained from several concentration levels, in such a way that five figures of merit are calculated for each experiment of the design. The analysis of these figures of merit for the 16 objects (4 compounds×4 heating ramps) using principal component analysis (PCA) shows that the shortest temperature program should be considered, since this is the one the best figures of merit for BPA and BFDGE (both banned) are achieved with. At these conditions and with probabilities of false positive and false negative fixed at 0.05, values of detection capability (CCβ) between 2.65 and 4.71 μg L(-1) when acetonitrile is the injection solvent, and between 1.97 and 5.53 μg L(-1) when acetone, are obtained. This GC/MS method has been applied to the simultaneous determination of BPF, BPA, BFDGE and BADGE in food simulant D1 (ethanol-H2O, 1:1 v/v), which had been previously in contact with PC tableware for 24h at 70 °C and then pretreated by a solid-phase extraction (SPE) step. The migration of BPA from the new PC containers analyzed is confirmed, and values between 104.67 and 181.46 μg L(-1) (0.73 and 1.27 μg L(-1) after correction) of BPA have been estimated. None of the results obtained exceeds the specific migration limit of 600 μg L(-1) established by law for BPA in plastic food materials different from PC infant feeding bottles. Severe problems of coelution of interferents have been overcome using PARAFAC decompositions in the analysis of these food simulant samples.

Determination of the capability of detection of a hyphenated method: application to spectroelectrochemistry

Abstract A procedure to evaluate the capability of detection of a second-order analytical technique to determine an analyte in presence of an interferent has been proposed taking into account α and β errors in a similar way as ISO norms indicate for the univariate analytical methods. The potentiality of spectroelectrochemistry as a quantitative three-way technique of analysis has been analysed. Trilinearity of spectroelectrochemical data has been studied since it is a necessary condition to apply the trilinear decomposition (TLD) method. As an example, the voltabsorptometric determination of o -tolidine in presence of high concentration of ferrocyanide was chosen to test the applicability of the proposed method. In the same way, the capability of discrimination has been determined. In addition, a second-order standard addition method (SOSAM) has been applied to calculate the concentration of the analyte of interest in the presence of this interferent, avoiding the need to previously identify and determine the quantity of the interferent.