PLS1 Algorithm Research Articles

Evaluation of Chemometric Models in an FTIR Study of the Gas Phase During Atmospheric-Pressure CVD of Tin Oxide Thin Films

The potentially fast data acquisition capability and high resolution of the Fourier transform infrared (FTIR) spectrometer makes it an attractive tool for in-situ analysis of CVD systems. However, real-time extraction of useful quantitative information from the spectra recorded at medium (4 cm-1) or high (0.5 cm-1) resolution can be a difficult and time-consuming process. In this study, we evaluate the performance of various chemometric models used to decompose spectral data recorded during a quantitative study of the dimethyltin dichloride (DMTC) and oxygen (tin oxide) CVD process. The chemometric models investigated were classical least squares (CLS), principal component analysis (PCA), principal component regression (PCR), and partial least squares (PLS). Preliminary findings indicate that the PLS-1 algorithm is the most promising candidate for accurately determining individual constituent concentrations from the complicated gas mixture inherent in this CVD system.

FT-IR Spectroscopy for the in-Flame Study of a 15 MW Dual Fuel Gas/Oil Burner

An FT-IR (Fourier transform infrared) analyzer has been configured for combustion research and allows the simultaneous quantification of CO, CO2, CH4, C2H6, C2H2, C2H4, NO, and N2O concentrations in the flame of natural gas burners. The calibration has been carried out by the multivariate PLS1 algorithm, with calculated standard spectra. For the multivariate CO/CO2 model, special attention has been given to nonlinear effects caused by interference in the 2210–2250 cm−1 region. For this model, two methods of calculating mixture spectra and their influence on prediction parameters are evaluated. Furthermore, the sensitivity of the CO/CO2 model for N2O interference has been tested. The presented results show that conclusions drawn from statistical prediction parameters should be treated cautiously in models that have to deal with spectral interference.

Experimental in-flame study of a 15 MW dual fuel gas/oil burner

Extensive in-flame measurements have been carried out on a 15 MW dual fuel burner, in order to obtain more information on the influence of different burner head configurations on the profiles of chemical species in the flame. Together with stack measurements of NO x concentrations, this will improve our knowledge on the formation of NO x in this installation. In a half plane of the installation at four axial distances radial concentration profiles were recorded. For these measurements an FTIR analyser has been configured to allow the simultaneous quantification of CO, CO 2, CH 4, C 2H 6, C 2H 2, C 2H 4 and NO concentrations on a continuous base. The calibration of the analyser has been carried out by the multivariate PLS1 algorithm, with calculated standard spectra, which avoids elaborate lab calibration procedures. The comparison of concentration profiles of CH 4 and C 2H 6 close to the burner mouth for different burner head configurations, gives a good indication of how the fuel is injected in the installation and together with the C 2H 4, C 2H 2 concentrations gives more information about the combustion itself.

Optimization of an automated FI‐FT‐IR procedure for the determination ofo‐xylene, toluene and ethyl benzene in n‐hexane

The development and optimization of an automated flow injection (FI) manifold coupled with a Fourier transform infrared (FT-IR) detector for the determination of toluene, ethyl benzene and o-xylene in an n-hexane matrix is described. FT-IR parameters optimized were resolution and number of co-added scans; FI parameters optimized were type of pump tubing, carrier flow rate and sample volume. ATR and transmission flow cells were compared for the determination of o-xylene, the ATR cell was easier to use and gave better figures of merit, except for sensitivity, for which the transmission cell was twice as good. Multivariate calibration routines were applied to the FI-FT-IR data and the PLS1 algorithm gave relative root mean standard errors of crossvalidation (RRMSECVs) < 7% for all three analytes using mean-centred data and the first derivative for o-xylene.

Simultaneous determination of the colorants tartrazine, ponceau 4R and sunset yellow FCF in foodstuffs by solid phase spectrophotometry using partial least squares multivariate calibration

A method is proposed for the simultaneous determination of the colorants tartrazine (TT), sunset yellow FCF (SY) and ponceau 4R (PR) in foods. The colorants were fixed in Sephadex DEAE A-25 gel at pH 2.0 and then packed in a 1-mm silica cell. The spectra of the analytes fixed in the solid support were recorded between 400 and 800 nm against a blank and the partial least squares (PLS) multivariate calibration was used to obtain the results. The linear dynamic ranges of the calibration graphs were from 50.0 to 650.0 ng ml −1 for the three colorants and these ranges were taking in account the optimisation of the calibration matrix using the PLS-1 algorithm. The experimental results showed that the optimum number of factors for the calibration matrix was four in all instances and the residual means standard deviation was 5.5267 for SY, 6.3878 for TT and 6.9816 for PR. The square of the correlation coefficients were 0.9977, 0.9978 and 0.9954 for SY, TT and PR respectively. The method was applied to the determination of the colorants in foods and results were compared with those obtained by means of HPLC as reference method. The results obtained can be considered as acceptable in most cases (eight of nine commercial samples). The relative standard deviations ranging between 0.5 and 10.8 for the commercial samples analysed.

Quantification of two chromatographic unresolved dichlorophenols using gas chromatography-direct deposition-fourier transform infrared spectrometry and multivariate calibration

A method for the quantification of two chromatographically unresolved dichlorophenol isomers in water is described. Acetylation and concentration on graphitized carbon cartridges are carried out as a preliminary step. Detection is made by gas chromatography (GC) coupled to Fourier transform infrared spectroscopy (FTIR), using a direct deposition interface (DD). Infrared spectra in the maximum of the unresolved peak of 2,5- and 2,4-dichlorophenol for a series of standards with different amounts of these two compounds are used, to elaborate a multivariate calibration model (PLS-1 algorithm). By the method described, concentrations of dichlorophenol isomers in water at ng/ml level can be determined.

Learning control strategies for chemical processes: a distributed approach

The design of a distributed learning system (DLS) which combines the features of instance-space and hypothesis-space methods is described. This algorithm decomposes a data set of training examples into subsets. After applying an inductive learning program on each subset, it synthesizes the results using a genetic algorithm. It is shown that this parallel distributed approach is more efficient, since each inductive learning program works on only a subset of data. Since the genetic algorithm searches globally in the hypothesis space, this approach gives a more accurate concept description. The implementation of DLS in Common LISP is discussed, and its distributed approach is compared to C4.5 and PLS1 algorithms.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>