Loading...

N-way partial least squares with variable importance in projection combined to GC×GC-TOFMS as a reliable tool for toxicity identification of fresh and weathered crude oils

Mostafapour, S ; Sharif University of Technology | 2015

994 Viewed
  1. Type of Document: Article
  2. DOI: 10.1007/s00216-014-8076-1
  3. Publisher: Springer Verlag , 2015
  4. Abstract:
  5. In this study, N-way partial least squares (NPLS) is proposed to correlate comprehensive two-dimensional gas chromatography-time of flight mass spectrometry (GC×GCTOFMS) data of different aromatic oil fractions (fresh and weathered) to their toxicity values. Before NPLS modeling, since drift and wander of baseline interfere with information of sought analytes in GC×GC-TOFMS data, a novel method called two-dimensional asymmetric least squares is thus developed for comprehensive correction of the baseline contributions in both chromatographic dimensions. The algorithmis termed comprehensive because it functions to correct the entire chromatogram in both dimensions and it preserves the separation information in both dimensions. In this method, a smoother is combined with asymmetric weighting of deviations from the (smooth) trend to get an effective baseline estimator in both chromatographic dimensions. After baseline correction, the NPLS model was calibrated with 20 oil fractions and evaluated by leave-one-out cross-validation. The number of latent variables was chosen on the basis of minimum root mean squares error of cross validation (RMSECV), and it was 7 (RMSECV=0.073). The developed NPLS model was able to accurately predict the toxicity effects in the five oil fractions as prediction sets which were independent of 20 oil fractions in calibration set (RMSEP=0.0099 and REP= 11.38 %). Finally, the newly developed n-way variable importance in projection (NVIP) was used for identification of the most influential chemical components on the toxicity values of different oil fractions. According to the high NVIP values in both chromatographic dimensions and their corresponding mass spectra, alkyl substituted three-and four-ring aromatic hydrocarbons were identified. It is concluded that multivariate chemometric methods (e.g., NPLS) combined to non-target analysis using GC×GC-TOFMS is a viable strategy to be used for analytical identification in fuel oil studies, with a potential to reduce the number of fractionation steps needed to obtain necessary chromatographic and mass spectral information
  6. Keywords:
  7. Aromatic compounds ; Aromatic hydrocarbons ; Calibration ; Chromatographic analysis ; Chromatography ; Gas chromatography ; Mass spectrometry ; Oil shale ; Smoke detectors ; Statistical methods ; Toxicity ; Ionization of gases ; Asymmetric least squares ; Chemometrics ; Comprehensive two-dimensional gas chromatography ; Oil analysis ; Partial least square (PLS) ; Least squares approximations
  8. Source: Analytical and Bioanalytical Chemistry ; Volume 407, Issue 1 , January , 2015 , Pages 285-295 ; 16182642 (ISSN)
  9. URL: http://link.springer.com/article/10.1007%2Fs00216-014-8076-1