Prediction of total acid number in distillation cuts of crude oil by ESI(−) FT‑ICR MS coupled with chemometric tools

Abstract

Competitive adaptive reweighted sampling-partial least squares (CARS-PLS) and negative-ion mode electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry (ESI(−) FT-ICR MS) data were adopted to assess the total acid number (TAN) of crude oil distillation cuts. Two crude oil samples and 24 derivatives with TAN ranging from 0.20 to 0.39 mg of KOH g-¹ were investigated. The multivariate calibration PLS model was built with 18 calibration samples and tested with 8 validation samples. CARS-PLS reduced the number of variables from 1610 to only 4, allowing the identification of molecular formulas that are truly related to the TAN. The root mean square error of prediction (RMSEP) obtained was 0.01 mg of KOH g-¹, which is lower than the error when using all variables (0.03 mg of KOH g-1). Finally, it was observed that the N and O2 compound classes are the most important classes for providing a better correlation between ESI(−) FT-ICR mass spectra and TAN values.