APPI(+)-FTICR mass spectrometry coupled to partial least squares with genetic algorithm variable selection for prediction of API gravity and CCR of crude oil and vacuum residues

Abstract

Positive-ion mode atmospheric pressure photoionization, APPI(+), with Fourier transform ion cyclotron resonance mass spectrometry (FTICR MS) was coupled to a Partial Least Squares (PLS) regression and genetic algorithm variable selection (GA-PLS) methods to estimate the API gravity and Conradson Carbon Residue of Colombian crude oil and vacuum residues (VR) samples. It was observed compositional differences between the crude oils, especially increase in relative abundances of the HC Class with API gravity. Principal Component Analysis (PCA) allowed distinguish crude oils and vacuum residues according to their API gravity value. GA-PLS calibration model provide root mean square error (RMSEC) of 0.13 and 0.33 for API gravity and CCR, respectively. The results here obtained allow to use FT-ICR MS data for quantitative analysis of crude oils and their fractions.