Petrophase 2014 Statistical analysis of crude oils and crude oil mixtures by LDI FTICR mass spectrometry

• Average mass list and logarithm scaling is preferred for PCA to separate oils in scoring plot. Fig. 1: a) Zoom-in of LDI spectra of three different crude oil samples. b) Zoom-in of LDI spectra of three repetitive measurements showing the good reproducibility of the measurement. Fig. 2: Class distribution plot of different crude oils detected as radical cations and protonated species [H] by LDI-FTICR mass spectrometry; oil 8 and oil 8a are identical oils from different bottles; oil 9a is oil 9 one day in ion source at 3 mbar. observed with logarithm scaling than with normal scaling. However, the statistical analysis also indicated that extremely high resolution of the mass spectra is needed for the separation of the oils by PCA (Figure 4). Using bucketing with mass resolution of 2 mDa for PCA calculation instead of average mass lists resulted already in efficient separation of all crude oils. Separation of the crude oils was impossible with a mass resolution of 20 mDa bucketing for PCA simulating mass spectrum with resolving power of only 20.000. Standard deviation was low for repetitive LDI-FTICR mass measurements.