Comparison of Near-IR, Raman, and Mid-IR Spectroscopies for the Determination of BTEX in Petroleum Fuels

Abstract

We report for the first time a direct comparison of the three most common vibrational analysis techniques for the determination of individual BTEX components (benzene, toluene, ethylbenzene, ortho-xylene, meta-xylene, and para-xylene) in blended commercial gasolines. Partial least-squares (PLS) regression models were constructed for each BTEX component by using each of the three spectroscopic techniques. A minimum of 120 types of blended gasolines were used in the training set for each BTEX component. Leave-one-out validation of the training sets yields lower standard errors for Raman and mid-IR spectroscopies when compared to near-IR for all six BTEX components. In general, mid-IR has slightly lower standard errors than Raman. These trends are upheld when the models are tested by using independent test sets with a minimum of 40 types of blended gasolines (all of which differ in composition from the training set).