Analytica Chimica Acta 573–574 (2006) 328–332 Characterization of oil spills in the environment using parallel factor multiway analysis Vassilis Gaganis ∗ , Nikos Pasadakis Mineral Resources Engineering Department, Technical University of Crete, Chania, Greece Received 30 November 2005; received in revised form 12 March 2006; accepted 20 March 2006 Available online 29 March 2006 Abstract The aim of this study was to characterize samples of petroleum spills derived from the oily free-phase zone located in the subsurface of a petroleum refinery and to reveal the contained distinct petroleum fractions, thus enabling the identification of the spill origin. The samples were collected from different monitoring wells and were analyzed using gel permeation chromatography (GPC) combined with a UV-diode array detector. The PARAFAC algorithm was employed for the analysis of the 3-D experimental data matrix, which contained the areas under the chromatographic trace, measured for distinct time slices over the 270–440 nm UV range for the whole sample population. The application of the PARAFAC method revealed two significant elution profiles possessing characteristic UV signals, which were attributed to the gasoline and diesel fractions, respectively. A third elution profile was also identified which corresponded to biodegraded heavy fractions. The relative contribution of these compositional features to the oil spill samples was also identified. The presented method can be employed as a rapid and reliable fingerprinting tool in environmental studies, where petroleum pollutants of unknown composition are expected. © 2006 Elsevier B.V. All rights reserved. Keywords: Oil spills; Parallel factor analysis; Environment 1. Introduction Oil spills in the subsurface occur during production, trans- portation and treatment of petroleum and its products. Their environmental impact depends on the type and the amount of oil as well as on the local subsurface conditions. Accurate and detailed compositional analysis of the non-aqueous free phase, that can be formed during oil spills, is a prerequisite for understanding the behavior and the fate of the pollutants in the local environment, for designing an effective clean-up strategy and for evaluating the efficiency of the remediation pro- cesses. The characterization of the pollutants, usually referred as fingerprinting, is an integrated analytical and data processing methodology, aiming at revealing the affiliation of the contam- inants to a group of chemically similar objects and at identi- fying the pollution source. Several fingerprinting methodolo- gies and applications have been demonstrated in the literature during the past 15 years [1–11] aiming at the characteriza- ∗ Corresponding author. Tel.: +30 28210 37692. E-mail address: gaganis@mred.tuc.gr (V. Gaganis). tion of pollutants and the allocation of oil spills sources in the environment. Oil spills in the subsurface of a refinery constitute complex mixtures of different petroleum fractions. The identification of their main constituents is an extremely difficult analytical work since the nature and the biodegradation level of the samples is often unknown and therefore the use of common analytical tech- niques employed in fingerprinting applications like gas chro- matography is problematic. In the current study, it was decided to employ the gel permeation chromatography (GPC) technique with ultra violet diode array detector (UV-DAD) which ensures sensitive detection and complete elution of the existing compo- nents while avoiding the commonly observed in gas chromatog- raphy incomplete analysis and column contamination. The aim of the analysis was the determination of a set of meaningful spec- tra and the corresponding elution profiles of the constituents thus enabling their identification. The GPC UV-DAD analysis produces a 2-D matrix sized N T × N W , where N T and N W denote the number of time seg- ments along the elution profile and the number of wavelengths of the spanned range, respectively. When multiple samples have to be analyzed together a 3-D matrix structure comprising of 0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.03.071