Practical aspects of chemometrics for oil spill fingerprinting

Tiered approaches for oil spill fingerprinting have evolved rapidly since the 1990s. Chemometrics provides a large number of tools for pattern recognition, calibration and classification that can increase the speed and the objectivity of the analysis and allow for more extensive use of the available data in this field. However, although the chemometric literature is extensive, it does not focus on practical issues that are relevant to oil spill fingerprinting. The aim of this review is to provide a framework for the use of chemometric approaches in tiered oil spill fingerprinting and to provide clear-cut practical details and experiences that can be used by the forensic chemist. The framework is based on methods for initial screening, which include classification of samples into oil type, detection of non matches and of weathering state, and detailed oil spill fingerprinting, in which a more rigorous matching of an oil spill sample to suspected source oils is obtained. This review is intended as a tutorial, and is based on two examples of initial screening using respectively gas chromatography with flame ionization detection and fluorescence spectroscopy; and two of detailed oil spill fingerprinting where gas chromatography-mass spectrometry data are analyzed according to two approaches: The first relying on sections of processed chromatograms and the second on diagnostic ratios.     

基于主成分和支持向量机浓度参量同步荧光光谱油种鉴别

基于浓度参量同步荧光光谱技术,对不同溢油类型不同油源原油样品集、引入外扰相似油源样品集进行光谱数据采集,获取其浓度同步荧光光谱矩阵Concentration-Synchronous-Matrix-Fluorescence(CSMF),利用主成分分析方法对两套不同层次的原油相关样品集进行了多类分类识别。结果表明:主成分载荷图可以很好地反映各个原油相关样品在油源上的相似程度,结合支持向量机可以实现不同溢油类型及不同油源原油的准确分类,对于引入风化和海水外扰相似油源溢油样品集,两类分类区分的结果远远高于多类分类识别的结果。通过详细的主成分分析讨论,为溢油油种鉴别提供了一种利用多类分类识别,逐步缩减嫌疑样本数量,最后通过两两分类实现溢油样品准确识别的新思路。     

The Analysis of Environmental Samples by Mass Spectrometry

Mass spectral analysis was carried out on shellfish samples taken in the vicinity of the Arrow oil spill at Chedabucto Bay, Nova Scotia. Oil samples were also analyzed, samples being obtained from the tanker and also from the beach.

After extraction of the oil and shellfish samples, group separations were made into aliphatic, aromatic and oxygenated fractions, which were analyzed separately by mass spectrometry. The validity of the group separations was established, since, in the analysis of the so-called aliphatic fraction, no evidence of aromatics could be seen by mass spectrometry.

On the other hand, polycyclic hydrocarbons were found in some of the aromatic fractions. Identification was based upon standard mass spectrometric analysis already performed on known examples of the polycyclics.

It was concluded that the possibility of oil contamination can be confirmed by the mass spectrometric analysis of appropriate samples. It was also concluded that the so-called control samples should have been taken at a greater distance from the oil spill.     

Oil spill identification by high-speed HPLC

The use of high-speed HPLC in oil spill identification problems has been evaluated in terms of analysis time and reliability. The aromatic fraction was analyzed by reverse-phase chromatography on a 3 μm packing, with detection at 210 and 287 nm, in less than 20 minutes. The profiles exhibited by several Spanish and Middle East crude oils were differentiated by simple statistical parameters. The effect of environmental weathering on the samples has also been investigated. An Arabian light crude oil was still identifiable after four months' simulated marine weathering.     

Identification of oil spills in harbours by means of pattern recognition

For environmental control purposes, floating oil spills in Europoort (Rotterdam) harbours must often be identified and their sources located. Pattern recognition, applied to gas chromatographic data for the spill and for various suspected sources such as oil from bunkers of passing ships and from harbour installations, can lead to definite conclusions, particularly after artificial weathering formulae have been used. The general approach for tackling these problems and the complicating factor of weathering of the oil spills is discussed.     

Characterization and identification of the Detroit River mystery oil spill (2002)

In this paper, a case study of the Detroit River mystery oil spill (2002) is presented that demonstrates the utility of detailed and integrated oil fingerprinting in investigating unknown or suspected oil spills. The detailed diagnostic oil fingerprinting techniques include determination of hydrocarbon groups and semi-quantitative product screening, analysis of oil-characteristic biomarkers and the extended suite of parent and alkylated polycyclic aromatic hydrocarbons (PAHs), and quantitative determination of a variety of diagnostic ratios of "source-specific marker" compounds. The detailed chemical fingerprinting data and results highlight the followings: (1) The spill samples were largely composed of used lube oil mixed with smaller portion of diesel fuel. (2) The diesel in the samples had been weathered and degraded. (3) Sample 3 collected from N. Boblo Island on 14 April was more weathered (most probably caused by more evaporation and water-washing) than samples 1 and 2. (4) All fingerprinting results clearly demonstrated oils in three samples were the same, and they came from the same source. (5) Most PAH compounds were from the diesel portion in the spill samples, while the biomarker compounds were largely from the lube oil. (6) Input of pyrogenic PAHs to the spill samples was clearly demonstrated. The pyrogenic PAHs were most probably produced from combustion and motor lubrication processes, and the lube oil in these spill samples was waste lube oil.     

Identification of traces of oil contamination of biogenic sediments by solid-phase extraction combined with thin layer chromatography

The possibility of identifying industry-related traces of oil contamination in biogenic sediments is investigated on an example of model oil-contaminated peat samples. It is shown that oil pollution of <1 wt % in biogenic samples cannot be identified using a conventional IR spectrometric procedure. An additional study of organic extracts, isolated in sample preparation, by solid-phase extraction and thin layer chromatography can detect specific oil aromatic hydrocarbons in contaminated samples uncharacteristic for background samples of the same type.     

Biodegradation of Crude Oil from the BP Oil Spill in the Marsh Sediments of Southeast Louisiana, USA

The significant challenges presented by the April 20, 2010 explosion, sinking, and subsequent oil spill of the Deepwater Horizon drilling platform in Canyon Block 252 about 52 miles southeast of Venice, LA, USA greatly impacted Louisiana??s coastal ecosystem including the sea food industry, recreational fishing, and tourism. The short-term and long-term impact of this oil spill are significant, and the Deepwater Horizon spill is potentially both an economic and an ecological disaster. Microbes present in the water column and sediments have the potential to degrade the oil. Oil degradation could be enhanced by biostimulation method. The conventional approach to bioremediation of petroleum hydrocarbon is based on aerobic processes. Anaerobic bioremediation has been tested only in a very few cases and is still considered experimental. The currently practiced conventional in situ biorestoration of petroleum-contaminated soils and ground water relies on the supply of oxygen to the subsurface to enhance natural aerobic processes to remediate the contaminants. However, anaerobic microbial processes can be significant in oxygen-depleted subsurface environments and sediments that are contaminated with petroleum-based compounds such as oil-impacted marshes in Louisiana. The goal of this work was to identify the right conditions for the indigenous anaerobic bacteria present in the contaminated sites to enhance degradation of petroleum hydrocarbons. We evaluated the ability of microorganisms under a variety of electron acceptor conditions to degrade petroleum hydrocarbons. Researched microbial systems include sulfate-, nitrate-reducing bacteria, and fermenting bacteria. The results indicated that anaerobic bacteria are viable candidates for bioremediation. Enhanced biodegradation was attained under mixed electron acceptor conditions, where various electron-accepting anaerobes coexisted and aided in degrading complex petroleum hydrocarbon components of marsh sediments in the coastal Louisiana. Significant degradation of oil also occurred under sulfate-reducing and nitrate-reducing conditions.     

A study of a crude oil spill site for selected metal concentrations remediated by a controlled burning in Southwest Louisiana

On August 17, 2002 an oil spill occurred at the Chevron Texaco No. 2 Bayou Tank Battery in the Sabine National Wildlife Refuge, in Southwest Louisiana. The spill covered approximately 3.5 acres of the surrounding marsh. A controlled and in-situ burn of the oil was conducted on August 20, 2002 along with some post-burn “mopping up” on the following day. Within a few days after the burn the growth of new vegetation was observed. A year after the burn, the affected marsh seemed to be no different from the rest of the marsh in the area, as if no spill or burn ever occurred. Analysis of the oil from the No. 2 Bayou Tank Battery showed that there were very low (below detection limit of ICP-OES), if any, concentrations of selected trace metals. Soil cores and grass samples were taken on December15, 2003 and July 15, 2004 from the Sabine National Wildlife Refuge oil spill site for metal determination and compared to cores and grasses taken from the surrounding marsh and confirmed that there was no increase in concentration of any metal that was examined as a result of the oil spill or controlled burn.     

Using GC × GC-FID profiles to estimate the age of weathered gasoline samples

Predicting the amount of time that a petroleum mixture has been exposed to weathering effects has applications in areas of environmental and other forensic investigations, such as aiding in determining the cause and intent of a fire. Historically, research on the evaporation rates of hydrocarbon mixtures has focused on forensic oil spill identification and predicting if a fresh sample could be weathered to give an observed composition in an aged sample. Relatively little attention has focused on approaching the problem from the other direction: estimating exposure time based on the observed composition of a weathered sample at a given time and assuming a prior composition. Here, we build upon our previous research into the weathering of model mixtures by extending our work to gasoline. Samples of gasoline with varying octane ratings and from several vendors were weathered under controlled conditions and their composition monitored over time by two-dimensional gas chromatography (GC × GC). A variety of chemometric models were explored, including partial least squares (PLS), nonlinear PLS (PolyPLS) and locally weighted regression (LWR). A hierarchical application of multivariate techniques was able to predict the time for which a sample had been exposed to evaporative weathering. Partial least squares discriminant analysis could predict whether a sample was relatively fresh (<12 h exposure time) or highly weathered (>20 h exposure time). Subsequent regression models for these classes were evaluated for accuracy using the root mean square error of prediction. LWR was the most successful, whereby fresh and highly weathered samples were predicted to within 30 min and 5 h of exposure, respectively.     

Numerical methods for comparing fresh and weathered oils by their FTIR spectra

Four comparison statistics ('similarity indices') for the identification of the source of a petroleum oil spill based on the ASTM standard test method D3414 were investigated. Namely, (1) first difference correlation coefficient squared and (2) correlation coefficient squared, (3) first difference Euclidean cosine squared and (4) Euclidean cosine squared. For numerical comparison, an FTIR spectrum is divided into three regions, described as: fingerprint (900-700 cm(-1)), generic (1350-900 cm(-1)) and supplementary (1770-1685 cm(-1)), which are the same as the three major regions recommended by the ASTM standard. For fresh oil samples, each similarity index was able to distinguish between replicate independent spectra of the same sample and between different samples. In general, the two first difference-based indices worked better than their parent indices. To provide samples to reveal relationships between weathered and fresh oils, a simple artificial weathering procedure was carried out. Euclidean cosine and correlation coefficients both worked well to maintain identification of a match in the fingerprint region and the two first difference indices were better in the generic region. Receiver operating characteristic curves (true positive rate versus false positive rate) for decisions on matching using the fingerprint region showed two samples could be matched when the difference in weathering time was up to 7 days. Beyond this time the true positive rate falls and samples cannot be reliably matched. However, artificial weathering of a fresh source sample can aid the matching of a weathered sample to its real source from a pool of very similar candidates.     

Bayesian predictive modeling and comparison of oil samples

Statistical comparison of oil samples is an integral part of oil spill identification, which deals with the process of linking an oil spill with its source of origin. In current practice, a frequentist hypothesis test is often used to evaluate evidence in support of a match between a spill and a source sample. As frequentist tests are only able to evaluate evidence against a hypothesis but not in support of it, we argue that this leads to unsound statistical reasoning. Moreover, currently only verbal conclusions on a very coarse scale can be made about the match between two samples, whereas a finer quantitative assessment would often be preferred. To address these issues, we propose a Bayesian predictive approach for evaluating the similarity between the chemical compositions of two oil samples. We derive the underlying statistical model from some basic assumptions on modeling assays in analytical chemistry, and to further facilitate and improve numerical evaluations, we develop analytical expressions for the key elements of Bayesian inference for this model. The approach is illustrated with both simulated and real data and is shown to have appealing properties in comparison with both standard frequentist and Bayesian approaches. Copyright © 2013 John Wiley & Sons, Ltd.     

Chemical fingerprinting of petroleum biomarkers in biota samples using retention-time locking chromatography and multivariate analysis

This work was conducted to study a new separation and evaluation approach for the chemical fingerprinting of petroleum biomarkers in biota samples. The final aim of this work was to study the correlation between the observed effects in the shore habitats (mussels and limpets) and one pollution source: the oil spill of the Prestige tanker. The method combined a clean-up step of the biota extracts (mussels and limpets), the retention-time locking of the gas chromatographic set up, and the multivariate data analysis of the chromatograms. For clean-up, solid-phase extraction and gel permeation chromatography were compared, and 5g Florisil cartridges assured the lack of interfering compounds in the last extracts. In order to assure reproducible retention times and to avoid the realignment of the chromatograms, the retention-time locking feature of our gas chromatography-mass spectrometry (GC-MS) set up was used. Finally, in the case of multivariate analysis, the GC-MS chromatograms were treated, essentially by derivatization and by normalization, and all the chromatograms at m/z 191 (terpenes), m/z 217-218 (steranes and diasteranes) and m/z 231 (triaromatic steranes) were treated by means of principal component analysis. Furthermore, slightly different four oil samples from the Prestige oil spill were analyzed following the Nordtest method, and the GC-MS chromatograms were considered as the reference chemical fingerprints of the sources. In this sense, the correlation between the studied samples, including sediments and biota samples, and the source candidate was completed by means of a supervised pattern recognition method. As a result, the method proposed in this work was useful to identify the Prestige oil spill as the source of many of the analyzed samples.     

Evolution of hydrocarbons and bacterial activity in the marine sediments contaminated by crude oil overflow and treated

The fate of an experimental oil pollution of intertidal sediments in a sheltered beach of North Brittany (France) has been investigated over a 16-month period. Chemical treatments were applied to two of the three contaminated plots by pre-mixing oil respectively with dispersant and biodegrading agents. The physico-chemical and bacteriological characteristics of the polluted areas were followed with the purpose of identifying the limiting parameters for oil microbial degradation and the effect of treatment. The concentration of hydrocarbons in the oiled sediments did not change significantly during the experimental period. Spectrofluorimetric and chromatographic data showed that the main evolution of oil concerns the degradation of n-alkanes and the removal of light aromatics. Biodegradation of hydrocarbons occurred at a measurable rate only during the warm seasons (average temperature 18 +/- 2 degrees C) causing after sixteen months the disappearance of more than 80% of the n-alkanes fraction independently of the pollution sediment level and the chemical treatment of the experimental plots. However, the biodegradation of n-alkanes proceeded during the first months, at different rates, inversely depending on oil content in the collected samples. The main limiting factor is dissolved oxygen according to the fact that spilled oil was located at 3-5 cm depth in a poorly oxygenated zone characterized by low redox potential. Nutrients were not a limiting factor probably due to domestic and agricultural inputs in this area. A marked bacterial growth was observed two weeks after the oil spill with a relative increase in hydrocarbon degrading bacteria with respect to total heterotrophs. Degradation rates, based on C14 n-hexadecane experiments, seem to follow the same way than specific bacterial counts (plate technique). Specific bacteria are always high at the end of our 16 months' field experimentation. In the laboratory as well as in the field experiments, the same behaviour of untreated and chemically treated oil was observed in partially anaerobic sediment.     

Spectrochemical study for the in situ detection of oil spill residues using laser-induced breakdown spectroscopy

Laser-induced breakdown spectroscopy (LIBS) has been used to identify the differences or similarities between crude oil and fuel residues. Firstly, a man portable LIBS analyzer was used for the on-site environmental control and analysis of the oil spill from The Prestige. An exhaustive analysis of crude oil and oil spill residues (collected during the field campaign in the Galician Coast) was performed in the laboratory. Characteristics elements in petroleum such as C, H, N, O, Mg, Na, Fe and V were detected. In addition, contributions from Ca, Si and Al in the composition of residues have been found. The use of intensity ratios of line and band emissions in the original fuel (crude oil) and in the aged residues allowed a better characterization of the samples than the simple use of peak intensities. The chemical composition between the crude oil and the fuel residues was found completely different. As well, a statistical method was employed in order to discriminate residues. Although significant differences were observed, no conclusions in terms of age and provenance could be reached due to the unknowledgment in the origin of the samples.     

Geochemical characterisation of Antarctic soils and lacustrine sediments from Terra Nova Bay

In Terra Nova Bay region (northern Victoria Land, Antarctica), the interactions among soil, meltwater and lakes are poorly understood with regard to the physicochemical transformations that occur when solid materials are exchanged among them. In order to ascertain the natural dynamics of several elements, namely Al, As, Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, P, Pb, Ti and Zn, as well as the perturbations by human activities and/or global contamination, soils and sediments from four catchments (Tarn Flat, Edmonson Point, Inexpressible Island and Northern Foothills) were studied.To accomplish the above mentioned objectives, the particle-size distribution and mineralogical composition, as well as the total C, H and N contents and the total concentrations of the investigated elements were determined. Finally in order to assess metal mobility, the modified BCR three-step sequential extraction procedure was applied to the samples and the partitioning of five metals, namely Cr, Fe, Mn, Pb and Zn, into different fractions was determined.Lithology and topography are the main factors that differentiate the physicochemical characteristics of the soils and sediments investigated. The lacustrine sediments and the soils sampled around the lakes showed a similar mineralogical and chemical composition, as evidence that rock- and soil-weathering processes occur primarily during the warm summer periods, when the lakes are partially ice-free. Nevertheless the presence of chlorite in these systems indicated that chemical weathering has taken place even in these extreme conditions. In general, the geochemical composition of soils and sediments in this Antarctic area seems to be influenced mainly by the nature of the bedrock and by the input of sea-spray, and, to a lesser degree, by mechanical and chemical weathering as well as by biological processes; no evidence of a local or global anthropogenic contamination was found.     

Characterization of polycyclic aromatic hydrocarbons in environmental samples by selective fluorescence quenching

Selective fluorescence quenching is used to profile polycyclic aromatic hydrocarbons (PAHs) in samples of environmental origin. After separation by high-efficiency capillary liquid chromatography, the PAHs are detected by laser-induced fluorescence spectroscopy. Nitromethane is added to selectively quench the fluorescence of alternant PAHs, whereas diisopropylamine is added to quench nonalternant PAHs. The chromatograms in the absence and presence of fluorescence quenching are evaluated by means of the product moment correlation method to quantify the statistical similarities and differences. This method is demonstrated by application to three samples: a standard mixture of 16 priority pollutants, a coal-derived fluid, and a contaminated soil. The correlation coefficients (r) are typically 0.99 or higher for samples that are identical in origin, 0.90-0.50 for closely related samples, and less than 0.50 for samples that are distinctly unrelated. This method can be used to confirm with high statistical confidence the cause or source of an event with environmental impact, such as an oil leak or spill, contamination or waste by-products from petroleum fuel production and processing, etc.     

Infrared and 13C MAS nuclear magnetic resonance spectroscopic study of acetylation of cotton

The acetylation of commercial cotton samples with acetic anhydride without solvents in the presence of about 5% 4-dimethylaminopyridine (DMAP) catalyst was followed using Fourier transform infrared (FTIR) and 13C MAS NMR spectroscopy. This preliminary investigation was conducted in an effort to develop hydrophobic, biodegradable, cellulosic materials for subsequent application in oil spill cleanup. The FTIR results provide clear evidence for successful acetylation though the NMR results indicate that the level of acetylation is low. Nevertheless, the overall results indicate that cotton fibres are potential candidates suitable for further development via acetylation into hydrophobic sorbent materials for subsequent oil spill cleanup application. The results also indicate that de-acetylation, the reverse of the equilibrium acetylation reaction, occurred when the acetylation reaction was prolonged beyond 3 h.     

Surfactants treatment of crude oil contaminated soils

This study reports experimental measurements investigating the ability of a biological (rhamnolipid) and a synthetic (sodium dodecyl sulfate, SDS) surfactant to remove the North Sea Ekofisk crude oil from various soils with different particle size fractions under varying washing conditions. The washing parameters and ranges tested were as follows: temperature (5 to 50 degrees C), time (5 to 20 min), shaking speed (80 to 200 strokes/min), volume (5 to 20 cm3), and surfactant concentration (0.004 to 5 mass%). The contaminated soils were prepared in the laboratory by mixing crude oil and soils using a rotating cylindrical mixer. Two contamination cases were considered: (1) weathered contamination was simulated by keeping freshly contaminated soils in a fan assisted oven at 50 degrees C for 14 days, mimicking the weathering effect in a natural hot environment, and (2) nonweathered contamination which was not subjected to the oven treatment. The surfactants were found to have considerable potential in removing crude oil from different contaminated soils and the results were comparable with those reported in literature for petroleum hydrocarbons. The removal of crude oil with either rhamnolipid or SDS was within the repeatability range of +/-6%. The most influential parameters on oil removal were surfactant concentration and washing temperature. The soil cation exchange capacity and pH also influenced the removal of crude oil from the individual soils. However, due to the binding of crude oil to soil during weathering, low crude oil removal was achieved with the weathered contaminated soil samples.