Supervisors: Örebro University Ingrid Ericson Jogsten - Örebro University
Jonas Malmborg – National Forensic Centre
Extraction efficacy of oil samples in
forensic investigations using solid phase
extraction (SPE)
Anna Karlsson Tufuga
VT 2020, 15 HP
Table of content
... 1 Abstract ... 4 Abbreviations ... 4 1. Introduction ... 5 1.1 Aim ... 5 1.2 Demarcations ... 6 2. Background ... 6 2.1 Environmental forensics ... 62.2 Crude oil and Heavy fuel oil ... 6
2.3 Hydrocarbon fingerprint ... 7
2.4 European Committee for standardisation (CEN) method ... 7
2.5 State of knowledge ... 8
2.6 Semi quantification ... 8
3. Method...10
3.1 Gas Chromatography setup ...10
3.2 Materials ...11
3.3 Replicates ...11
3.4 Acquisition function number ...11
3.5 Integration ...13
3.6 Quality assurance/Quality control ...13
3.7 Sample preparation ...14
3.8 Sample extraction...14
3.9 Comparison of different solid phase extractions ...15
4. Intensity of analytes from GC-MS analysis ...16
5. Data evaluation ...16
5.1 PW-plot ...16
6. Result ...16
6.1 Matrix effect on extraction ...17
6.1.1. Soil ...17
6.1.2. Ash ...18
6.1.3 Fabric ...19
6.1.4 Cotton swab...20
6.2 Solid phase extraction efficiency ...20
6.2.1. Silica gel...21
6.2.3. Florisil and Florisil/Na2SO4 ...22
6.3 Blank sample/matrices ...22
6.4 General result ...25
7. Discussion ...25
7.1 Similar studies ...25
7.2 SPE clean-up effect ...26
7.3 Blanks ...26
8. Conclusion ...26
9. Future Perspectives ...27
10. References ...28
Abstract
This study aims to complement the internationally implemented method “CEN/TR 15522-2: 2012 WATERBORNE PETROLEUM AND PETROLEUM PRODUCTS - PART 2” (CEN). It is a method for forensic investigations on oil spill identification using gas chromatography coupled with low resolution mass spectrometry in electron ionisation mode (GC-EI-MS), in single ion monitoring mode (SIM). The method uses hydrocarbon fingerprints and biomarker abundances to compare oils from spill sources with oil from suspected sources. This method is implemented by the national forensic centre (NFC) with their main object to perform and develop forensic investigations for successful law
enforcement.
The experiment uses four different matrices common within the NFC department: wood ash, soil, fabric and cotton swabs. The method evaluates how different sample preparation and clean-up techniques can extract crude oil and heavy fuel oil without losing important information such as the relative abundance of so-called biomarkers typically looked for in international standard praxis in forensic investigations.
In conclusion the implemented CEN method showed a reasonably good extraction from matrixes. Extraction of biomarkers were generally quantitative. Extractions of PAHs worked best in soil and cotton swab matrices. In ash samples, the extraction was not very efficient (between 20-80%). It seems that the PAHs strongly bind to active coal in the ash and cannot be extracted fully. It was also evident that the fabric matrix used was problematic for PAH extraction. The fabric itself seemed to release compounds which interfere with the analysis. In soil samples, 31abR (a biomarker compound) was a reoccurring interference from the matrix. Furthermore, analysis of isoprenoids and alkanes had a very broad analytical variation, seen by that the analytical response for these compounds vary greatly among different samples. SPE-extractions did not work well enough following the protocol in this study to be included as a sample preparation step at the moment. More optimization would be needed before the method could be included as an implemented method.
Abbreviations
NFC- National forensic centre
CEN- European Committee for standardisation HFO- Heavy fuel oil
GC-MS – Gas Chromatography Mass Spectrometer EI- Electron ionization
SIM- Selective ion monitoring SPE- Solid phase extraction
1. Introduction
Oil spills, in oceans and on land has for a long time been a threat to both marine life and animals living at the shores. Especially large oil spills such as the EXXON VALDEZ accident can have
detrimental consequences following long periods of time and not to mention very costly to clean up. [1] It is therefore of greatest interest to allocate the responsible party for the oil spill so that the right offenders are held accountable for the financial liabilities.
In Sweden, it is commonly the coastguard that detects oil spills and carry out sampling for further analysis at the National Forensic Centre (NFC). NFC is a department within the Swedish police. Their main object is to perform and develop forensic investigations for a successful law enforcement. [2] Samples from oil spill events are analysed with the aim to trace the source of the emissions and thus prevent more environmentally hazardous emissions in the future. Together with about 40 different countries NFC has developed a method for analysing these type of oil spills in water, with the aim of tracing the source of the emissions. This method has since been further implemented by NFC in several other areas where different oil types have been analysed on several other matrices. For example, there may be oil spills on the ground from leaking cars or petrol stations, oil residues in ash from fires where there are suspicions of an arson or in sexual offense cases where lubricating oils can be sampled on cotton swabs as a typical matrix.. [3] (Examples from the Swedish National Forensic Centre, oral communication.)
The technical method used by NFC today is gas chromatography coupled with low resolution mass spectrometry in electron ionisation mode (GC-EI-MS), in single ion monitoring mode (SIM). This instrumentation allows focusing on the occurrence and abundance of different biomarkers in different samples. This is fundamental when comparing the chemical composition in samples from crime sites with suspected sources to see if a definite chemical profile match can be made. These types of determinations can be used in an environmental court case as evidence to prosecute a possible responsible party. [4]
1.1 Aim
The aim of this thesis was to complement the internationally implemented “European Committee for standardisation” (CEN) [4] method by evaluating how efficient different sample preparation and clean-up techniques can extract different oils of interest (crude oil and heavy fuel oil). In this study, the experiment was carried out using four different matrices that are commonly investigated at the NFC department: wood ash, soil, fabric and cotton swabs. The method will be evaluated by
examining if this can be performed without losing important information such as the relative abundance of so-called biomarkers typically looked for by NFC and the international standard praxis in forensic investigations.
Specific objectives of the study are to prepare oil samples for GC injection by extraction with dichloromethane (DCM), centrifugation and extraction of supernatant and filtration through glass wool in Pasteur pipettes. These samples will then be compared with the same sample extracts that have been cleaned up by four different solid phase extraction (SPE) methods. The SPE cartridges tested in this study are Alumina, Silica gel, Florisil and dual layer Florisil/Na2SO4. Blank matrices will also be prepared in the same manners to deduct if matrices contain any interferences and if they can be resolved by using SPE clean-up.
1.2 Demarcations
Weathering of samples due to sunlight, wind, rain and other conditions such as in most real cases will not be considered in this study. This study will only consider sample preparation and clean-up under laboratory conditions.
2. Background
2.1 Environmental forensics
In environmental forensic cases such as oil spills in water or on land, sampling may be difficult. There are also many parameters to consider during the analytical procedure since analytes are usually exposed to different weather conditions, which can also make dating of samples difficult. The focus of this study will be on sample extraction and analysis under laboratory conditions.
In this study, oils that are included are crude oil and heavy fuel oil (HFO). The main difference between these two oils is that heavy fuel oil is a condense oil consisting of heavy oil compounds with high boiling points while crude oil contains all components of an oil, even the lighter compounds that are extracted as diesel and gasoline in refineries. Analysing compounds with high boiling points is an advantage in environmental forensics since it decreases the chances of losing analytes and altering sample composition due to evaporation or weathering effects of specific samples before analysis [4]. This method uses GC-EI-MS as analytical technique. GC is an instrument that separates analytes based on their boiling points (or size of the molecule, as larger molecules are heavier and have higher boiling points) via a set temperature programme which creates a chromatographic separation of compounds, which are detected in the MS. In the inner tube of the capillary column there is a stationary phase coated as a film which interacts with compounds and affect their retention times, and a carrier gas which acts as the mobile phase that drives the movement of the compounds through the column. [5] The original CEN (European Committee for standardisation) method is described in Annex B for a low-resolution GC-MS system simply because they are more commonly found in most laboratories to facilitate that all laboratories can follow the same protocol. But a GC high-resolution MS can still be used for higher mass spectrometric resolution, but this instrument is very expensive and not readily available in most laboratories. It is also important to use similar injection concentrations of spill sample and suspected sources to minimise the variance of shifts in diagnostic ratios due to concentration variations. [6]
2.2 Crude oil and Heavy fuel oil
Crude oil is the raw material used in oil refineries for extracting different types of oil, for example gasoline and diesel [7]. HFO is another example of a by-product from oil refineries where different molecular weight hydrocarbons are separated in the distillation procedures for commercial use. [8] Crude oil has a density of between 870-920 kg/m3 [9] and HFO has a maximum density of 1010 kg/m3. The distillation point for HFO is at a very high temperature and is amongst the last products in the oil to evaporate and be extracted in refineries. [10]
When conducting forensic investigations on oils, usually specific biomarkers are searched for in the target sample and compared with a reference sample. Biomarkers are a group of hydrocarbons found in petroleum which are significant for hydrocarbon fingerprinting determination in
environmental forensics. They are complex molecules that are derived from ancient living organisms and can be found in petroleum, rocks and sediments for example. Biomarkers are stable compounds and degrade slightly or not at all from their biogenic precursors. This means they carry information
about origin, source, geological conditions and more, which makes them as perfect molecules for characterization of a sample. [7]
A reference sample is a standard oil which is known to contain almost all considered compounds (see table 1). In this study crude oil was used as the standard oil that the other samples (see tables 3-6 in method for time windows) were compared against, as it contains all compounds in the study except methyl anthracene. By finding a possible sample composition match in both spill samples and in suspected sources, conclusions about the emission source can be supported by this method. Therefore, it is beneficial to look at biomarkers that are heavy and does not evaporate easily. This makes it easier to investigate the “hydrocarbon fingerprint” in which the types of hydrocarbons present in a sample are evaluated. [8]
2.3 Hydrocarbon fingerprint
Hydrocarbon fingerprint analysis is a method for oils, petroleum and other hydrocarbon derived combustion sources, where distinct chemical patterns can be distinguished. These are called the chemical fingerprints of the hydrocarbons and are mainly concluded from thermal separation of compounds in the sample. Therefore, GC is a suitable instrument to use in these types of
investigations where we want to identify the composition of a sample that is thermally stable such as these oils. [8] Polycyclic aromatic hydrocarbons (PAH) for example are a prime example of aromatic hydrocarbon compounds that can reveal much about a sample by the PAH diagnostic ratios. The diagnostic ratios of PAHs depend on the thermal process that were involved in their formation. For example, low molecular weight PAHs are derived from low temperature processes, such as wood burning, and high molecular weight PAHs are derived from high temperature processes such as combustion of fuels in engines. [11] In oils however, PAHs are formed from being contained millions of years below the earth’s crust and exposed to high temperatures and massive pressure [3]. The current standard method for oil fingerprinting analysis is implemented by an international congregation of experts in the field and is standardised by the European Committee for standardisation (CEN). The standardised method is based on a two-pieced approach. First by screening via gas chromatography coupled with a flame ionisation detector (GC-FID) and then confirmatory analysis by gas chromatography – mass spectrometry (GC-MS). These two methods combined allows for a definite analysis of hydrocarbons in oil. [4] [12]
2.4 European Committee for standardisation (CEN) method
The CEN method is based on analysis of the volatile components in both crude oil and HFO that are amenable to GC analysis. In this study there were 44 compounds of interest that were examined, including 15 alkanes and 26 different PAHs, among other compounds listed below in the results section. Alkanes are the simplest family of hydrocarbons, containing hydrogen and carbon atoms only. The boiling points of alkanes increase proportionally with their length. [13]
As mentioned previously these two oils are raw material and by-product in oil refinery processes and they can also contain asphaltenes as non-volatile residues. Asphaltenes are a non-volatile heavy fraction of these oils that are discarded in sample clean-up before analysis. They are preferably avoided as they have been known to clog instruments and cause instrumental damages and wear. However, discarding part of the sample may be problematic since important information about the sample composition might be lost, causing incorrect identification of the oil sample. [12] As of today, sample clean-up is not used regularly in oil-related analysis at the Swedish National Forensic Centre.
However, issues with inconsistent instrument performance has been evident when dealing with oils of the HFO type or oils extracted from complex matrices (Jonas Malmborg, personal communication).
2.5 State of knowledge
Solid phase extraction (SPE) was used as the sample clean-up method in this study. The idea was to retain compounds from a sample onto a sorbent. The sorbent is then commonly washed to remove unwanted compounds and then finally eluted by another solvent and collected. In the present study the washing step after sample addition was excluded as a mechanical filtration was performed by elution with DCM. Having a normal-phase sorbent is advantageous when working with oils due to their low polarity that will allow oils to elute through while retaining polar compounds or soot particles which may occur in heavy fuel oils. Previous studies have shown that fully activated silicic acid and neutral alumina SPE columns have effectively been used when eluting PAHs for GC-MS analysis. Florisil, alumina and Silica SPE column have also been tested to work well with extracting PAHs from soil samples. [14] Compounds with very high boiling points are sometimes not fully eluted from the GC column due to their requirements of very high temperature. This can cause a strong interfering effect on chromatograms as well as deteriorating the overall quality of the GC by affecting inlet system, column and the detector. Crude oils and HFOs both contain asphaltenes and other heavy compounds with high boiling points, therefore the clean-up step with SPE is essential to sustain the quality of the instrument. [6] However, evaluation of asphaltenes was not included in this study.
There is a general need for sample clean-up before running “black oils” such as crude oil through GC-MS analysis due to the column becoming damaged from clogging and wear by for instance soot particles and asphaltenes. In the present study, SPE was utilised as a physical separation method to mainly eliminate these larger particles to prevent them from damaging the injector or clogging the GC-column. [15]
In other studies, both silica and florisil cartridges has been shown to remove these interferences allowing the instrument to extend its lifetime on heavy oil samples. Another common interference may be matrix interferences that are co-eluted unintentionally along with the analytes. Here, alumina and/or silica SPE has been shown to eliminate or minimize those interferences. [6] Separation of compounds in oils by SPE are based on the principle “like dissolves like. Thus, by altering solvent compositions and extraction solvents, specific compounds can be eluted or retained from the sorbent. From previous studies on oils using SPE clean-up, it has been shown that florisil SPE cartridges is able to separate oil samples into aliphatic and aromatic fractions. Separation was
achieved by using DCM and hexane in a mixture (3:1) for extraction. It has been seen in the same study that PAHs are efficiently extracted with a 1:1 ratio of DCM and hexane (and a volume over 6 mL). PAHs which are mid polar compounds and generally have high polarity compared to non-aliphatic hydrocarbons, require high polarity solvents to be extracted. Such as DCM/hexane or DCM/benzene mixtures. The same study also showed that there was a higher background
interference from the SPE sorbent or the SPE tube material. When altering the tube material to glass, the background interferences was greatly decreased. This suggests that there is a great importance in using high quality SPEs as this may significantly affect the results. [15]
2.6 Semi quantification
Semi-quantitative determination without using a calibration curve (perhaps the most common) is possible using a few different approaches. The mutual idea is to integrate peaks and relate them to something other than an external calibration curve. Peaks could be related to e.g. an added internal standard, the weight of the sample or, as in the present study, a stable non-weathered substance
ubiquitously present in oil (30ab which is a hopane). The compound 30ab can be regarded as a non-added internal standard. Hence, the absolute amount of substances present in the sample is not determined, instead focus is directed to their relative abundance. Relative quantification works well when aiming for sample composition comparison, as in oil spill fingerprinting, since the absolute amount/concentration of substances is of little importance and labour-intensive to determine [16].
Table 1. All compounds analysed in the study.
Group Compound m/z ~Retention time Peaks
Alkanes
Alkanes C
10-C
2585
10–40
15
Isoprenoids
Pristane
85
26.3
1
Isoprenoids
Phytane
85
28.3
1
Two-ring biomarkers Ses8
123
23.5
1
PAHs
C2-naphthalenes
156
19.7–21.2
8
PAHs
C1-fluorenes
180
26.3–26.7
5
PAHs
3-methyl-phenantrene
192
30.1
1
PAHs
2-methylphenanthrene
192
30.2
1
PAHs
4-methyldibenzothiophene
198
29.3
1
PAHs
1-methyldibenzothiophene
198
30.1
1
PAHs
benzo(a)fluorene
216
35.2
1
PAHs
2-methylpyrene
216
35.8
1
PAHs
4-methylpyrene
216
36.2
1
PAHs
C1-chrysenes
242
40.8–41.4
4
Tricyklisk diterpanes C23 Tr
191
36.8
1
PAHs
Retene
234
35.5
1
PAHs
Tetra-methyl-phenantrene
234
36.9
1
Norhopanes
27 Tm
191
45.2
1
Norhopanes
29 ab
191
46.8
1
Hopanes
30 ab
191
47.8
1
Homohopanes
31 abS
191
49.0
1
Homohopanes
31 abR
191
49.1
1
Diasteranes
27 dbS
217
41.9
1
Diasteranes
27 dbR
217
42.4
1
Steranes
29 aaR
217
46.9
1
Steranes
27bbR/27bbS (choose higher)
218
44.0
2
Steranes
29bbR/29bbS (choose higher)
218
46.3
2
Triaromatic steranes C21 TA
231
40.1
1
Triaromatic steranes RC28TA
231
48.0
1
3. Method
3.1 Gas Chromatography setup
The CEN method used by NFC provides detailed information for the GC analysis, both regarding peak patterns, approximate abundance, retention times and GC settings (see table 1 below). The
temperature program should be adjusted so that all analytes elute at set retention times. The retention times are held within a range of (+/-) 0.02 min and settings in the temperature programme on the GC-oven are altered to match this. This was achieved by altering the initial hold time (which shifts the entire chromatogram) and by altering the temperature ramp program slightly to find the precise retention time match. [6]
The GC was fitted with a fused silica capillary column (HP-5MS column and (5%-phenyl)-methylpolysiloxane phase) with a length of 30 m, an internal diameter of 0.250 mm and film thickness of 0.25 µm. A 1-meter guard column was coupled to the HP-5MS column. A guard column is a short (1-5 meter) piece of uncoated deactivated fused silica tubing which is placed in-line between the GC injection port and the capillary column. The guard column is used to take the brunt of the contamination/damage from the solvent and sample. The inlet was set to spitless and injection volume set to 1 µL. The temperature programme (described in next paragraph) was fitted so that gammacerane eluted before the final temperature is reached, according to the CEN method. The capillary GC was equipped with an electron pressure control under the GC conditions below. Table 2. GC- settings
Parameter Setting
Oven-programme 42 °C for 1.30 min, XX °C/min to 330 °C (hold time 10 min) Injection 325 °C 1 µL, Splitless
Transfer line temperature 230 °C (instrument dependent) Ion source temperature 230 °C (instrument dependent) Ionisation Electron ionisation 70 eV
Two specific fix points were used to set the correct GC programme, these were 3-methyl
phenantrene (3-MP) with a retention time for its first peak at 30.00 min and hopane (30ab) at 47.80 min. These points were fitted within a maximum limit of (+/-) 0.02 min in variation. Two parameters in the heat programme was altered to achieve this fit. The initial hold time, which causes shift of the entire chromatogram in either direction depending on increased or decreased hold time. As well as temperature ramp setting, in which the initial ramp in the method was 5.5 °C/min and was ultimately altered to 5.52 °C/min. The final hold time was set to 1.22 min. This resulted in a 3-MP peak at 30.01 min and 30ab peak at 47.82 min, which was within the approved range. The total run-time with these settings for one sample was 63.39 min.
3.2 Materials
Crude oil of type Romaskino (Russian), Heavy fuel oil (HFO) type 5, cotton fabric from sweatpants and cotton swabs were obtained from the National Forensic Centre. Pottery soil of brand
“Planteringsjord” was purchased from Plantagen, Sweden. Wood ash from birch tree were collected from a home heating stove.
3.3 Replicates
Three replicates were chosen for this procedure due to fulfilling the following statement from the CEN implemented method: “Duplicate preparation and analysis of samples provides a measure of the handling and analytical portions of the analytical system, while replicate analyses of a single extract provides a measure of the analytical portion of the system.” [6]
Blank extractions with dichloromethane of the different matrices were included to provide information on what is extracted from the matrix itself affecting the results of the analysis.
3.4 Acquisition function number
The mass spectrometer was set up using single ion monitoring in four time window acquisitions to increase total dwell time of target compounds and thus increased sensitivity of the analysis.
Compounds in the table below were included in time window acquisition 1, ranging from 4.0 min to 28.8 min. Note that m/z 85 were included in all windows due to the wide retention range and m/z 191 were included in both time window three and four. The reason why only one m/z was used for each compounds was to lower the total dwell time, and the pattern analysis does not require a further confirming ion.
Table 3 - Acquisition function number 1 of biomarkers from oil analysis using gas chromatography mass spectrometry.
Group Compound m/z Expected
retention time (min)
Peaks
Alkanes
Alkanes C
10-C
2585
10–40
31
Isoprenoids
Pristane (peak after C17)
85
26.3
1
Two-ring
biomarkers
BS10
123
23.5
1
PAHs
C2-naphthalenes
156
19.7–21.2
8
PAHs
C1-fluorenes
180
26.3–26.7
5
In the second time window acquisition, compounds with retention times ranging from 28.8 min to 34.5 min were included, see table 4 below.
Table 4 - Acquisition function number 2 of biomarkers from oil analysis using gas chromatography mass spectrometry.
Group Compound m/z Expected
retention time (min) Peaks
Alkanes
Alkanes C
10-C
2585
10–40
31
PAHs
3-methyl-phenantrene
192
30.1
1
PAHs
2-methyl-phenantrene
192
30.2
1
PAHs
4-methyldibenzothiophene
198
29.3
1
PAHs
1-methyldibenzothiophene
198
30.1
1
In the third time window acquisition, compounds with retention times ranging from 34.5 min to 37.5 min are found. See table 5 below.
Table 5 - Acquisition function number 3 of biomarkers from oil analysis using gas chromatography mass spectrometry.
Group Compound m/z Expected
retention time (min) Peaks
Alkanes
Alkanes C
10-C
2585
10–40
31
PAHs
benzo(a)fluorene
216
35.2
1
PAHs
2-methylpyrene
216
35.8
1
PAHs
4-methylpyrene
216
36.2
1
Tricyklisk
diterpanes
C23 Tr
191
36.8
1
PAHs
Retene
234
35.5
1
PAHs
Tetra-methyl-phenantrene
234
36.9
1
In the fourth and final time window acquisition, compounds with retention times between 39.0 min to end of programme are found. See table 6 below.
Table 6 - Acquisition function number 4 of biomarkers from oil analysis using gas chromatography mass spectrometry.
Group Compound m/z Expected
retention time (min)
Peaks
Alkanes
Alkanes C
10-C
2585
10–40
31
Norhopanes
27 Tm
191
45.2
1
Norhopanes
29 ab
191
46.8
1
Hopanes
30ab
191
47.8
1
Homohopanes 31 abS
191
49.0
1
Homohopanes 31 abR
191
49.1
1
Diasteranes
27 dbS
217
41.9
1
Diasteranes
27 dbR
217
42.4
1
Steranes
29 aaR
217
46.9
1
Steranes
27bbR/27bbS (choose higher)
218
44.0
2
Steranes
29bbR/29bbS (choose higher)
218
46.3
2
Triaromatic
steranes
C21 TA
231
40.1
1
Triaromatic
steranes
RC28TA
231
48.0
1
3.5 Integration
According to the method all compounds (44 chosen to evaluate in this study) were integrated based on peak height. An exception to this were a few compounds which used the peak area of several adjacent peaks or when over-loading of the column stationary phase was at risk due to high
compound concentration. These are C2-naphthalenes, C1-fluorenes, C1 chrysenes, alkanes, pristane and phytane.
3.6 Quality assurance/Quality control
Instrumental performance was assessed by injecting a control sample (hydrocarbon reference sample) which was used to evaluate the chromatographic separation (resolution of peaks).
Resolution was calculated on phytane and pristane by the equation below (Eq.1). This was coupled with a visual evaluation of the previously mentioned compounds and the sample resolution overall.
𝑅𝑅 =
1
𝑡𝑡
𝑅𝑅2− 𝑡𝑡 𝑅𝑅12 (𝑊𝑊
1 + 𝑊𝑊2)Equation 1 . Resolution equation used to calculate resolution of peaks in phytane and pristane. A saturation sample was injected for every 12-15 oil injections to ensured that no discrimination of heavier alkanes occurred in the injector. The abundance of the samples was also continuously checked to ensure the consistency of the signal and no instrumental errors occurred.
Procedure blank sample was also used to be able to detect any possible contamination from external sources. The procedure blank contained pure DCM (Honeywell, Riedel-de Haën, 602-004-00-3) and went through the same method without presence of matrix. Matrix blanks were also prepared and put through the same method without the presence of oils. These blank samples were prepared with similar amount and weight of matrix as the previous samples to achieve as good similarity as possible to previous samples (see table 7). Instrumental blanks were also injected periodically to check for cross contaminations or other contaminations from the instrument.
The analysis of this method does not aim to quantify included compounds but to analyse patterns of compounds normalized against 30ab, to investigate the impact from the different steps on the sample processing.
3.7 Sample preparation
Following the described protocol for the CEN method, approximately 12 teaspoons of each of the four matrices seen below in the table were added to amber glass bottles and their weight was measured to ensure similar compositions in all three replicates. As can be seen in the table below the approximate weight of the soil was 8 g per sample and approximately 6 g per sample of wood ash. For the fabric approximately 10 x 10 cm squares were cut and weighed to approximately 3 g for each sample. The cotton swabs samples consisted of eight cotton ends pealed from the cotton swabs in each sample and had a weight of about 0.25 g.
Table 7. Sample weight of matrices in grams for three replicates of each matrix. Matrices Rep 1
(Crude Oil) Rep 2 (Crude Oil) Rep 3 (Crude Oil) Rep 1 (HFO) Rep 2 (HFO) Rep 3 (HFO)
Soil 8.001 g 8.006 g 8.005 g 8.000 g 8.006 g 8.007 g
Wood ash 5.998 g 6.005 g 6.003 g 6.000 g 5.998 g 6.006 g
Fabric 2.921 g 2.986 g 2.999 g 2.919 g 2.926 g 2.968 g
Cotton
swab 0.246 g 0.246 g 0.249 g 0.237 g 0.246 g 0.237 g
After weighing of samples, approximately 15 drops of each oil were added to each replicate. For HFO between 12-15 large drops, which is approximately equivalent to crude oil volume due to dilution with DCM to enable pipetting of the solid oil which makes the actual volume of HFO less. Three replicates of each matrix were prepared with crude oil and three of the remaining replicates with HFO. As HFO was rather difficult to pipette some DCM was added to dissolve the oil. Once all replicates had been added with oil, all samples were left for a few minutes to ensure evaporation of DCM and that the oil properly adhered to the matrices.
Table 8. Blank matrix sample weight.
Blank matrice Weight
Soil 8.001 g
Wood ash 6.001 g
Fabric 2.981 g
Cotton swab 0.217 g
3.8 Sample extraction
The first step in sample extraction was to add approximately 20 Pasteur pipette volumes of DCM to each replicate for extraction of the oils. Lids were closed and each replicate was shaken by hand and left for 20-30 min to settle. The DCM only was then extracted with Pasteur pipettes into 22 mL glass tubes suitable for centrifugation.
Thereafter centrifugation was carried out for 5 min at 3000 revolutions per minute (RPM) to spin down particulate matter not to be injected into the GC column. The supernatant was extracted again and filtered through glass wool Pasteur pipettes into new 22 mL labelled glass tubes. An aliquot of each sample was transferred to a GC vial for further analysis.
Blank matrices were also extracted in the same manner as the other samples. They were centrifuged for 5 min in 3000 RPM and filtered through Pasteur pipettes with approximately 1 cm of glass wool packing.
The rest of the sample was stored in refrigerator until the next step where they were cleaned-up by SPE. GC vials were also stored in refrigerator/freezer until use, to minimise evaporation of the volatile DCM. Initially all samples were stored in refrigerator but as they were found to evaporate over time they were later switched to freezer.
3.9 Comparison of different solid phase extractions
Four different SPE sorbents were tested in this study:• Fluorisil (magnesiumsilicate) • Dual layer fluorisil /Na2SO4 • Silica gel
• LC Alumina-N Sigma-Aldrich 57087
The above mentioned four pre-packed SPE columns were used in the sample clean-up. A vacuum pump was also used to increase the flow rate through the columns. The flowrate was adjusted at approximately 1-2 drops per second.
For all the SPE types, DCM was used as the extracting solvent. First, the columns were cleaned with three bed volumes of DCM, which was not collected. The DCM was drained from the cartridge until the head of the liquid in the column was just above the filter phase in the cartridge. Then two Pasteur pipette volumes of each sample (approx. 2 mL) was added to each SPE cartridge and collected into new glass-vials. An exception is for some Alumina SPE samples, where four Pasteur pipettes were added to each cartridge in the initial SPE trials, which was then altered to two Pasteur pipettes due to some sample shortages. Additionally, three bed volumes of DCM were added and collected in the same glass vial and finally the extract volume was evaporated to achieve the previous sample concentration that was added to cartridge before the SPE. The samples with four Pasteur pipettes were crude oil std, HFO std, ash HFO (1,2 &3) and fabric HFO (1,2 & 3).
For each SPE sorbent, a crude oil and an HFO standard solution sample was extracted and analysed for comparison with the oil prepared and extracted samples (approximately 15 drops of each oil dissolved in 20 mL DCM – to correspond to extracted volume from matrices). The reason 20 mL DCM was used was due to the glass tubes available, with a total space for 22 mL. All four blank matrices (soil, ash, cotton swab and fabric) were also extracted on each SPE type (see flow chart below). These samples were all treated equally to the other samples in their SPE clean-up.
Figure 1. Flow chart of SPE layout including sample volumes
15 drops HFO + 20 mL DCM
Four matrices & three replicates on each matrice
Four SPE types - two pasteur pipettes of sample in each cartridge
15 drops crude oil + 20 mL DCM
Four matrices & three replicates on each matrice
Four SPE types - two pasteur pipettes of sample in each cartridge
4. Intensity of analytes from GC-MS analysis
Extracted samples were analysed using GC-MS and analyte intensity was compared. For accurate comparison according to the CEN method, analyte response should be within 10% for all samples. Absolute intensity of 30ab was evaluated for all samples. To reach intensities within 10%, (a dilution factor for all samples was calculated and) samples were diluted to reach this range.
Dilution of samples were also performed to reach a final abundance of 1 million for the total ion chromatogram (TIC). Peak heights were calculated to match the smallest height of the sample range. This was found in sample Ash crude oil ALU 2 with a 30ab peak height of 90885 (30ab was used since this is the compound thought to best be extracted in general). This sample also had a TIC of 1,7 x 106. All samples were calculated on which volume to forward to a new vial for dilution to 1 mL (total volume) to achieve the same concentration as the ash crude oil ALU 2. Then a new calculation was made to figure out which factor multiplied with the current abundance would make an abundance of 1 million. This was factor 0.590 and so all new volumes were multiplied with the same factor to achieve an equal abundance for all samples (see Excel file). However, this try was not successful and abundances were found to still vary quite a bit. Perhaps this was due to pipette errors or that DCM had evaporated from the GC vials while in the fume hood or storage (refrigerator/freezer) so that the actual concentration was no longer that which was calculated with.
This measure was performed to avoid overloading the capacity of the analytical column while taking into consideration that analytes with the lowest intensity would not be over-diluted and fall below the detection limit.
5. Data evaluation
5.1 PW-plot
Full descriptions of the GC/MS PW-plot (percentage-weathering plot) can be found in the
literature (CEN 2012; Malmborg 2017)[18]. In short, the plot is a tool for visualization of weathering and identification purposes where all compounds are normalized towards a substance that is a presumed non-weathered substance. The plot functions by normalizing the value of each compound towards the non-weathered compound. Then the normalized value of the weathered spill sample is compared by the normalized value of a non-weathered source sample. The x-axis plots the
compounds in order by retention time and the y-axis shows each compounds intensity, which allows easy detection of differences between compounds. This results in a clear graphic picture for easy detection of compound variance and comparison of multiple samples. If a compound is extracted to 100% this means they are extracted equally well as 30ab.
6. Result
The two standard oils that were prepared containing crude oil dissolved in DCM and HFO dissolved in DCM were analysed and used as references. TIC chromatograms of the oils are shown in Figures 3 and 4 below. The chromatograms show that oil compositions span over the major part of the distillation interval allowing for the validation of various compound classes.
Figure 2. GC-MS chromatogram of the total ion chromatogram of a high fuel oil (HFO) reference sample.
Figure 3. GC-MS chromatogram of the total ion chromatogram of crude oil reference sample.
6.1 Matrix effect on extraction
Four independent matrices were tested for extraction of oils using DCM. Some general observations from the extractions performed are described. Biomarkers (illustrated as red squares in Figures 4-10) showed the best extraction capacity throughout the study. This was seen by the fact that these compounds were often found near the 100 % line in the PW plots. Similar intensities, when
normalised by 30ab, in extracted samples compared to the non-extracted reference sample(s) where achieved as was illustrated by an extraction efficiency close to 100 %. Even in samples where the other compounds were found both well above and below 100 %, biomarkers were quite steadily near 100 % extraction. These results were obtained for both HFO and crude oil.
6.1.1. Soil
As can be seen in Figure 5 the biomarker compounds ( ) was satisfactory extracted from the soil. One biomarker, 31abR, was found at about 130 % in all soil samples. In all the other matrices 31abR was found close to 100 %. This indicates that this particular soil contains an interference of this HFO Std Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200508_042 Scan EI+ TIC 1.79e6 39.48 38.05 36.57 35.02 33.40 31.71 29.95 28.10 26.16 24.12 21.98 19.72 17.33 14.81 12.17 7.19 25.14 26.28 40.87 45.92 44.72 47.08 48.21 49.30 50.36 51.40 52.41 53.38 54.40 55.55 56.85 58.34 Råolja Std Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200506_332 Scan EI+ TIC 1.77e6 19.77 17.37 14.85 12.21 9.49 12.94 16.47 19.21 22.03 20.27 24.17 26.21 25.17 28.14 26.32 29.98 31.74 33.43 35.04 38.07 36.59 40.88 42.21 45.93 44.74 47.10 48.23 51.43
compound that is both co-extracted from the matrix and co-eluted in the GC-analysis when aiming to extract the oil. This was found in both HFO and crude oil for soil samples.
Figure 4. PW-plot illustrating HFO extracted from soil. Marks in the plot illustrates alkanes, biomarkers, isoprenoids, PAHs and triaromatic steranes. 31abR is found furthest to the right (marked by arrow).
It was also found that PAHs ( ) was quantitatively extracted in both HFO and crude oils. However, the alkanes ( ) and isoprenoids ( ) were difficult to reproducibly analyse due to possible
discrimination effect in the liner. Hence, replicates would differ substantially in alkane/isoprenoid content and firm conclusions were difficult to make. Their extraction levels varied quite significantly in most samples, both above and below 100 %, as can be seen in the list of PW-plots (see appendix). Alkanes were commonly found above 100 % in crude oil samples, while in HFO samples the alkanes were spread both above and below 100 %. Some evaporation among the alkanes was also evident. This can be demonstrated as the lighter alkanes often showed a larger variance than the heavier ones.
6.1.2. Ash
It was found that it was difficult to quantitatively extract PAHs from ash samples. In most ash samples the general PAH level was found between 20-80 %. This indicates that ash would be a problematic matrix to analyse extracted PAHs since they seem to adhere to the matrix. It is hypothesized that the ash acts as an active carbon trap for the PAHs, and that these compounds would require further extractive measures to be desorbed from the matrix, e.g. sonification or other solvents. Thus, PAHs should not be included in a hydrocarbon fingerprint to identify matches from a known oil source with oils found in ash. This phenomenon can be seen in the figure below.
Figure 5. PW-plot illustrating HFO extracted from Ash. Marks in the plot illustrate alkanes, biomarkers, isoprenoids, PAHs and triaromatic steranes. 20-80 % recovery of PAHs.
6.1.3 Fabric
The fabric included in this study was found to be a difficult matrix to extract and analyse. In one of the replicates an error seems to have occurred and it could be considered an outlier in comparison with the other two replicates and to the samples analysed using crude oils. Although no apparent explanation for the deviation, this sample was excluded from further analysis and thus the results displayed in Figure 7 include only data from replicate 2 and 3 from extractions of fabric. It was also very apparent it was the first replicate which was inconsistent since the other two gave reasonably similar results. See results below of the PW-plot after excluding the first replicate. There was a co-eluting peak at C23Tr, also found in the blank samples of the fabric. A tendency for low PAH extraction was found in both HFO and crude oil.
Figure 6. PW-plot illustrating HFO extracted from fabric (two replicates). Marks in the plot illustrates alkanes, biomarkers, isoprenoids, PAHs and triaromatic steranes.
6.1.4 Cotton swab
The cotton swabs, generally considered as a non-complex matrix, were found to work well for biomarkers and PAHs. The deviance from 100 % for alkanes and isoprenoids are still considered to be analytical variances.
Figure 7. PW-plot illustrating HFO extracted from cotton swabs. Marks in the plot illustrates alkanes, biomarkers, isoprenoids, PAHs and triaromatic steranes.
6.2 Solid phase extraction efficiency
SPE was evaluated as a clean-up step of extracted oil samples. For comparison, non-matrix extracted oil samples were also extracted using SPE. Four different SPE sorbents were used: silica, alumina, florisil and florisil with sodium sulphate on top. PW-plots of HFO extracted before vs. after the four
different SPEs can be seen in Figures 8-11. PAHs and biomarkers were the most efficiently extracted compounds in all SPE types (near 100% extraction for most compounds). The most problematic group was alkanes which seem to be varying the most in extraction. The outlying biomarker 31abR is marked by an arrow in the HFO extracted soil (in Figure 4) and was not removed during SPE clean-up. Thus, none of the SPE sorbents included in the study were efficient to remove the matrix
interferences of 31abR in soil extracts. One of the main purposes of the solid phase extraction is a physical separation of the physical material extracted in the solvent (soot particles, asphaltenes etc.), while not changing the composition of analytes of interest.
6.2.1. Silica gel
As can be seen from Figure 8, the biomarkers were all extracted to similar extent as 30ab. The PAHs were in average extracted to approximately. 120 %, giving the impression that biomarkers were not fully extracted from the SPE-column using three bed volumes of DCM. However, this was not
repeated using crude oil, where PAH and biomarkers both were extracted in the vicinity of 100 %. For the samples extracted from ash, fabric and cotton swabs, cleaned using silica gel the PAH content varied more. In general, the PAH content increased in these samples compared to the biomarkers, indicating less than 100 % biomarker elution. However, considering the reproducible extraction rate among the biomarker compounds it must be considered unlikely that the extraction rate was much lower than 100 %. In conclusion, it was difficult to draw conclusions about the silica gel elution efficiency of the applied protocol.
Figure 8. PW-plot illustrating HFO oil after vs. before silica gel cleaning. Marks in the plot illustrates alkanes, biomarkers, isoprenoids, PAHs and triaromatic steranes.
6.2.2. Alumina
As for silica gel, the biomarkers elute to a very similar extent in all samples. However, the
interpretation of the PAH content was more complex. In general, the PAH content from the eluting solvent was above 100 % when normalized to 30ab. As mentioned above, the interpretation of alkanes and isoprenoid was very difficult.
Figure 9. PW-plot illustrating HFO oil after vs. before alumina cleaning.
6.2.3. Florisil and Florisil/Na
2SO
4The Florisil samples repeated the patterns of silica gel and alumina, yielding 100 % recoveries for the biomarker compounds, and recoveries >100 % (as normalized to 30ab) for PAHs, alkanes and
isoprenoids. This was repeatable for samples extracted from all matrices and did not differ
substantially between Florisil and Florisil/Na2SO4. Hence, only one example is shown in this report, see Figure 10.
Figure 10. PW-plot illustrating HFO oil after vs. before Florisil/Na2SO4 cleaning.
6.3 Blank sample/matrices
Figures 11-15 illustrates total ion chromatograms (TIC) of blank samples included in the study. The procedure blank (Figure 11) was used as a quality assurance to prove that no compounds came from
cross contamination or from the instrument itself. The matrix blanks were used to deduce which of the compounds in the sample that derived from the matrix itself. The chromatograms plots intensity of the signal on the y-axis and time along the x-axis. Coupled with information about our compounds of interest, this chromatogram can be used to deduct which compounds were present in the sample.
Figure 11. GC-chromatogram of Procedure blank sample (DCM)
Figure 12. GC-chromatogram of soil extracted blank sample (DCM extracted from soil without addition of oil).
The soil blank shows the presence of several substances, especially in the high-boiling point region of the chromatogram. However, co-elution with analytes of interest was scarce in the selected ions and influential only on 29ab and, in particular, 31abR compounds. Thus, it should be possible to compare oil extracted from soil to a potential source if the presence of oil in the sample is sufficiently high so that the smaller baseline disturbances found in some ions could be neglected. Furthermore, it would be recommended to produce a blank sample from soil close to, but not influenced by, the spill to examine the soil at hand.
Blank DCM 200421 Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200421_007 Scan EI+ TIC 4.51e4 57.05 55.63 54.27 53.67 53.37 58.66 60.25 Blank Jord 200421 Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200421_011 Scan EI+ TIC 1.07e5 46.60 26.81 7.26 4.63 19.80 44.71 42.19 39.48 29.00 36.57 42.41 52.07 48.42 48.30 49.65 50.30 50.37 52.71 52.88 54.82 62.88 60.40 56.65
Figure 13. GC-chromatogram of wood-ash extracted blank sample (DCM extracted from wood-ash without addition of oil).
The wood ash blank revealed limited peak occurrence in the ions of interest.
Figure 14. GC-chromatogram of fabric extracted blank sample (DCM extracted from fabric without addition of oil).
The fabric releases many compounds potentially disturbing to analysis. In particular, ions in the PAH class was affected by interfering peaks. This was in accordance with earlier findings at NFC, were PAH compounds are usually hard to interpret in comparative analysis of oils soaked into fabric (Jonas Malmborg, personal communication). In addition, a interfering peak co-eluting with C23Tr was found in this cotton fabric. This means that it could be preferred to mechanically remove oil from fabric, if possible. Extraction of oils from fabric will reduce the probability of finding a successful oil
correlation, due to the interfering matrix.
Blank Aska 200421 Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200421_008 Scan EI+ TIC 3.25e4 58.53 57.28 54.43 53.89 53.80 53.55 53.05 52.23 59.06 61.35 62.63 63.18 Blank Tyg 200421 Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200421_009 Scan EI+ TIC 4.88e4 36.76 36.43 32.00 26.21 19.35 7.26 4.63 11.03 16.39 22.0324.17 26.74 31.02 35.03 55.95 55.90 50.36 48.52 39.48 40.86 42.19 43.47 44.71 47.07 54.91 51.51 58.99 59.60 59.84 62.07 62.94
Figure 15. GC-chromatogram of cotton swab extracted blank sample (DCM extracted from cotton swabs without addition of oil).
The cotton swab blank shows limited disturbance in the regions of interest in the ions. The dominant peaks seen in the chromatogram are likely sterols, extracted from the cotton.
As can be seen in these blank matrix chromatograms above, soil and fabric matrices extracted the most compounds when using DCM for extraction (these blank matrices showed most peaks). Soil blank had the highest intensity at 10 times higher than the fabric blank. This meant that these matrices contributed the most to matrix interferences in this study. In fabric it seems that we have a wide variety of interfering compounds, while in soil it was mostly the later eluting compounds which were extracted by the solvent. All analytes were evaluated in SIM and peak height/area of detected peaks in blank chromatograms were compared against peaks from oil samples.
When analysing the data from the blanks more numerically (see excel file), the majority of alkanes were present in many of the blank matrix samples, including those samples extracted by SPE. However, these were generally at very low levels.
6.4 General result
In general, it was found that the CEN method worked efficiently for extracting HFO and crude oil from different matrices. Most compounds had an efficient extraction similar to 30ab. Biomarkers and PAHs were most efficiently extracted, while alkanes and isoprenoids had the largest variation in extraction. PAHs were found to be bound to the active coal in ash which proved to be a complex matrix for oil extraction. For the SPE, it was seen that this method did not work efficiently and matrix interferences that were found (such as for 31abR) were not removed by the use of different SPE materials.
7. Discussion
7.1 Similar studies
In similar studies on environmental forensic investigations the weathering effect is often considered. As mentioned in the demarcations, this has a great effect in environmental forensic samples and there are many studies that have considered this. A weathering “check” can easily be done by overlaying chromatograms of a spill sample and a suspect source. Another way of checking a sample for weathering is by using a PW-plot which has been used in this study as well. This is done by
Blank Tops 200421 Time 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 % 0 100 LRH-200421_010 Scan EI+ TIC 6.11e4 61.46 61.02 58.99 56.00 54.47 50.36 49.70 47.08 44.71 4.63 7.25 40.8742.19 61.81
integrating the height or area of a spill sample and a suspect source and comparing these in a PW-plot to see how well these two samples match. [19]
The semi-quantification method used in this study was based on normalization towards a non-weathered compound (30ab) compared with a study on bio-oil compounds using external calibration to quantify compounds of interest. This was done using 10 standards, which are plotted in a
calibration curve. By using an external calibration, quantification of analytes can be done with high accuracy. [20]
Another similar study used oil fingerprints for oil spill identification and used a spectroscopic analysis method. This study conducted analyses on asphaltenes which are discarded before analysis in many current methods. It was proven that much information for distinguishing oil samples can be found in the asphaltene phase, and by examining these along with the more volatile fractions, ten different oil samples could be distinguished from each other. It was also found that ATR-FTIR was not able to distinguish weathered oil samples from its source sample, which further demonstrates the significance of asphaltenes in oil fingerprinting. [12]
7.2 SPE clean-up effect
The CEN describes a method with SPE clean-up which use DCM both as sample solvent and
extraction solvent. Normally SPE uses the affinity of solutes to the sorbent to either retain desired or undesired compounds. This allows to separate components of a sample quite efficiently, by choosing a sample solvent which extracts all the compounds of interest. By extracting that sample through the SPE cartridge and then eluting (washing out) with the same solvent, we therefore mainly achieved a physical clean-up of particles that are large enough to get stuck in the filter phase. Therefore, if the aim of the SPE would be to clean up interfering compounds from matrices, which may be smaller than the filter, two different solvents would be needed as sample solvent and as extraction solvent. Or there would need to be a pH (charge) or polarity change of some kind to first extract only the compounds were interested in, in the sorbent and then eluting those with another extraction solvent.
7.3 Blanks
Although alkanes were found at very low levels in both SPE matrix blanks (e.g. aluminium fabric blank) and matrix blanks (e.g. fabric blank), it is quite concerning that several alkanes were present and that it was a continuous trend through the blank samples. It remains unclear where the alkanes present were derived from. A guess is that it might be derived from insufficient washing of SPE vacuum manifolds between samples which has caused a cross contamination of samples.
8. Conclusion
The overall conclusion from this thesis was that the implemented CEN method had a reasonably well extraction performance from various matrices. The extraction efficiency was close to 100 % for many compounds, especially biomarkers, but considerably more variable for some other compounds (e.g. alkanes), which may be partially related to evaporation or injector discrimination problems.
Extractions of PAHs worked best in soil and tops matrices. In ash samples however, the extraction was not very efficient (between 20-80%). It seems that the PAHs interact with the active coal in the ash and cannot be fully extracted. It was also evident that the fabric matrix used was problematic for PAH extraction. It seems that the fabric itself releases compounds which interfere with the analysis. Although this was based on one specific type of fabric, similar problems with PAHs and fabric matrices have been experienced by NFC before (oral communication, Jonas Malmborg). In soil samples it was found that 31abR was affected by a reoccurring interference from the matrix yielding
about 130 % extraction, in comparison to other samples where it had an extraction efficiency of around 100%. Furthermore, the analysis of isoprenoids and alkanes was not sufficient since the analytical variation was very high (especially for samples far from each other in the analytical sequence order). Some deviations may also be due to errors in the laboratory work or instrument. The method also described how SPE can be used to process extracts from oils in different matrices and NFC wished to validate this method in order to perhaps implement it in their analyses.
SPE-extractions were found to be insufficient to be included as a sample preparation step and that results from studies using SPE are quite difficult to compare, especially if they don’t include a clear validation of the extraction. However, all SPE sorbents used displayed similar results in the tests. It was also seen that biomarkers showed a conformity in composition before and after processing in SPE. The conformity for PAHs was however lower, partially due to larger variance but also because most PAHs increased in relation to 30ab in SPE processed samples. The increase of PAHs could possibly be explained by the fact that biomarkers were not eluted fully from the column.
Some problems experienced with this method has been to relate to the intensity interval of 10 % for all samples. Combining GC-vials with various enclosures with the very volatile DCM as solvent it was apparent that evaporation was a major issue difficult to surpass. Eventually a fair way was found to minimize evaporation: keep all vials in freezer and only load a few (soon to analyse) samples at a time onto the GC. Normalising against 30ab was also found a bit problematic in some samples where it was suspected that 30ab was not fully extracted. This causes incorrect normalization which is difficult to determine the exact effect of. By using e.g. an internal standard like some previous studies have, this problem could be subsided as the concentration of the added standard is known [20]. From this study it was apparent that each laboratory should organize, perform and evaluate this method in their own laboratory before use as an implemented method. It was also clear that there are many future perspectives which are advantageous to study (see section below).
9. Future Perspectives
As mentioned above in discussion, the SPE method could be altered by solvent exchanges to achieve a better clean up effect to solve for matrix interferences that were observed in this case. A test on different elution protocols for compounds from the column could also be used, alternatively usage of a larger volume DCM than the three bed volumes used in this study.
Due to the mechanical filtration on samples by SPE that was achieved in this study, perhaps a filtration by particle filters could be used with the same results while achieving a faster sample processing.
It would also be advantageous to compare these results by an analysis of samples analysed closer in time and sample order in the sequence to evaluate any possible effects of this.
10.
References
[1] ‘OilSpillstats_2013.pdf’. Accessed: Apr. 07, 2020. [Online]. Available:
http://www.itopf.org/fileadmin/data/Documents/Company_Lit/OilSpillstats_2013.pdf.
[2] ‘NFC:s uppdrag | Nationellt forensiskt centrum’, nfc.polisen.se.
https://nfc.polisen.se/link/ca077872c73c45bbb6a4677a574c4719 (accessed Apr. 07,
2020).
[3] ‘Jonas Malmborg - NFC, Oral communication [2020-03-10]
[4] S. Stout and Z. Wang, Oil Spill Environmental Forensics Case Studies.
Butterworth-Heinemann, 2017.
[5] C. Poole, Gas Chromatography: Gas Chromatography. Saint Louis, UNITED STATES:
Elsevier, 2012.
[6] ‘CEN/TR 15522-2 : 2012 | OIL SPILL IDENTIFICATION - WATERBORNE PETROLEUM AND
PETROLEUM PRODUCTS - PART 2: ANALYTICAL METHODOLOGY AND INTERPRETATION
OF RESULTS BASED ON GC-FID | SAI Global’.
https://infostore.saiglobal.com/en-us/Standards/CEN-TR-15522-2-2012-332165_SAIG_CEN_CEN_763642/ (accessed Apr.
08, 2020).
[7] ‘Full article: Forensic Fingerprinting of Biomarkers for Oil Spill Characterization and
Source Identification’.
https://www.tandfonline.com/doi/full/10.1080/15275920600667104 (accessed May 21,
2020).
[8] J. C. Raia, C. R. Blakley, A. N. Fuex, D. C. Villalanti, and P. D. Fahrenthold, ‘Evaluation of
Environmental Samples Containing Heavy Hydrocarbon Components in Environmental
Forensic Investigations’, Environ. Forensics, vol. 5, no. 1, pp. 21–32, Mar. 2004, doi:
10.1080/15275920490424015.
[9] ‘Crude Oil Definitions’. http://www.genesisny.net/Commodity/Oil/ODefinitions.html
(accessed Jun. 07, 2020).
[10]
‘HFO’. https://powerplants.man-es.com/fuels/hfo (accessed Jun. 07, 2020).
[11]
M. Tobiszewski and J. Namieśnik, ‘PAH diagnostic ratios for the identification of
pollution emission sources’, Environ. Pollut., vol. 162, pp. 110–119, Mar. 2012, doi:
10.1016/j.envpol.2011.10.025.
[12]
B. J. Riley, C. Lennard, S. Fuller, and V. Spikmans, ‘An FTIR method for the
analysis of crude and heavy fuel oil asphaltenes to assist in oil fingerprinting’, Forensic
Sci. Int., vol. 266, pp. 555–564, Sep. 2016, doi: 10.1016/j.forsciint.2016.07.018.
[13]
‘Alkanes Background’, Chemistry LibreTexts, Oct. 03, 2013.
https://chem.libretexts.org/Bookshelves/Organic_Chemistry/Supplemental_Modules_(O
rganic_Chemistry)/Alkanes/Properties_of_Alkanes/Alkanes_Background (accessed May
21, 2020).
[14]
R. M. Marcé and F. Borrull, ‘Solid-phase extraction of polycyclic aromatic
compounds’, J. Chromatogr. A, vol. 885, no. 1, pp. 273–290, Jul. 2000, doi:
10.1016/S0021-9673(00)00428-3.
[15]
Z. Yang, C. Yang, Z. Wang, B. Hollebone, M. Landriault, and C. E. Brown, ‘Oil
fingerprinting analysis using commercial solid phase extraction ( SPE ) cartridge and gas
chromatography-mass spectrometry ( GC-MS )’, Anal. Methods, vol. 3, no. 3, pp. 628–
635, 2011, doi: 10.1039/C0AY00715C.
[16]
V. Ruiz-Hernández, M. J. Roca, M. Egea-Cortines, and J. Weiss, ‘A comparison of
semi-quantitative methods suitable for establishing volatile profiles’, Plant Methods, vol.
14, no. 1, p. 67, Aug. 2018, doi: 10.1186/s13007-018-0335-2.
[17]
‘About Resolution, Part 1 : SHIMADZU (Shimadzu Corporation)’.
https://www.shimadzu.com/an/hplc/support/lib/lctalk/resol-1.html (accessed May 25,
2020).
[18]
J. Malmborg, ‘Examining the weathering susceptibility of light biomarkers’,
Environ. Forensics, vol. 18, no. 3, pp. 197–206, Jul. 2017, doi:
10.1080/15275922.2017.1340367.
[19]
M. Fingas, ‘Handbook of Oil Spill Science and Technology’, p. 724.
[20]
A. Artigues, N. Puy, J. Bartrolí, and E. Fábregas, ‘Comparative Assessment of
Internal Standards for Quantitative Analysis of Bio-oil Compounds by Gas
Chromatography/Mass Spectrometry Using Statistical Criteria’, Energy Fuels, vol. 28, no.
6, pp. 3908–3915, Jun. 2014, doi: 10.1021/ef5005545.
Appendix
PW-plot (Percentage weathering plot)
Figure 1. PW-plot illustrating HFO extracted from soil.
Figure 3. PW-plot illustrating HFO extracted from fabric (two replicates- Fabric 1 excluded)
Figure 4. PW-plot illustrating HFO extracted from cotton swabs.
Figure 6. PW-plot illustrating HFO extracted from wood ash on silica SPE
Figure 7. PW-plot illustrating HFO extracted from fabric on silica SPE (two replicates)
Figure 9. PW-plot illustrating HFO extracted from soil on florisil SPE
Figure 10. PW-plot illustrating HFO extracted from wood ash on florisil SPE
Figure 12. PW-plot illustrating HFO extracted from cotton swabs on florisil SPE
Figure 13. PW-plot illustrating HFO extracted from soil on alumina SPE
Figure 15. PW-plot illustrating HFO extracted from fabric on alumina SPE (two replicates)
Figure 16. PW-plot illustrating HFO extracted from cotton swabs on alumina SPE
Figure 18. PW-plot illustrating HFO extracted from wood ash on sodium sulphate florisil SPE
Figure 18. PW-plot illustrating HFO extracted from fabric on sodium sulphate florisil SPE (two replicates)
Figure 19. PW-plot illustrating HFO extracted from cotton swabs on sodium sulphate florisil SPE
Figure 21. PW-plot illustrating crude oil from soil
Figure 22. PW-plot illustrating crude oil from wood ash
Figure 24. PW-plot illustrating crude oil from cotton swabs
Figure 25. PW-plot illustrating crude oil from soil on silica SPE
Figure 27. PW-plot illustrating crude oil from fabric on silica SPE (two replicates)
Figure 29. Replicates of HFO standard samples 1-3
