Interlaboratory Comparison of Quantifying Hydrocarbons and Trace Elements in Sediment Samples from a Tropical Estuary

The Marine and Environmental Studies Laboratory at PUC-Rio, in partnership with the PETROBRAS R&D Center (Brazil), conducted an interlaboratory comparison of aliphatic and aromatic hydrocarbons and trace element determinations. The goal of this study was to produce a homogeneous and stable material and obtain consensus concentration values for its use as a reference material (RM) for sediments from tropical regions. The prepared material is adequate for use as a reference material because it has remained stable (Test F, ANOVA, p < 0.05) since its preparation in 2007. In this study, several alkylated polycyclic aromatic hydrocarbon (PAH) compounds were included, and the resulting dispersion was the same order of magnitude as the parental PAH (16 PAH USEPA).


Introduction
Polycyclic aromatic hydrocarbons (PAHs), n-alkanes, and other aliphatic hydrocarbons, have been used as indicators of oil and incomplete combustion residues in the marine environment.In addition, PAHs are considered as contaminants of special concern due to the carcinogenic and mutagenic properties of some of its components. 1,2etals and metalloids are components that occur naturally in the environment.In addition, metals and metalloids have historically been released in the environment due to human activities.Unlike organic pollutants, which can be degraded into less harmful components through biological or chemical processes, trace elements are non-degradable.Thus, trace elements can accumulate in sediments and act as sources of natural or anthropogenic disturbances.The effects of metal pollution in the environment can be substantial and long lasting.More importantly, toxic metals can enter the food chain and adversely affect humans when contaminated seafood is consumed. 3he process of quantifying organic and inorganic trace substances is complex because they are present in environmental matrices at low concentrations and include a broad spectrum of components or species that must be identified and quantified individually.In addition, environmental matrices are complex and require the use of careful procedures and methods with very low detection limits.Thus, laboratory intercomparisons represent an essential method for validating methods and verifying the accuracy of results.Accuracy and precision are critical when analyzing hydrocarbons and trace-metals in environmental samples when assessing and monitoring coastal and/or ocean environments.
As in many other countries, laboratories in Brazil depend on reference materials (RMs) prepared in North America, Europe or Japan to apply such quality control.In addition to the inherent barriers for importing them, these RMs are based on the matrices of temperate climates and often have geochemical characteristics that differ from those of tropical regions.
The goal of this study was to establish the capacity of Brazil for producing reference materials for petroleum hydrocarbons and trace elements in an environmental matrix, specifically from a tropical region.In addition, this study describes and evaluates the steps that were taken to obtain the reference material SED001.Once prepared and established, the consensus values of these materials can be made available to universities, research centers and commercial laboratories.The prepared sediment sample (SED 001) was distributed to several regional and international laboratories in February 2008.The procedures used for preparing this material were based on ISO Guide 35 4 and Villeneuve et al. 5 The preparation steps for the reference material included the selection of a sampling site containing sediments with the characteristics of interest (presence of hydrocarbons and trace-elements at detectable quantities), particulate size reduction, homogenization and distribution in glass bottles (the three last steps were performed by CETEM Mineral Technology Center, RJ, Brazil).
For trace elements, the participant laboratories were requested to report the individual metal concentrations using three different digestion methods: (i) 1.0 mol L -1 HCl -a milder extraction that provides almost all bioavailable elements; 6 (ii) EPA 3050b 7 (concentrated HNO 3 treatment) -which does not allow total digestion but also releases bioavailable elements; and (iii) EPA 3052 8 (HNO 3 + HF) -which performs total digestion.The targeted elements included Mg, Al, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Sr, Cd, Sn, Sb, Ba, Hg and Pb.The trace elements were chosen based on their environmental importance and the interests of the petroleum industry.Three different methods were provided so that each laboratory could perform the trace element determination using at least one method.
The requested methodology for determining hydrocarbons and trace elements would be established based on the routines of the respective participating laboratories.

Sample preparation
A sufficiently large amount of sediment (130 kg wet weight) was collected from Guanabara Bay, Rio de Janeiro, after screening the initial sampling site (coordinates: -43.087910 and -22.765130).The material was collected using a Van Veen sampler in containers that were previously decontaminated using acetone and dichloromethane.The collected material was stored in a refrigeration unit until the beginning of the process steps.To this batch, the following preparation procedures were applied: freezedrying, grinding, and sieving through a 75 µm stainless steel sieve.The fraction smaller than 75 µm was homogenized in a stainless steel rotating drum (200 L), followed by stepwise separation and packing using a stainless steel rotating separator with 12 collectors.The obtained aliquots contained approximately 40 g of sediments.These aliquots were stored in glass bottles with aluminum screw caps and sealed with Teflon tape.

Homogeneity test
The homogeneity of the material was initially assessed by determining the concentrations of the hydrocarbons (n = 10) and trace elements (n = 13) in the sediment samples from different glass bottles.The F test was used to evaluate the homogeneity between the different bottles and within each bottle (analysis of variance (ANOVA); software SPSS 8.0). 4 A preliminary PAH analysis was conducted at the Laboratory of Marine and Environmental Studies (LABMAM) using EPA method 8270D (gas chromatography-mass spectrometry (GC-MS)).In contrast, the trace element determinations were performed at the Laboratory of Atomic Spectrometry (LABSPECTRO) using EPA method 3050 (inductively coupled plasmaoptical emission spectrometry (ICP-OES)).All samples were analyzed in triplicate.Both laboratories are part of the Department of Chemistry at PUC-Rio.

Stability study
Several determinations were performed following the production of the sediment samples (n = 6) to assess the stability of the sediment material (SED 001).The results showed that the material is stable under actual conditions regarding the tested compounds (F test, ANOVA, p < 0.05).Vol. 26, No. 2, 2015

Moisture content
The initial moisture content established for the material was 3.95 ± 0.05% (n = 10).With time, the moisture contents were determined from aliquots of the lyophilized material by drying in an oven at 105 ºC to a constant weight.Currently, the moisture content is 4.00 ± 0.05%.

Information regarding the shipping and data return
In February 2008, 42 bottles of sediment sample (SED 001) were distributed to laboratories that responded positively to the invitation letters, which were forwarded to 100 labs.In addition, a sample description and general instructions for analysis were included.
This study followed the structure suggested by Carvalho et al. 9 Participants were requested to analyze hydrocarbons and trace metals by using the current techniques and methods that were applied in their laboratories.In addition, participants were also requested to perform three separate determinations for each analyte and to include a short description of their methods with the results (extraction, clean up, gas chromatographic conditions, etc.).For this purpose, a reporting form was attached to the information sheet.The concentrations were reported as net values (i.e., after correcting based on a blank, moisture content, etc.), with as many significant figures as justified by the precision of the method.For each group of targeted substances, the participants were requested to report the average weights of the samples that were taken for analysis, the moisture content, the methods used to confirm the identities of the compounds, the calculations used for the results, a summary of the quality control procedures that were routinely employed in the laboratory, the arithmetic mean value and the relative standard deviations (in percentage) of their determinations.

Homogeneity test
The results and the statistical analyses (F test (ANOVA); SPSS 8.0 applied for evaluating the homogeneity) indicated that the SED 001 could be considered as homogeneous between the bottles and within each bottle.The material did not show any significant inhomogeneity (F = MSB / MSW < F crit for α = 5%, where MSB and MSW are the mean squares between and within-bottles, respectively). 4Tables 1 and 2 show the preliminary results for the trace elements and hydrocarbons (mean and standard deviation (SD)), respectively.

Stability study
Several determinations were performed following the production of the sediment samples (n = 10) to assess the stability of the material (SED 001).Table 3 shows the results produced by LABMAM between 2007 and 2014 for some analytes.The results showed that the material is stable under actual conditions (F test, ANOVA, p < 0.05).The bottles were stored in the laboratory at a temperature of 21.0 ± 1.0 °C.

Data summary
Forty two laboratories agreed to participate, and only 27 laboratories returned their results.Of these laboratories, 8 provided data for both hydrocarbons and trace elements, 11 only provided results for hydrocarbons, and 8 only provided results for trace elements.Each laboratory was identified by a code number to preserve anonymity.The analytical methods that were used by the participants for petroleum hydrocarbon analysis, including the pre-treatment of the sample and the GC conditions, are given in Table 4.For trace elements (Table 5), the methods of quantification used by the laboratories ICP-OES, atomic absorption spectrometry (AAS), inductively coupled plasma mass spectrometry (ICP-MS), and neutron activation analysis (INAA).

Statistical evaluation
Results were considered acceptable when they were obtained from laboratories that reported a minimum of three replicates.The laboratories that only reported one or two replicates were disregarded.The range, mean, and standard deviation values for each analyte were based on the reported laboratory means.Next, the outliers were removed using Grubbs test, and the range, mean and standard deviation were recalculated.
To determine the laboratory performances, z-scores 10,11 were calculated according to the following equation: where x i is the mean of the reported analyte concentrations in the samples from each laboratory; x a is the assigned value (a mean value of the acceptable results in the intercomparison run); and s b is the target standard deviation. 10The z-score is based on the properties of a normal distribution.These values express the performance of each laboratory and its deviation relative to the average results.The performance is considered acceptable if /z/ < 2. The measurement is regarded as out of control when /z/ > 3.This score represents a simple method for providing each participant with a normalized performance score for bias.3][14] In addition, this calculation accounted for the variations of repeatability (which estimate the variation in a single laboratory over a short period of time) and reproducibility (which estimate the variations of the measurements between the laboratories). 10iteria for assigning a reference material Data for many of the analytes were sufficiently grouped to characterize the sediment samples as a reference material.The criteria used in the certification of petroleum hydrocarbon are presented by Villeneuve et al. 5 and Carvalho et al., 9 i.e., (i) for a range of PAH concentrations.In this case, the relative uncertainty of the overall median does not exceed the following limits: > 10,000 µg kg -1 (30%), 1,000-10,000 µg kg -1 (50%), 100-1,000 µg kg -1 (60%), 10-100 µg kg -1 (80%) and < 10 µg kg -1 (100%).Furthermore, (ii) the overall mean is based on the data obtained by at least two different methods.(iii) The relative number of laboratory means that were rejected as outliers did not exceed (a) 20% or (b) 50%.Finally, (iv) the overall mean was calculated based on the following number (or more) of laboratory means: (a) 10 or (b) 5. Based on these criteria, the certification classifications for aliphatic and aromatic hydrocarbons are provided in Tables 6 and 7, respectively, and can be defined as follows: Class A: when all criteria are met; Class B: when criteria i, ii, iiib, and iva or i, ii, iiia, and ivb are met; and Class C: when at least three laboratories give results that reasonably agree with each other.
These tables provide information regarding the final results, including the consensus values, the standard deviations, the numbers of participants who reported the concentrations of each substance, and the classification obtained by the analyte. 5or trace elements, the consensus values are presented in Table 8 for the three different digestion methods that were used (after verifying the outliers).We only considered the analytes for which at least three replicates were reported.

General remarks regarding the intercomparison of the hydrocarbons and trace elements
Overall, 19 laboratories provided data for hydrocarbons.Among these laboratories, only two did not report results for aliphatic compounds.Three of these laboratories only sent one analysis result and were excluded from the final evaluation.Generally, 64% of the laboratories indicated at least one outlier when determining the aliphatic hydrocarbon content, and 68% of the laboratories presented at least one outlier when determining the PAH content.Some laboratories did not report any correction for soil moisture content.The performance assessed by the z-score was satisfactory for all laboratories.Despite the large variability that is inherent to the intercomparison exercise, the typology of the n-alkanes and the PAH was As expected, the results obtained for the trace elements using the different digestion methods showed a gradient that was related to the efficiency of the procedure.Milder digestion (with HCl) 6 resulted in lower concentrations than the samples analyzed by the two other methods, EPA method 3050B 7 (concentrated HNO 3 treatment) and EPA method 3052 8 (HNO 3 + HF).Similarly to the hydrocarbons, the z-scores for the metals were satisfactory.Three laboratories (or 25%) reported outlier values for the methods using 1 mol L -1 HCl, 15% reported outlier values when using EPA method 3050B, and 33% reported outliers when using EPA method 3052.Figure 3 shows some of these results.

Comparison of SED 001 with other materials
Table 9 shows the results and standard deviations that were obtained in the current study (SED 001) and those reported for two commercial reference materials (NIST 1941b 14 and IAEA-435 15 ).The data variability observed in this study was in the same range as the variabilities reported for other materials prepared by specialized groups in this area.In fact, the reference materials are classified into two groups, reference materials (RMs) and certified reference materials (CRMs).In the second group, the traceability chain is much larger and should generally present lower variability.
Very few reference materials report alkylated compound contents.In this study, the results obtained for alkylated PAH are of similar quality to those reported for the 16 PAHs.In addition, no increase in variation was observed due to the limited availability of commercial standards for alkylated compounds and due to the different approaches used to determine their concentrations (Table 10).Exceptions occurred for the lighter alkylated PAHs, such as alkyl naphthalene and alkyl phenanthrene, which presented higher variation.In addition, the higher molecular mass homologues tended to show larger variability, which was potentially derived from the growing uncertainty in defining the area in the ion chromatogram when transitioning from C 0 to C n compounds.
Table 11 shows the results for trace elements in the current study (SED 001 by EPA 3052 method 8 ) and of other commercial reference materials (NIST 8704, 16 NIST 695, 17 and IAEA-433 18 ).The range of the results for the reference material (SED 001) was 4.0 to 25%, which was higher than the range presented by NIST RM 8704 (from 3.0 to 8.0%), NIST 695 (from 0 to 4.0%) and IAEA 433 (5.0 to 12.0%).
A list of laboratories that participated in this study is presented in Table 12.preserved.In Figures 1 and 2, both classes of compound are shown for the data obtained in this study.This result shows that the RM SED 001 is suitable for calibrating laboratory procedures that target the appraisal of hydrocarbon sources.Overall, 13 laboratories presented results regarding the trace element concentrations in sediment sample SED 001.

Conclusions
The produced material is suitable for use as a reference material in processes that require routine quality control when analyzing hydrocarbons and trace elements in sediments.In general, the laboratories that participated in this program attained satisfactory performance, which was due to the large confidence intervals in some cases.

Figure 2 .
Figure 2. z-Scores for the aromatic hydrocarbons in sediment sample SED 001.2*: Laboratory that performed the determination using two methods.

Table 1 .
Trace element results of the homogeneity test.Mean values (triplicate) from each vial are expressed in mg kg -1 , dry weight

Table 2 .
Results of the homogeneity test for PAH.Mean values (triplicate) from each vial are expressed in µg kg -1 , dry weight

Table 3 .
Results of the stability study for PAH, data analyses and sample size.Values are expressed in µg kg -1 , dry weight

Table 4 .
Summary of methods used for quantifying the petroleum hydrocarbons (aromatic and aliphatic compounds) in sediment sample SED 001 a Laboratory that performed the determination using two methods.ASE: accelerated solvent extraction; HPLC: high-performance liquid chromatography; GC-FID: gas chromatography-flame ionization detector.

Table 5 .
Summary of the methods used for quantifying the elements in sediment sample SED 001.Laboratory code numbers and the instrumental methods used in each procedure

Table 6 .
Aliphatic compounds, dry weight basis.The results are expressed in µg kg-1

Table 7 .
Aromatic hydrocarbons, dry weight basis.The results are expressed in µg kg -1 Average values are expressed on a dry weight basis; b standard deviation; c number of accepted laboratory means that were used in calculating the recommended values and the standard deviation.Vol. 26, No. 2, 2015 a Figure 1.z-Scores for the aliphatic hydrocarbons in sediment sample SED 001.2*: Laboratory that performed the determination using two methods.

Table 8 .
Trace elements, dry weight basis.The results are expressed in mg kg-1 a Average values are expressed on a dry weight basis; b standard deviation; c number of accepted laboratory means that were used in calculating the recommended values and the standard deviation.

Table 9 .
Comparison of SED 001 with the other reference materials.The results are expressed in µg kg-1

Table 10 .
Results for the alkylated compounds in SED 001.The results are expressed in µg kg-1

Table 11 .
Comparison of SED 001 with other reference materials.The results are expressed in µg kg-1

Table 12 .
List of participants