Publication Abstract

Project ME4117 – Development and Improvement of Analytical Methods for Marine Monitoring. 1.2 – Improvement of analytical methods for monitoring polychlorinated biphenyls, organochlorine pesticides and polybrominated diphenyl ethers in marine samples

Project ME4117 – Development and Improvement of Analytical Methods for Marine Monitoring. 1.2 – Improvement of analytical methods for monitoring polychlorinated biphenyls, organochlorine pesticides and polybrominated diphenyl ethers in marine samples

J. L. Barber, S. Losada, B. Wilford, S. Morris and P. Bersuder

The Cefas laboratory routinely performs analysis of polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs) and polybrominated diphenyl ethers (PBDEs) in marine matrices such as sediments and biota. Current methods of sample preparation and analytical determinations are resource intensive and costly. With the aim of reducing costs and increasing sample turnaround times associated with these processes, method developments involving an automated, accelerated solvent extraction (ASE) technique and extract clean up steps using gel permeation chromatography (GPC) and solid phase extraction (SPE) were undertaken. The application of programmed temperature vapourisation (PTV) sample extract injection and gas chromatography with mass spectrometry (GC-MS) and were also investigated with the objectives of improving sensitivity and specificity of analysis. From the outcomes of these experiments, selected and optimised methodologies were then validated using sediment certified reference materials (CRM) and fish tissue.

Programmed temperature vapourisation low flow injection was optimised with a splitter flow pressure of 6 psi and final temperature of 270˚C. The inlet temperature was set at 45˚C with a vent time of 0.8 min. However, compared to the splitless injection mode, later eluting PCBs showed a reduction in peak responses. For optimisation of a multiple injection method, a 1 min vent time was applied along with an inlet temperature of 40˚C. Hexane as an injection solvent proved to be promising.  Increasing the vent pressure did not have an effect on low boiling point compounds. However, later-eluting peaks showed a much lower response, possibly due to these compounds being retained in residual solvent. No real effect was seen by reducing the heating rate and this is expected for PCBs as they are not thermally labile. Further work on this issue would be necessary using an OCP standard mixture where several OCPs are labile. Increasing the transfer time improved responses significantly for all peaks with mass spectrometric detection, but less so when using electron capture detection, which is the conventional method of detection.

It was found that the operation of the automated ASE apparatus required frequent maintenance to ensure satisfactory performance. With in-cell clean up and three extraction cycles, extracts from CRM sediments were free of most co-extracted matrix components. However, the presence of elemental sulphur remained an issue even after reacting extracts with copper to remove it. Where CRMs were analysed using ASE, acceptable analyte recovery appeared to be a product of the sample intake mass. One gram samples were better dispersed with a greater contact with the extracting solvents compared to 10 g sample masses. The inclusion of internal and labelled PCB standards prior to extraction also improved recovered amounts (66 to 121%) of most of the native analytes. Recoveries >120% (i.e., for CBs 49, 128 and 138 plus pp’-TDE) were most likely due to effects of co-eluting interferences during GC separation.  For the isolation of PBDEs, ASE proved problematic and did not provide a satisfactory and robust within-cell, extract clean up technique. Low recovered concentrations were found when compared to the conventional method of Soxhlet extraction followed by column chromatography clean up.

An analytical method involving gas chromatography with electron impact ionisation and single quadrupole mass spectrometric detection (GC-EI-MS) was developed for the analysis of PCBs and OCPs. Compared to the existing approach of using GC with electron capture detection (ECD), the optimised GC-EI-MS sample cycle time using selected ion monitoring (SIM) was 10 min shorter. An improvement was thus made on sample turnaround times. Acceptable linearities of detection were evident and correlation coefficients from calibration plots were all >0.99. For most analytes, instrumental limits of detection (iLODs) were between 0.5 and 1 ng/mL. With the exception of CB194 and HCHs, the sensitivity of the method was found to be acceptable to meet the Green Book requirements. On applying the developed method to the analysis of PCBs and OCPs extracted from sediment CRMs, within-batch variations of determined concentrations were lower compared to those found with GC-ECD. With the selective method of detection, elimination of interfering peaks especially associated with the internal standard CB53 was apparent. This resulted in a greater accuracy of analysis and, thus, an improved confidence in the quality of data in comparison to data derived after the less selective, GC-ECD approach.

Deploying a triple quadrupole mass spectrometer coupled to GC, an analytical method with electron impact ionisation tandem mass spectrometry (GC-EI-MS/MS) was optimised and developed to detect and quantify 17 PBDEs prepared in solvent. With a DB5MS GC column for analyte separation, data acquisition using multiple reaction monitoring (MRM) of specific transition ions per analyte was applied. Instrumental LODs ranged ranged from 0.2 to 25 pg on-column and the method was least sensitive for higher molecular weight compounds such as BDE183, 204 and 209. The most likely explanation is due to the design of the quadrupoles of the particular instrument. High mass-to-charge fragment ions may be unstable in the third quadrupole and an insufficient abundance of these may not have been transmitted to the detector.

An automated extraction procedure using a SpeedExtractor was also developed for the isolation of PBDEs from fortified fish (salmon) muscle tissue. An optimised method based on recovered analytes used a solvent combination similar to the conventional Soxhlet extraction approach (i.e., hexane plus acetone; 1:1) and three extraction cycles. Compared to the existing and time intensive, Soxhlet procedure, similar quantities of lipid were obtained from a cod liver oil sample using the SpeedExtractor. In order to remove co-extracted interferences such as lipids, gel permeation chromatography (GPC) and solid phase extraction (SPE) using Florisil, alumina and/or acidified silica were also developed, optimised and their application compared. All target BDE congeners eluted after the lipid peak during GPC clean up, indicating an efficient method for the preparation of lipid-free tissue extracts. In addition to hexane, dichloromethane (DCM) was necessary to quantitatively elute the more retained PBDEs from Florisil SPE cartridges. It was observed that acidified silica degraded some of the novel brominated flame retardants (BFRs) which were also included in these studies. Since it was the objective to develop a generic method to include PBDEs and related BFRs, the use of acidified silica was found to be inappropriate.

A combination of optimised GPC and SPE (with Florisil and sequential elution using hexane and DCM) were selected for method validation. Extracts derived from PBDE-fortified salmon liver replicates were analysed by GC-NCI-MS and GC-EI-MS/MS. For PBDEs with more than eight bromine ions, the GC-EI-MS/MS technique was not sufficiently sensitive to meet the Green Book guideline limit of 100 ng[analyte]/kg. Theoretical, method limits of quantitation (mLOQ) ranged from 5.5 to 685 ng/kg. To meet the guideline limit, it would be necessary to increase the sample intake mass or the volume of extract to be cleaned up. In turn, this would require a more robust, clean-up method. For analysis by GC-NCI-MS, mLOQs were better and in the range of 1.7 to 8.3 ng/kg. The benefit of selectivity of the MS/MS method was proved when compared to the MS method. Extraneous, halogenated and non-PBDE peaks observed by the latter approach were absent during MS/MS analysis. This was expressed in the recoveries of some of the PBDEs after GC-NCI-MS. The non-selective nature of this method where m/z 79 and 81 are monitored means that other and novel BFRs that were simultaneously spiked into the samples for purposes of their validation and which also share similar retention times and fragment ions with some of the PBDEs, enhanced the responses of several PBDEs. This led to an over recovery for some congeners, such as BDE17, BDE66, BDE99 and BDE154. This may also reflect future chromatographic observations from the analysis of real environmental samples. Novel BFRs are now replacing PBDEs, formerly among the most widely used flame retardants, in some applications. For this reason, there is justification for GC-MS/MS methods to replace GC-NCI-MS for the analysis of samples where a mixture of brominated flame retardants is expected. Using MS/MS, recoveries for tri- to hepta-BDEs were acceptable for all levels of spiking and ranged from 70-120%. However, recoveries of nona-BDEs and BDE209 were not satisfactory. Of the nona-BDEs, only BDE208 demonstrated a recovery of >50%.

It is recommended that further work is undertaken to refine in-cell clean up of sediment extracts during automated and accelerated solvent extraction to improve the removal of co-extracted elemental sulphur. Further assessments should be made of the application of different solvent combinations and extraction cycles for the efficient recovery of not only PCBs and OCPs but including PBDES from sediments and both lipid-rich and lipid–lean biota samples. Continued investigations are recommended for the application of different GC column types and higher injection volumes for GC-EI-MS analysis of PCBS and OCPs. The analysis of the high molecular weight nona-BDEs and BDE209 proved problematic and particular effort should be directed at improving and optimizing isolation and clean up methods for their determination.

J. L. Barber*, S. Losada*, B. Wilford, S. Morris* and P. Bersuder*

Cefas Technical Report