A03009: Methods and searchable spectroscopic libraries to identify substances migrating above a one part per billion (109) threshold
Wednesday 18 June 2003
This research project aims to provide information on the 'threshold of regulation' approach that is under consideration in the EU.
Background
The threshold of regulation (ToR) approach is under consideration in Europe as a means to limit the amount of migration testing and toxicological testing needed in the area of migration from food contact materials. It is based on the premise that there could be an exposure level below which substances pose negligible health risk to consumers. Therefore, substances migrating below this level may not need to be specifically permitted for use. The ToR approach has been proposed as a way to deal with otherwise seemingly intractable issues, such as contaminants potentially present at very low levels in recycled materials, secondary packaging and ink components.
One scientific and technical problem is that the approach has little practical value if there are no analytical methods capable of detecting and identifying unknown substances at the ToR level required by toxicologists. A 1 ppb migration threshold has been proposed.
In this project we evaluated a suite of four complementary mass spectrometry (MS)-based analytical approaches for their capabilities at the 1 ppb ToR level. The objectives of the project were:
- To establish the capabilities of advanced analytical techniques for detecting and identifying unknown substances down to the level of 1 ppb; and
- To create polymer-specific libraries of EU-listed plastics ingredients and non-listed (but still permitted) impurities so that decisions could be made on which substances detected above any ToR are 'permitted' and which are not.
Research Approach
The experimental work was preceded by a literature review of the threshold of regulation concept and mass spectrometric methods for part-per-billion measurements and identification of substances migrating from food contact materials. This review was used to guide the selection and refinement of analytical methods. Four MS-based analytical techniques were then investigated.
- Headspace gas chromatography mass spectrometry (HS-GC-MS) for the analysis of volatile substances.
- Liquid injection gas chromatography mass spectrometry (GC-MS) for the analysis of semi-volatile substances either directly in a food simulant or in a solvent extract.
- Liquid chromatography mass spectrometry (LC-MS) for the analysis of non-volatile substances.
- Inductively coupled plasma mass spectrometry (ICP-MS) for the analysis of inorganic/metallic elements.
These techniques were first used to analyse mixtures of standard substances, to determine their sensitivity towards representative chemicals that may migrate from packaging materials. Analysis was then performed in the presence of crude solvent extracts of paper and plastics packaging materials, to see if the presence of co-extractives impaired the performance of the methods. Finally, the methods were used to collect spectra of typical plastics extractives, to assist in any future assessment of migrates from plastics intended for food contact.
Results and findings
Detection limits of 10 ppb or lower, using full scan mass spectrometry, were attainable for volatile and semi-volatile substances analysed by headspace and liquid injection GC-MS respectively. The main limitation was ‘swamping’ of regions of the chromatogram by the volatile components in the simulants (ethanol, acetic acid, iso-octane) where the response to the solvent component dwarfed the low level analytes. This would not be a problem in analysis of most foodstuffs because they do not contain these volatile organic solvents.
Inductively-coupled plasma MS (ICP-MS) was found to reach a sensitivity of 10 ppb in simulants for most metals. Some, notably magnesium, aluminium and chromium, suffered from polyatomic interferences and this impaired their detection. The use of a high resolution MS may remove this limitation.
For non-volatile substances particle beam MS coupled to an LC inlet provided electron impact ionisation spectra which were searchable in data libraries. However, this technique proved to be insufficiently sensitive. In the best case, the antioxidant Irganox 1010 was analysed at the 1 ppm level in organic solvent to give an identifiable mass spectrum. Atmospheric pressure chemical ionisation (APcI) gave much improved sensitivity, down to 10 ppb in some cases, but further work would be required to develop appropriate library data bases to make possible the identification of substances through matching of their spectra obtained using this ionisation technique. The major stumbling block in developing such libraries is the dependence of fragmentation patterns on cone voltage and between-instrument variability.
The ToR approach is feasible from an analytical chemistry viewpoint, but with some current limitations. For volatile and semi-volatile substances, the application of GC-MS can achieve good sensitivity and can give characteristic spectra. Headspace GC-MS analysis has some limitations when applied to volatile food simulants. Liquid injection GC-MS analysis also has some limitations when applied to non-volatile matrices such as the simulant olive oil or foods. Volatiles and semi-volatiles, are of major interest because their relatively low molecular sizes mean that they are the type of substances that are most likely diffuse from packaging into foods or food simulants.
For non-volatile substances, the ToR approach is not currently feasible. The best sensitivity achieved using LC-PB-MS was 1 ppm, which is considerably inferior to the performance of GC-MS methods. It may be possible in the future to achieve lower detection limits using LC-APcI-MS once the problems of the variation in fragmentation patterns between instruments have been overcome and APcI libraries have been built up.