Nontarget migration testing

By applying the procedures mentioned above, monomers and additives can be adequately tested. However, other components, the NIASs, can be present in the food contact material and potentially migrate to food. Such unlisted components are gaining renewed attention. In principle, they are already regulated in Article 3 of Framework Directive EC 1935/2004, i.e., food contact materials should not endanger human health. Unlisted substances that could be found in food contact materials may include:

• impurities in starting substances

• reaction intermediates

• decomposition products

• reaction products

Renewed attention to unlisted substances is stimulated by efforts to enhance consumer protection. Recent examples of unlisted substances originating from packaging materials are:

• semicarbazide formed as a decomposition product of azodicarbonamide used as a blowing agent for plastics (e.g. Stadler et al. 2004)

• chlorohydrins and cyclic reaction products formed from epoxidised soybean oil used as a stabiliser (e.g. Biedermann-Bremm et al. 2001).

In these cases migrants from food packaging, with possible toxicological alerts, were found in food although these substances were not intentionally added or used for the food packaging, but were formed during preparation and/or use of the food packaging. These examples show the need to pay attention to unlisted substances. It is therefore necessary to develop a pragmatic and cost-effective strategy to address the issue of unlisted substances that is accepted by both industry and legislative authorities.

5.4.1 Current status and opinions on screening and analysis of migrants

Compared to target analysis used for specific migration studies, trying to detect all possible migrants in food simulants by screening methods offers a whole new perspective. Due to the lack of legislative requirements to analyse all substances migrating from food contact materials, only very few papers deal with this subject. The main principle for methods suitable for screening of migrants is that they should be able to detect, identify and quantify a large range of chemical components varying in chemical structure, polarity and molecular weight. Some possible strategies for the analysis of unlisted substances have already been addressed by a few authors (Feigenbaum et al 2002, Grob 2002). The scheme proposed by Feigenbaum et al. has also been published in Annex II of the EC Practical Guide on materials and articles in contact with food. In principle, the schemes combine well-known analytical techniques for analysing a wide range of components, such as headspace gas chromatography (GC) for volatile components, atomic absorption spectroscopy (AAS)/induced coupled plasma (ICP) for elemental analysis, GC for semi-volatiles and liquid chromatography (LC) for non-volatiles. An additional step may be the application of gel permeation chromatography (GPC)/size exclusion chromatography (SEC) to determine and isolate the fraction of migrants with a molecular mass below 1000 daltons (Da), which is thought of as the toxicologically relevant limit. Figure 5.4 shows a schematic representation of an analytical strategy based on those of Feigenbaum et al. and Grob. In principle, with this analytical strategy a wide range of different migrants can be detected.

Headspace GC-MS is the preferred method for the analysis of very volatile migrants. Practically the same GC conditions can be used as for GC-MS. Due to the coupling to MS, identification is also relatively easy. The heating time and temperature are the main experimental variables. The major drawback of headspace GC-MS is quantification. As a result of the principle of headspace GC-MS, i.e., partitioning of compounds between gas phase and liquid phase, the chemical properties will have a significant influence on the partition of each molecule between gas phase and liquid phase. Therefore, quantification is almost solely possible by using external standards of the same compound (Grob and Barry 2004).

Fig. 5.4 Possible analytical scheme to detect substances migrating from food.

Size exclusion chromatography (SEC) or gel permeation chromatography (GPC) separate samples based on molecular size and thus in most cases on molecular weight. It is the preferred technique for separating low molecular weight components from high molecular weight components based on their molecular mass. As mentioned above, the fraction of migrants below 1000 Da is thought to be toxicologically relevant. As a result it may be advisable to isolate the fraction of migration extracts below 1000 Da before any further analysis. However, a major drawback of SEC or GPC is that the separation is in principle based on molecular size. Separation purely on molecular mass is possible only using calibration standards of specific compounds, e.g., polystyrene. This implies that only polystyrene can be exactly separated on molecular mass. For every other compound separation on molecular mass is approximate. It is therefore difficult to separate the fraction below 1000 Da exactly. In many cases the fraction below ~1500 Da is isolated. GPC/SEC separation can be carried out according to standardised methods DIN 55672, OECD 118 and 119. However, it is questionable whether isolation of the fraction below 1000 Da is really necessary.

Further analysis of the migration extract is preferably carried out using GC-MS and LC-MS. For GC-MS it is impossible to detect components with a molecular mass above 1000 Da due to volatility and the mass range of quadrupole MS detectors. With LC-MS the mass range can be chosen and thus the highest mass can be set at 1000 Da. Multi-methods exist with which almost all elements in the periodic system can be analysed in extracts at low levels. Combination of induced coupled plasma (ICP) and atomic absorption (AAS) coupled to MS is the preferred method. Quantification of each element is also possible with this method.

GC-MS is the method of choice for medium volatile migrants. Relatively standard GC methods can be applied that are able to detect many different types of components, except for very polar components. The only variable that might influence the application of GC is the food simulant or extraction solvent. Almost all organic solvents can be directly injected into the GC system using non-polar columns where only the start temperature has to be adapted to the specific boiling point of the solvent. Aqueous solvents are more troublesome for standard GC methods and special GC columns that are compatible with water should be used. With these columns the separation of polar components is also possible. MS detection coupled to large databases makes GC-MS ideal for identification of migrants. The major drawback of GC-MS is the separation and detection of less volatile components and very polar components, although high temperature columns might increase the volatility range significantly. For example, Irganox 1010 (m/z 1075) has been detected using HT-GC-FID (Feigenbaum et al. 2002).

LC coupled to a combination of MS, PDA and fluorescence is the preferred complementary technique. With LC basically all components can be separated that are soluble in the mobile phase. Separation in LC can be based on polarity, solubility and molecular mass. Moreover there are large numbers of columns each with different properties and the combinations of mobile phases and additives are numerous. Separation and sensitivity are highly influenced by different experimental conditions like mobile phase, temperature, pH, column, etc. It is therefore very difficult to obtain a single LC method that is suitable for a large number of different types of migrants. This is complicated further when MS detection is used whilst there are also numerous variations in type of MS detectors (ion trap, time-of-flight, triple quad), type of ionisation (APCI, ESI) and settings. Furthermore, ionisation is highly influenced by mobile phases and pH. The most often used method is a reversed phase LC method with a C18 column and a mobile phase gradient from water to acetonitrile or methanol. The optimal additives and pH depend mostly on the type of detection. However, before applying this analytical strategy to screen for migrants in food packaging some considerations have to be made with respect to:

• realistic (migration) extract

• sample work-up/pre-treatment

• identification of migrants

• quantification of migrants

• sensitivity/limit of detection.

Realistic (migration) extract

It is very important that a realistic extract be analysed that reflects the food as accurately as possible. For safety reasons it is necessary to obtain a worst-case extract although this should not be exaggerated. In food packaging legislation food simulants can be used that resemble officially the various food types, i.e., 3% acetic acid, water, 10% ethanol and olive oil. Aqueous simulants are relatively easily compatible with LC and can therefore be used for screening of migrants. However, GC is less compatible with aqueous solutions and care should be taken. Possible solutions are the use of water-compatible GC columns, e.g., Aquawax, or the use of derivatisation in which the aqueous extract is freeze-dried, followed by derivatisation of polar functional groups using, e.g., methylation, butylation or silylation. With the latter method polar components are made less polar and can thus be analysed with 'regular' GC methods. However, derivatisation is not straightforward and should be carefully optimised before application. Furthermore, databases like those of Wiley and NIST do not contain many EI spectra of derivatised components.

Olive oil is a very troublesome food simulant with respect to analysis. It is therefore advised not to use olive oil for screening of migrants. Alternative fat simulants are 95% ethanol and iso-octane and these may be more compatible with techniques like GC and LC. However, these two simulants are very different from each other. Iso-octane is very non-polar and therefore exaggerates the migration of non-polar components; 95% ethanol is very polar and therefore exaggerates the migration of polar components. Another option might be to carry out mild extraction using conditions that are relatively worst-case but still realistic, and solvents that are readily compatible with the various analytical methods. Possible examples of these solvents are iso-propanol/iso-octane 1: 1 (v/v), dichloromethane, diethyl ether and tetrahydrofuran. Further research is necessary to show whether these solvents are realistic options and which extraction conditions resemble worst-case, but realistic migration conditions.

Sample work-up/pre-treatment

Ideally (migration) extracts are analysed directly without any sample work-up or pre-treatment. Every extra step in the sample work-up or pre-treatment might result in the loss of (volatile) migrants that may be toxicologically relevant, or the 'creation' of migrants due to contaminants. This can be partially solved by carrying out blank experiments in which a blank solvent undergoes the same procedure. There is a large chance that direct injection of the migration extracts does not lead to sufficient sensitivity and that further concentration is necessary. In this case contamination and loss of migrants should be minimised at all times. Further complication might occur when the solubility limits of specific compounds are exceeded and the components precipitate.


The first aim in the screening of migrants is the ability to detect and separate as many components as possible. A combination of headspace GC, ICP/ AAS, GC and LC is an adequate set-up to cover a large range of different compounds from very volatile compounds to non-volatiles, non-polar to polar and organic versus inorganic. Individual methods should be optimised to increase separation. Migration extracts might contain a large amount of migrants with a large variation in concentration that might interfere with separation. This often results in co-elution thereby complicating both identification and quantification. In the case of small and large peaks co-eluting, the former one might not be visible at all and remain unnoticed. Multi-dimensional separation like GC x GC and LC x LC are the answers to co-eluting peaks (Dalluge et al. 2003). In these methods, different separation mechanisms are applied in two dimensions. For example, in GC x GC components can be separated in the first dimension on volatility whilst in the second dimension the peaks are separated on polarity. A demonstrative example is shown in Fig. 5.5 where a GC x GC-TOF-MS is shown for a migration extract of a polyethylene (PE) food packaging.

It can be clearly seen that along the first dimension a series of large peaks due to PE oligomers are separated whilst in the second dimension more polar components, like additives, are separated. With GC-MS, small peaks due to degradation products of additives might be invisible due to the large peaks of the PE oligomers. The advantage of GC x GC-TOF-MS is that the same mass spectral information and databases can be used and sensitivity is equal to or higher than that of GC-MS. An additional advantage of GC x GC is the possibility of group-type separation where components with similar

Fig. 5.5 GC X GC-TOF-MS chromatograms of a PP migration extract: (a) view along the first dimension (separation on boiling point), (b) view on the second dimension (separation on polarity).

Fig. 5.5 GC X GC-TOF-MS chromatograms of a PP migration extract: (a) view along the first dimension (separation on boiling point), (b) view on the second dimension (separation on polarity).

functionality form groups on the GC X GC chromatogram (Mondello et al. 2003). This might be a useful way to screen classes of compounds, e.g., aromatic amines or isocyanates.

For LC X LC the situation is a bit different. First of all this method is not as well developed as GC X GC and is mostly applied for proteins where a combination of ion exchange-RPLC is used or to polymers where a combination of LC X SEC is most often used. Moreover, a major drawback of LC X LC

is the fact that due to the two dimensions the peaks are diluted and therefore sensitivity decreases, which is already a critical factor for LC. Furthermore, on-line LC x LC-MS methods are not available at the moment. There is a lot of development in the field of LC x LC and without a doubt this technique might play an important role in the future for screening of migrants. The combination of NPLC x RPLC might be very helpful to increase separation based on polarity.

Identification of migrants

A wide range of detection techniques can be used, although unknown migrants can be identified almost solely by means of mass spectrometry (MS). The advantage of GC-MS is the relatively straightforward method used to detect a wide range of components. Identification is relatively easy on the basis of existing mass spectral databases. Although ionisation of molecules may differ amongst molecules, almost all molecules ionise under the conditions of GC-MS, i.e., electron impact ionisation. Identification with GC-MS mainly depends on the presence of large databases. This often results in either complete identification or no significant match. In some cases it is possible to characterise the component using specific mass traces. For example, a mass/charge ratio (m/z) of 149 is often indicative for phthalates. If identification is not possible using databases, other types of MS might be helpful. For example, chemical ionisation (CI) gives in addition to electron impact (EI) the molecular mass of the component. This is often difficult to determine due to strong fragmentation with EI. Chemical ionisation in combination with high resolution MS might give the elemental composition of the molecule. By applying CI and ion trap MS it is then possible to obtain structural information by MS/ MS experiments. A possible identification strategy for unknown migrants using GC-MS based techniques is shown in Fig. 5.6.

MS is the preferred technique for screening of migrants using LC. The ionisation process in LC is relatively soft and does often lead to very simple mass spectra containing mainly the molecular ion. In certain cases the component can be identified based on retention time and molecular mass derived from the molecular ion. However, this is mainly true for well known

Structural Elemental Molecular ion information composition

Fig. 5.6 Possible identification strategy for unknown migrants using GC-MS.

Structural Elemental Molecular ion information composition

Fig. 5.6 Possible identification strategy for unknown migrants using GC-MS.

components. Moreover, the mass spectra can be complicated by adduct formation, especially in electrospray ionisation. Very apolar components (e.g. aromatics, alkanes) cannot be ionised at all with LC-MS and will therefore not be visible in LC-MS chromatograms. No large databases exist for LC-MS and therefore identification of unknown peaks in the chromatogram is often very difficult, expensive and time-consuming, if possible at all.

Recent developments in MS analysers are giving some new perspectives in the field of identification. Especially with high resolution MS, like TOF-MS and even better FT-ICR-MS, the exact molecular mass and thus the elemental composition can be obtained (Hendrickson and Emmett 1999). Using various databases of chemicals on the Internet, it is often possible to identify the component based on the chemical composition. Of course this should be validated by analysing the specific component using LC-MS and comparing the retention time and mass spectrum. MS/MS experiments, using ion trap MS analysers, are in certain cases necessary to obtain some additional structural information in order to decide which chemical composition obtained from high resolution LC-MS is the most likely.

Quantification of migrants

The most common method of quantifying components is the use of calibration standards from the pure compound. However, in the case of degradation products or decomposition products this is not possible due to the absence of a commercially available reference compound. Furthermore, the quantification of all migrants present in a migration extract is very time-consuming and thus expensive. A possibility might be the use of a mix of reference compounds that resemble different types of compounds. This would assume similar ionisation efficiency for compounds with similar chemical structures. In this way a specific peak is, after identification, categorised in a specific class of compounds and is semi-quantified using the relevant reference compound. For example, for a migration extract of PE food packaging a mixture of alkanes, butylated hydroxytoluene and Irgafos 168 might be used as reference compounds for PE oligomers, anti-oxidants and degradation products of anti-oxidants. It is essential that information from the supplier can be obtained regarding what type of ingredients are used and thus what type of migrants can be expected. Although this approach is not 'waterproof', it is relatively easy and holds true in many cases for GC-MS. However, for LC-MS it would be very difficult to use the same approach as ionisation depends very much on chemical structure. For certain types of molecules parallel detection using PDA or fluorescence might be helpful and these types of detection might be more quantitative. After a component is identified and seems to be toxicologically relevant, exact quantification might be possible using reference compounds.

Sensitivity/limit of detection

Assuming that all migrants can be detected by the various analytical methods, the concentration of the migrant in the extract determines whether the migrant is visible in the chromatograms. If a component is not visibly present in the chromatograms it cannot be further addressed. However, this does not mean that the component is of no toxicological risk. The toxicological properties of a specific compound and the extent of its consumption via food determine what concentration is allowed in the migration extract, using the assumptions made in food contact legislation. As a result the acceptable concentration of compounds in migration extracts might differ by orders of magnitude. As a result analytical methods should be sensitive enough to detect the most toxic compounds above certain toxicological thresholds. An example of such a threshold is the Threshold of Regulation of 1.5 mg/day as proposed by Begley (1997). This means that every compound is allowed if the consumption does not exceed 1.5 mg/day. At first instance this seems an adequate rule of thumb for the screening of migrants. However, using the current assumptions in the EU that 1 kg of food is consumed every day and is in contact with 6 dm2, this means that 1.5 mg/day corresponds to 1.5 mg/kg food = 1.5 mg/6 dm2 packaging. Assuming standard migration conditions of 2.34 dm2 packaging in contact with 100 ml simulant, this results in a concentration of (1.5/6)*(2.34/100) = 0.00585 mg/ml = 5.85 ng/ml. Hence a detection limit of 5.85 ng/ml is necessary to ignore all undetected migrants. For most components this is not feasible using the analytical techniques discussed above. Possible solutions to this problem are enrichment of the migration extract by evaporation, also as discussed above. Concentration by at least a factor of 10-100 is necessary. Another option might be the use of actual exposure data instead of relying on the various assumptions made in the current legislation. This is explained in some more detail in a later section.

At this moment, no analytical technique or strategy exists with which all possible components can be detected, identified and quantified. Despite this, the current status of analytical techniques is able to at least give a good indication of the presence of migrants in food simulants. Analysis of migration extracts by the scheme shown in Fig. 5.4 enables the detection of a wide range of non-polar/polar compounds, small/large compounds and volatile/ non-volatile compounds. Table 5.7 gives an overview of literature dealing with the screening of migrants from food contact materials including the type of food packaging and the analytical methods used.

0 0

Post a comment