CHAPTER 1: Risk Assessment of Plastic-based Food Contact Materials: Focus on Polyolefins

Polyolefins is a term indicating a class of polymers commonly used for, among other applications, food packaging.¹  They are produced starting from one or more monomers, polymerization aids (catalysts, peroxides, etc.) and polymer production aids (i.e. in-process antistatic agents). Solvents can also be present. The resulting base polymer is then formulated with additives. Polymers in their final form (generally pellets, granules or flakes) are sold to converters to produce articles.

Monomer(s), additives, polymerization aids (PAs) and polymer production aids (PPAs) are considered intentionally added substances (IASs). Impurities within starting raw materials and reaction products from the manufacturing process (i.e. oligomers, additive by-products) are defined as non-intentionally added substances (NIASs) and are subjected to risk assessment, in addition to PAs and PPAs if not listed in EU or National Food Contact Regulations.

Regulation (EC) No. 1935/2004 of the European Parliament and of the Council of 27 October 2004 provides general principles of safety and inertness for all food contact materials (FCMs). In addition, EU Regulation No. 10/2011 as amended sets out rules for plastic materials in contact with food. EU Regulation No. 10/2011 describes requirements for migration tests. There are two types of limits: OML (overall migration limit) and SML (specific migration limit). OML is based on the total mass of substances that can migrate (gravimetric analysis), whereas SML is based only on quantification of one specific substance that can migrate into food.

EU Regulation No. 10/2011 is based on the principle of a “positive list” and Annex 1 contains the list of authorized chemicals (monomers, additives and certain PPAs). Some listed substances can have an SML. Other substances that are not required to be listed, such as PAs and NIASs, need to be risk assessed.

When a polymer is used for food contact applications, there are important material features that must be considered before approaching any further analytical challenge related to risk assessment:

• ODOUR: According to current regulation:²  “… any material or article intended to come into contact directly or indirectly with food must be sufficiently inert to preclude substances from being transferred to food in quantities large enough to endanger human health or to bring about an unacceptable change in the composition of the food or a deterioration in its organoleptic properties.” This means that it is necessary to monitor packaging material components that could migrate into food, deteriorating safety and quality, starting from a low molecular weight.

• NIASs: According to EFSA (European Food Safety Association),³  reaction products and impurities (NIASs) that are not included in the EU list of authorized substances in EU Regulation No. 10/2011 are required to be risk assessed, irrespective of their source or intended function. Oligomers are also considered as NIASs. Their identification and quantification are required for risk assessment by Art. 19 of EU Food Contact Regulations 10/2011/EC, specifically for those oligomers with molecular weight (MW) <1000 dalton (Da), as potential migrants of concern which, after transfer (migration) into food, may be physiologically resorbed after ingestion.

Polymer characterization usually consists of fractionation/extraction or thermal desorption. Fractionation/extraction can be carried out using different solvents and techniques. Aiming to investigate all the species present in an FCM, a combination of analytical methods is used. With polypropylene, a commonly used approach is the separation of the insoluble crystalline part (X_i) from the soluble atactic fraction and oils.^4,5  Methylene chloride (CH₂Cl₂) is another solvent that is commonly used. The fractionated or extracted part can be analysed using different instrumental techniques: NMR (nuclear magnetic resonance), GPC (gel permeation chromatography), GC-FID (gas chromatography coupled with flame ionization detection), GC-MS (gas chromatography coupled with mass spectrometry), HPLC-MS or HPLC-DAD (high-performance liquid chromatography coupled with mass spectrometry or diode-array detection, respectively). The polymer as such can also be thermally desorbed and studied by GC-MS and GC-FID (see Figure 1.1).

What emerges is a very complex matrix, with species covering a very wide MW range, starting from a few daltons (residual monomer) up to millions of daltons (high-MW polymers with high crystallinity), as shown in Figure 1.2. This scheme is characteristic and different for each individual polymer.

Considering a very common FCM, i.e. a printed film to wrap food, a deeper investigation reveals a very complicated situation (see Figure 1.3). At least four different polymers were used, laminated with a printed surface and glue.

Analysis of a final item is very difficult and the result is a sum of the contributions from each individual layer component, including the printed layer (with or without paper, when present) and glue, and heating effects during extrusion/lamination. Moreover, it is worth remembering that each film layer can be formed by more than a single polymer.

Very complex matrices, due to different plastic materials and printing and lamination processes, are usually challenging to analyse if the whole production chain is unknown. The final item is analysed to characterize NIASs, but it would often be desirable to obtain more information about the origin of the NIASs.⁶ 

For the above reasons, to achieve a full characterization and understanding of FCMs, each individual polymer should be analysed as such, together with the laminated final film with and without the printed and glued layer.

Polymer analysis is usually carried out by GPC, in particular to describe finally the MW distribution. It could also be considered as a first approach to characterizing oligomers with MW <1000 Da. The separation path is based on the different retention times of molecules of a polymer in solution. Chemical interaction between the particles and the stationary phase can be considered negligible. In addition, no information about the chemical nature of compounds is obtained.

Starting from polypropylene homopolymers (PP HOMOs), Figure 1.4 shows three different materials with the same average MW but different MW distributions, due to different synthesis conditions.

The resulting GPC traces are shown at top left in Figure 1.4 and the NIAS–oligomers region is the grey circled zone. Even if clear differences in the MW distributions for the three samples can be measured in terms of average MW (M_n, M_w, M_z), no accurate speculations are feasible for the NIAS region via GPC. To overcome the poor resolution and separating power of GPC, GC must be considered.

For comprehensive and full characterization of ODOUR, NIAS (or IAS) substances, including oligomers, a deep and more sophisticated analytical approach is needed. Generally, an extraction step of the potential migrants from the matrices is applied and then GC or LC separation is performed. Thermal desorption (static or dynamic) is applied for volatile compounds, whereas solvent extraction is needed for the characterization of less volatile species. Capillary GC and HPLC provide the required separation, selectivity and efficiency.

As for the detector to be used, not only should it give as much information as possible for each separated organic compound, but it should also have good sensitivity. A mass spectrometer coupled with capillary GC or HPLC fits all these requirements.

Capillary GC, coupled with FID, is also widely used in many studies as it provides a relatively easy and universal quantitative approach. The FID response for hydrocarbons can be considered as weight percent response, even without any use of external standard (ESTD) or internal standard (ISTD) calibration.

Within capillary GC, the main technique used for volatile organic compounds (VOCs), three different sample analyses are necessary, starting from static headspace (SHS) for the C₃–C₂₄ range, passing through a dynamic headspace–thermal desorption system (TDS) for C₉–C₄₈, and arriving at solvent extraction for C₉–C₆₀ or heavier compounds. Hydrocarbons up to C₁₀₀ can also be eluted, if high-temperature-resistant capillary columns are used. C_n roughly indicates the number of carbon atoms in the possible eluted species.

To detect additives (IASs) or some high-MW additive by-products (NIASs), HPLC-DAD-MS is necessary. HPLC cannot be used to quantify oligomers because, owing to their chemical nature, they are not detectable with either MS or DAD.

APCI (atmospheric pressure chemical ionization), ESI (electrospray ionization) or APPI (atmospheric pressure photoionization) sources are not effective in aliphatic hydrocarbon ionization. Different extraction methods have been described,⁷  but essentially they are based on UASE (ultrasonic-assisted solvent extraction), OSM (one-step microwave-assisted extraction) or TSM (two-step microwave-assisted extraction).^8–11 

In order to detect NIASs or IASs with SML using HPLC-MS, a quadrupole instrument is not sufficient. APCI, ESI and APPI are considered soft ionization techniques,¹²  so they do not produce EI (electron ionization)-like spectra, with fragmentation patterns of analytes, but only a few ions for each compound (i.e. molecular ion and one or two fragments). These few ions and their relative abundances (fragmentation pattern) depend on the solvent and temperature. Consequently, these fragmentation patterns differ considerably from those available in the database currently used for species recognition analysis and they are not searchable in commercially available mass spectral libraries. Only by using the exact mass (2 ppm or better resolution) it is possible to establish the empirical formula of the analyte. If CAD (collisionally activated dissociation) is used both in single-stage MS and multi-stage mass spectrometry (MSⁿ), it is possible to obtain a fragmentation pattern.¹³ 

Since DAD is not a destructive detection method, it can be used before MS with very good results. Coupling DAD with MS (TOF), identification of some IASs (see Figure 1.5) or NIASs (see Figures 1.6 and 1.7) can be carried out using exact mass calculation. The UV spectrum, when reported in the literature, can also be helpful.

Samples suitable for GC-MS or GC-FID investigation can be spheres or particles (polymer before extrusion), pellets, films, items or food as such. Usually, the sample amount is 1 g, for the SHS method a few grams (2–5 g) for solvent extraction and less than 1 g for TDS. If solvent extraction or SHS is used, spheres and pellets require grinding before treatment. An FCM (final item) requires cutting or grinding. TDS can be used with a few spheres or pellets, but manufactured items and films need to be cut into small pieces before thermal desorption.

Any sample manipulation should be carried out wearing cotton gloves, to avoid sample contamination. Plastic gloves should not be used, to avoid “ghost peaks.” Contamination due to ubiquitous substances or interfering compounds should be considered in solvents used for HPLC and/or extraction.

SHS is wrongly considered a relatively easy technique. Reality is different, and the results should be carefully evaluated before drawing any conclusion. Together with TDS-GC-MS, it is the main technique used to investigate odour.

Very exhaustive discussions about SHS, the amount sample and its treatment, relative vial pressure, etc., can be found in the literature.¹⁴  In the following, a few practical considerations are discussed.

Mechanisms of odour formation and its transport,¹⁵  together with the odour threshold,¹⁶  have been described in the literature. If food as such is analysed, useful references can be found in a flavour reference handbook.¹⁷ 

SHS does not require any particular equipment except a glass vial that can be closed with an inert septum. Vials of size 20 mL are the most commonly used, sealed with silicone or a butyl rubber internally PTFE-coated septum. Sample heating, vial gas-phase injection and chromatogram evaluation complete the analysis (see Figure 1.8).

The sample amount should not be greater than 1–2 g.¹⁴  The maximum and analyte quantitation are described by the equation

where C_G = concentration of the analyte in the headspace, C₀ = original sample concentration of the analyte, K (partition coefficient) = C_S/C_G, C_S = concentration of analyte remaining in sample, β (phase ratio of the vial) = V_G/V_S, V_G = volume of the gas phase and V_S = volume of the liquid–solid sample.

If peak identification only is required, a search of mass spectral libraries is the first necessary step. If quantitation is also necessary, it is very important not to forget some fundamental concepts.

For each individual component, starting from residual monomer(s), up to oxygen-based or odorous substances, passing through hydrocarbons (C₂–C₂₄), the partition coefficient K must be considered in order to avoid incorrect analytical results.

The first approach in material investigation can be carried out by heating the closed vial containing the sample without any grinding. GC-MS and interpretation of the mass spectra give a rough indication of the compounds present in the gas phase. Chromatographic separation can be achieved with apolar columns with a thick film phase (VOC-like) or polar columns (e.g. Carbowax™-like).

Of course, the gas-phase concentration and K differ not only according to the heating temperature but also to the nature of the matrix (different polymers give different K values for the same substance), matrix particle size, nature of the analyte [polar, non-polar, its boiling point (bp), etc.] and equilibration time. The heating temperature and K are the limiting factors.

If an internal standard (ISTD) is used, it must be selected to be as similar as possible to the substance or chemical clusters to be quantitated (i.e. comparable K or similar bp). An example, obtained using FID, is shown in Figure 1.9.

If toluene is used, hydrocarbons with similar bp (C₉) can be quantified. If hexadecane is used, C₁₂–C₁₅ hydrocarbons can be studied. Of course, a single ISTD must not be used to quantify the whole chromatogram.

To determine the effect of different post-polymerization treatments on the VOC population, SHS coupled with GC-MS can be used. An effective comparison can be obtained by heating the same sample type and amount at the same temperature for the same time (see Figure 1.10).

Grinding the cooled polymer before applying the SHS technique is strongly suggested. For some analytes, too long a period of time would be necessary to reach equilibrium with the gas phase if the sample has not been milled. Acetone and TBA (tert-butyl alcohol), typical by-products of some peroxides used for “visbreaking” high-MW distribution polymers, are good examples (see Figure 1.11). Visbreaking is a process commonly used in polymer production. A peroxide is used to decrease the average MW of the polymers (viscosity reduction). It involves radical scission of high-MW chains.

Consider further that not only do acetone and TBA have different K values, but also K depends strongly on the heating temperature and on the polymeric matrix. A homopolymer (Base 1) and a heterophasic copolymer (HECO 2), both milled, were studied at different SHS temperatures (60 and 100 °C). The K values obtained for acetone (together with a co-eluted by-product interferent) and TBA are given in Table 1.1.

When the ESTD approach is considered, a very common error consists in preparing the reference standard calibration in an empty vial, using TVT (total vaporization technique). In this case, the entire liquid sample (a few microlitres), which contains the analytes to be quantified, is vaporized. Only a single phase is present in the vial, hence the concentration of analyte itself in the vial is

where C_G = concentration of the analyte in the headspace, W₀ = amount of analyte and V_v = volume of the vial.

The described procedure does not take into account either K or matrix effects, and also the phase ratio (β) of the vial containing the unknown sample. According to good laboratory practice (GLP), heat should be provided overnight, at about 100–120 °C, with the milled matrix in a ventilated oven, to obtain a “cleaned matrix.” The latter represents the same matrix as the unknown sample, but without analytes. The “cleaned matrix” can be used to prepare an external standard calibration vial, weighing the same amount of polymer of the “unknown” sample and spiking the former with a well-known number of analytes to be quantified. After the same thermostating time, at the same temperature, samples can be analysed by first injecting a gas-phase amount of the “unknown” sample and second the calibration standard. ESTD calculation can now be carried out correctly.

Considering the data in Table 1.1, if TVT is used instead of the “cleaned matrix” procedure to prepare ESTD calibration, quantitation of TBA in HECO 2 (heterophasic copolymer) at 100 °C will be underestimated by more than 50%.

TDS could be considered as a dynamic headspace technique. It is used both to identify odour and NIASs. Instead of heating a closed vial containing the sample and analysing the gas phase at equilibrium, as is done in SHS, the sample is heated at the desired temperature, an inert gas flow (usually helium) passes over it and all the stripped chemical species are cryofocused into a glass liner. After 10–20 min of extraction, analytes in the glass liner are quickly heated and injected with a PTV (programmed-temperature vaporizing injector) or MMI (multi-mode injector). A capillary GC column for separation and a mass spectrometer in scan mode will complete the analytical run.

If the same amount of different samples is thermally desorbed using the same GC conditions, a comparison can be carried out (semiquantitative analysis).

A reliable analysis is obtained if a further thermal desorption on the already analysed sample produces a chromatogram with less than 20% of the peaks obtained in the first run. In order to comply with this rule, sample heating for 20 min at 280 °C could be a good starting point. At this temperature, the polymeric matrix is molten but additives do not degrade and do not produce any by-products or “ghost peaks” (see Figure 1.12).

Extracting the sample in a helium atmosphere greatly preserves it, together with the polymeric matrix. A temperature of 280 °C is sufficient to strip volatile compounds from C₃ to about C₄₈. A commonly used melt stabilizer such as Irgafos^® 168 and its phosphate by-product are visible on the chromatogram when present. Their MWs are 646 and 662 Da, respectively (see Figure 1.13).

The glass liner used for cryofocusing is usually filled with a polymer-based phase (e.g. Tenax^®). This or a similar absorbent guarantee better absorption, but full desorption is sometimes critical. If possible, baffled deactivated liners are preferred, kept at −50 °C during sample thermal desorption. If used, the chromatogram range will be reduced from C₉ to about C₄₈. Residual monomers and lighter chemicals will not be fully trapped, but no memory effects and faster desorption will be guaranteed.

The TDS technique is very effective and can give a considerable amount of information about FCMs. TDS is much more sensitive than SHS and it can analyse a broader range of compounds, not easily extracted by SHS.

In Figure 1.14, a comparison between two equal weight PP homopolymers (pellets) obtained on the same polymerization plant, but using two different catalytic systems, is shown. Zooming part of the chromatogram (see Figure 1.15), it is possible to identify different C₁₂ oligomeric isomers deriving from the use of the two different catalytic systems.

The overall oligomer profiles can be obtained by summing the areas of peaks. For PP, C₉, C₁₂, C₁₅ … up to C₄₈ oligomers can be desorbed and detected under suitable conditions. Eluting peaks that are attributed to isomers with the same polymerization degree are grouped. The outcome consists of a graph where the relative amount of the several oligomer clusters is plotted versus the corresponding different and increasing number of carbon atoms. To exemplify this, data from an experimental case describing three different materials are reported in Figure 1.16.

Using this approach, it is possible to characterize t qualitatively he oligomer distribution. Samples 2 and 1 are similar in terms of both oligomer content and oligomeric MW distribution (bottom left graph in Figure 1.16). Samples 1 and 3 have different oligomer contents and different oligomeric MW distributions (bottom centre graph). Samples 2 and 3 have different oligomer amounts but the same oligomeric MW distribution (bottom right graph).

Thermal desorption applied to polymeric samples, coupled with GC-MS, allows the detection of volatile and semivolatile species. The relevant chromatogram is generally characterized by a high number of species that hardly ever can be completely identified. Trying to characterize each single peak, starting from the very first to the last one, is not only time consuming but sometimes even useless. Frequently, it happens that even the smallest peaks cannot be ignored because they can play a crucial role in NIAS identification. A comparison between a clear monolayer film and a yellowing one, both produced using the same polymer, is shown in Figure 1.17. Notwithstanding the relatively few peaks, peak-by-peak identification appears to be a titanic task.

Working with “ionic series” in the chromatogram extraction mode could be the best approach. Quadrupole-based MS software permits the extraction of the chromatogram obtained in scan mode, considering only the ions that may be of interest in NIAS identification; e.g. extracting m/z 220, only peaks due to quinone-like substances, derived from some antioxidant (AO) additives, are enhanced (see Figure 1.18). The latter are NIASs and they should be quantified to comply with the risk assessment if the film is used as an FCM.

If the TDS approach is used in a first screening for the presence of by-products of some additives (e.g. Arvin's by-product for Irganox^® 1010¹⁸ ), characteristic ions of substances of interest can be used to interpret the chromatograms of different extract ions.

Ion series can be deduced based on experience or considering well-known ones: e.g. amine series: m/z 30, 44, 58, 72, 86, 100, 114, and it is useful to check if these types of components are present. An exhaustive book about MS could be very useful¹⁹  for recognizing substances from their fragmentation patterns.

TDS GC-MS is a powerful technique that is very useful for performing the characterization of C₉–C₄₈ oligomers, most additive by-products, some additives with low–mid MWs, white mineral oils, etc.

Quantitative analyses are not very simple, not only because a single thermal extraction does not allow the complete release of substances, but also because ESTD reference materials are difficult to prepare and scattered results, in different runs with the same calibration mixture, are very common. Each individual case must be optimized, considering both the analytes and matrix features.

Solvent extraction can be considered one of the most important techniques for quantifying NIASs and IASs with SMLs. There are no generally accepted or recognized methodologies and several different solvent extraction procedures have been reported²⁰  that use different solvents, different extraction conditions, different samples and different sample preparation methods and, therefore, they give different results.

First, it could be better to separate IASs with SMLs or general NIASs from oligomers or hydrocarbons in general. The latter are usually described as follows:²¹ 

• polyolefinic products (POHs), divided into:

  • POSHs (polyolefin oligomeric saturated hydrocarbons)

  • POHAs (polyolefins monounsaturated hydrocarbons)

• mineral oils (MOHs), divided into:

  • MOSHs (mineral oil saturated hydrocarbons)

  • MOAHs (mineral oil aromatic hydrocarbons).

EU regulation No. 10/2011 ²²  requires, when possible, analyses on base food. If this is not possible, tests have to be managed using food simulants. As an alternative, a screening approach on plastic materials should be carried out, using “worst case simulation”.^23–25 

Direct analyses on foods are very complex, results are quantitative and it is almost impossible to assess the origin of oligomers and NIASs in general. This is also valid for analyses carried out on food simulants.

IASs with SMLs and mid–low MW additive by-products (NIASs) can commonly be extracted using ultrasound-assisted solvent extraction (UASE) with methylene chloride. Extraction for 4 h at room temperature of foods, milled polymers or other items could be considered suitable for this purpose. Filtration with a 0.2 µm PTFE filter is usually needed. Methylene chloride-based extraction is very common in most testing laboratories involved with NIASs.

Analytical approaches should be carried out using GC-MS in the SIM (selected ion monitoring) mode if a quadrupole is used or extraction of characteristic ions with exact mass when the time-of-flight (TOF) method is used. Triple quadrupole or TOF methods are less affected by matrix effects owing to their ion selectivity. Sometimes, characterization of NIASs is relatively easier in food matrices than in polymers. From a GC point of view, analysis of 2,4-di-tert-butylphenol (a very common by-product of Irgafos^® 168 additive) in white bread using UASE with methylene chloride as solvent is easier than its quantitation in polymers. The white bread matrix is “cleaner” than polymers in terms of peak profile, even if it is much more complex in terms of organics content (fibre, sugars, starches, proteins, etc.). A typical example is shown in Figure 1.19.

As a rule of thumb, UASE with methylene chloride is acceptable for mid-MW IASs with SMLs. Heavier IASs with SMLs or heavier additive by-products require OSM or TSM. Derivatization can be used if everything else tried proved ineffective. Occam's razor is the guideline: the simpler, the better.

Sometimes, owing to a very complex polymeric matrix, it is not very easy to distinguish between oligomers and NIASs (see Figure 1.20). Figure 1.20 shows part of chromatogram of a thermally desorbed copolymer (TDS-GC-MS) (upper chromatogram). The mass spectrum of the peak at 14.9 min is not the same on the left and right parts of the peak itself, which means that at least two different substances are co-eluted. The spectra obtained are well identified as a saturated branched hydrocarbon (oligomer) and 3,4-dimethylbenzaldehyde. The latter is a common by-product of Millad^® 3988, a well-known clarifying agent for PP. Extracting the ion at 133 Da (centre chromatogram) gives a well-resolved peak. The Peak Purity software tool, bottom right, confirms the presence of two different analytes. The chromatogram was obtained using a 60 m × 0.25 mm id 5% phenylmethylsilicone capillary column with 0.25 µm film thickness. A longer capillary column, a thicker film or a stationary phase of different polarity are not the right approach to achieve better selectivity, according to tests carried out in our laboratories.

A different GC approach is needed. Multi-dimensional GC (also denoted GC×GC) could be the way to achieve the needed separation. A useful literature reference is available.²⁶ 

When the same sample was extracted using UASE with methylene chloride and injected (on-column mode) into a GC×GC system equipped with the same capillary column as used previously, but coupled with a 1.8 m × 0.18 mm id RTX^® 200 column with 0.18 µm film thickness, mounted in a secondary oven, Figure 1.21 was obtained.

Considering the 3D chromatogram obtained, it is possible to see how the situation is much easier to understand. The so-called “forest of peaks”^27,28  can now be better resolved. The x-axis represents time in minutes on the primary apolar column and the y-axis represents time in seconds (6 s of modulation) on the more polar secondary capillary column. On the right part of the chromatogram, the oligomeric pattern is clearly visible: it looks like a mountain chain, delimiting the part of the chromatogram closer to the x-axis. The more distant the substances are from the x-axis, the more polar they are. Close to the end of the x-axis run, but shifted to the y-axis (they are more polar than oligomers), heavier additives and some of their by-products are present. 3,4-Dimethylbenzaldehyde is now well separated along the y-axis. A 5 ppm (w/w) concentration based on polymer is readily visible in the scan mode. If the ion at 133 Da is extracted, the peak is well resolved in a completely empty zone of the 3D chromatogram (see Figure 1.22). The SIM mode is often denoted SIC (selected ion chromatogram) in multi-dimensional GC.

Quantitation of the aldehyde is possible by injecting an ESTD reference solution in methylene chloride containing a precisely weighed amount of the by-product. The comparison was carried out not by considering the peak area, but its relative volume. This means that, the faster the mass spectrometer, the better defined is the peak volume. A quadrupole is not fast enough, so for this purpose the TOF mode is needed. The mass spectrometer used to perform this analysis is a high-resolution TOF instrument, which can acquire spectra at a maximum of 25 Hz. Faster TOF at low resolution can reach an acquisition rate of 200–400 Hz. Usually, 200 Hz is high enough to obtain very good 3D peak shapes and correct peak volumes, together with a manageable file size.

Regarding the investigation of POHs and MOHs, two different approaches can be followed:

• (a) UASE in methylene chloride, followed by GC-MS, GC-FID and GC×GC-MS techniques.

• (b) Static extraction in a solvent (hexane), separation of aliphatic and aromatic hydrocarbons by HPLC, followed by separate injection of the two fractions into a capillary GC-FID system.^29–33  Recently, in addition to this, GC×GC has been implemented to distinguish oils from synthetic hydrocarbons in food.⁶ 

When it is possible to have complete traceability of the production chain and samples of each individual step are available, approach (a) is strongly preferred. This will be considered later.

Approach (b) is usually used on food as such or on a final FCM. Unfortunately, the standard technique with HPLC-GC-FID can provide only quantitative analyses. No further specific characterization or identification of POHs and MOHs is possible.

Before starting to investigate the characterization of aliphatic and aromatic hydrocarbons in detail, it would be useful to evaluate their origin in plastic materials. “Real oligomers” (POSHs) are produced during the polymerization process and they are due to weakly activated catalytic sites or to a premature interruption of the polymeric chain, caused by hydrogen addition to limit chain transfer and polymeric growth [Ziegler–Natta (Z/N) catalysts]. If a radical reaction occurs, oligomers can be formed by termination of radical chain growth at an early stage.

Generally, pelletization and addition of polymeric particles obtained in polymerization reduce the amount of oligomers (see Table 1.2). Such components are basically saturated branched aliphatic hydrocarbons. If polyethylene (PE) is considered, saturated linear aliphatic hydrocarbons are the main components.³⁴  Unsaturated branched aliphatic hydrocarbons (including cyclic hydrocarbons) are usually not present or are detected at a very low concentration. MOAHs are absent, because they are not produced during olefin polymerization.

Together with Z/N catalyst, mineral oil³⁵  (sometimes mixed with grease) is added to prepare a catalytic paste for olefin polymerization. Added oil (and grease) can have an MW distribution very similar to those of POSHs or higher (see Figures 1.23 and 1.24), but below or very close to 1000 Da.

Oil and grease are both considered as IASs and, if they are of food grade, do not need to be risk assessed. Figure 1.25 exemplifies the situation commonly found for polyolefin pellets. The chromatogram represents a polymer obtained in an autoclave, pelletized and adding additives together with OB22AT oil (viscosity 2.0 cP at 100 °C). The relatively high amount of added oil is clearly visible. POSHs are also present and they are almost indistinguishable from the mineral oil itself. Most of the “hill” is due to mineral oil, but oligomers and one additive are also present. From the chromatographic point of view, it is almost impossible to distinguish between what are POSHs and what are MOSHs if the MW distributions are very close to each other. Additives can be separated using MS in SIM mode (their ions are very different from those of hydrocarbon series), but even if GC×GC is used, POSHs and MOSHs cannot be distinguished from each other. A mass spectrum obtained by TDS-GC-MS of a commercially available HECO is shown in Figure 1.26. The oligomeric pattern is easily recognized. The “hill” co-eluted between C₂₄ and C₃₃ is due to a residual mineral oil, caused by a low polymerization yield, or added as a lubricant. Both are to be considered as IASs.

When GC×GC in TIC mode was used to analyse 140 ppm of a commercially available food-grade mineral oil, Figure 1.27 was obtained. MOSHs are visible together with impurities in the oil: triterpene-like components, which are a well-known impurity in mineral oils.^6,36 

When a UASE solution of a PP copolymer was spiked in the previously analysed sample, Figure 1.28 was obtained. In the TIC mode, there is no possibility of separating MOSHs from POSHs. However, when the SIC mode was used, extracting only ions belonging to triterpene-like components, Figure 1.29 was obtained.

Monitoring white mineral oil impurities, it is now possible to quantify and separate POSHs from MOSHs, even if the separation is not efficient. Consider also that some MOSHs and POSHs could even have the same CAS number, but they can be classified as IASs or NIASs, depending on their origin.

The ever-increasing need for supporting robust risk assessment of FCMs requires continual improvement of the relevant analytical characterization techniques. Generally accepted unique analytical methods are rarely available. Moreover, owing to the complexity of FCMs, even though continual improvements in sensitivity may be achieved by the analytical tools and databases, unclear traceability of all processing steps can be the cause of only partial and non-robust characterization.

A combination of analytical methods with very high sensitivity and resolution allows a thorough description of the whole species that must be monitored (IASs with SMLs and NIASs).

Regarding polyolefins, this chapter has summarized some of the knowledge achieved so far. The available results, obtained by coupling different techniques and knowledge significantly based on mass spectrometry solutions, show that MOAHs can be excluded as migrants from PP-based matrices, obtained with Z/N catalysts. MOAHs are not produced during the polymerization process and, when detected in the final packaging materials, they derive from oils – greases not food approved³⁶  or from external pollutants.³⁷ 

¹H NMR

Proton nuclear magnetic resonance

AO

Antioxidant

APCI

Atmospheric pressure chemical ionization

APPI

Atmospheric pressure photoionization

CAD

Collisionally activated dissociation

C_n

Number of carbon atoms in the possible eluted species

DAD

Diode-array detection

EFSA

European Food Safety Authority

ESI

Electrospray ionization

ESTD

External standard

FCM

Food contact material

GC-FID

Gas chromatography coupled with flame ionization detection

GC-MS

Gas chromatography coupled with mass spectrometry (quadrupole mass spectrometer)

GC×GC

Multi-dimensional (3D) gas chromatography

GLP

Good laboratory practice

GPC

Gel permeation chromatography

HPLC

High-performance liquid chromatography

IAS

Intentionally added substance

ISTD

Internal standard

LSD

Light-scattering detector

MMI

Multi-mode injector

MOH

Mineral oil

MOAH

Mineral oil aromatic hydrocarbon

MOSH

Mineral oil saturated hydrocarbon

M_n

Polymer number-average molecular weight

M_w/M_n

Ratio of weight- and number-average molecular weights

M_w

Polymer weight-average molecular weight

MW

Polymer molecular weight (Da)

M_z/M_w

Ratio of Z- and weight-average molecular weights

M_z

Polymer Z-average molecular weight

NIAS

Non-intentionally added substance

OML

Overall migration limit

OSM

One-step microwave-assisted extraction

PE

Polyethylene

POH

Polyolefinic product

POHA

Polyolefin monounsaturated hydrocarbon

POSH

Polyolefin oligomeric saturated hydrocarbon

PP COPO

Polypropylene copolymer

PP HECO

Polypropylene heterophasic copolymer

PP HOMO

Polypropylene homopolymer

PPA

Polymer production aid

PTFE

Poly(1,1,2,2-tetrafluoroethylene)

PTV

Programmed-temperature vaporizing injector

SHS

Static headspace

SIC

Selected ion(s) chromatogram

SIM

Selected ion(s) monitoring

SML

Specific migration limit

TBA

tert-Butyl alcohol

TDS

Thermal desorption system

TIC

Total ion chromatogram

TOF

Time-of-flight

TSM

Two-step microwave-assisted extraction

TVT

Total vaporization technique

UASE

Ultrasound-assisted solvent extraction, usually with methylene chloride (CH₂Cl₂)

UV

Ultraviolet

VOC

Volatile organic compound

Z/N

Ziegler–Natta

These studies have been made possible thanks to the valuable contributions and chromatographic support of M. Cavallazzi, P. Marisaldi and G. Tani. Thanks are also due to A. Medri for discussions of regulatory aspects.