- Email: [email protected]
Release of non-intentionally added substances (NIAS) from food contact polycarbonate: Effect of ageing
Food Control 71 (2017) 329e335
Contents lists available at ScienceDirect
Food Control journal homepage: www.elsevier.com/locate/foodcont
Release of non-intentionally added substances (NIAS) from food contact polycarbonate: Effect of ageing , Paola Salvadeo, Claudio Corradini Chiara Bignardi, Antonella Cavazza*, Carmen Lagana degli Studi di Parma, Parco Area delle Scienze 17/A, 43124, Italy Dipartimento di Chimica, Universita
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2016 Received in revised form 27 June 2016 Accepted 14 July 2016 Available online 16 July 2016
Non-intentionally added substances (NIAS) released from polycarbonate tableware were evaluated through untargeted analysis by UHPLC-ESI-Orbitrap. Migration experiments from fifteen samples of different age were performed using ethanol 95% and isooctane as food simulants. High-resolution mass spectrometry permitted the identification of oligomers derived from polycarbonate degradation and traces of colouring agents. Data were analysed with the aim of exploring a possible correlation between type and amount of different oligomers and age of the samples. It has been found out for the first time that the pattern of polycarbonate degradation products observed in new samples was different from that of old samples. Colouring agents were found to be released in higher amount from new samples than from old ones. The work shows the high potential of the technique employed and its importance in the field of control related to safety concern. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Polycarbonate Migration Degradation products Colourants Ageing
1. Introduction The migration of substances from materials in contact with food is considered a positive event when active packaging is realized (Hosseinnejad, 2014). In such case, molecules such as antioxidants (Corradini et al., 2013) and antimicrobials (Lantano et al., 2014) are intentionally incorporated in a plastic film put in contact with a food product with the goal to protect it from spoilage. On the other side, migration from plastic is undesired when potential contaminants occurring in the material reach a food product and may act as toxic compounds affecting health (Lau & Wong, 2000). Therefore, to establish the identity of all substances released by plastic, and evaluate the main parameters affecting their migration (time of contact, temperature, type of food, etc.) is of paramount importance (Hoekstra & Simoneau, 2013). This matter does not regard only food, but involves the pharmaceutical and cosmetic fields too, since those products are often packaged in plastic, and many medical devices are made of plastic. Besides, cosmetics regulation does not require the indication of expiring date, therefore products can be stored for many years in plastic and/ or may remain for a long time on shelves before consumption. Furthermore, products such as sunscreens are generally subjected
* Corresponding author. E-mail address: [email protected] (A. Cavazza). http://dx.doi.org/10.1016/j.foodcont.2016.07.013 0956-7135/© 2016 Elsevier Ltd. All rights reserved.
to high temperature when left for several hours under the sun. In addition, it is worth to consider that the process of migration is not only related to storage and packaging, but also to industrial processing and manufacturing; in fact industrial plants often contain plastic parts, and contact at high temperature and for long time may occur. Many molecules representing potential migrants are well known, and some of them, such as bisphenol A (BPA) and additives including antioxidants and UV absorbers have been studied since n García, Cooperb, Franz & Paseiro long time (Sanches Silva, Sendo Losada 2006; Gao, Gu, & Wei, 2011). Besides, in parallel to a target analysis, it is important to take into account the possible release of unknown substances, such as the so-called non-intentionally added substances (NIAS) that sometimes are found in a product without a clear explanation about their origin. They may represent impurities of raw materials employed, or may derive from degradation of the material components (Nerin, Alfaro, Aznar, ~ o, 2013). The identification of these molecules, together & Domen with the estimation of a potential threshold effect for risk assessment is a relevant issue that deserves great attention in terms of safety concern (Brüschweiler, 2014; Hollnagel, van Herwijnen, & Sura, 2014; Ketelslegers, 2014). In this context, the development of innovative analytical methods based on the employment of modern techniques such as high resolution mass spectrometry (HR-MS) has been demonstrated to be very useful and effective for the identification of
330
C. Bignardi et al. / Food Control 71 (2017) 329e335
unknown compounds released by plastics used in food contact materials (Bignardi, Cavazza, Corradini, & Salvadeo, 2014; Canellas, Vera, & Nerin, 2015a; Canellas, Vera, Ner, í, & n, 2015b). An important point that it is worth to be considered, regards the re-usable plastic containers aimed at storing food products, since they are submitted not only to ageing process, but also to mechanical damage due to the repeated use and frequent washing cycles. Factors such as high temperature and ultravioletevisible light exposure, humidity, atmospheric components, contact with liquids are responsible for chemical and physical ageing (Factor, 1996) that provoke the formation of tiny scratches and small cracks on the surface. As a direct consequence, the object becomes opaque and foggy, and formation of micro-cavities occurs. All these phenomena can be believed to be responsible for the release of polymer components (Struik, 1977; Harvey, 2012), and previous studies showed an increment of bisphenol A released from aged and damaged bottles and tableware respect to unused (Bignardi, , Salvadeo, & Corradini, 2015; Brede, Fjeldal, Cavazza, Lagana Skjevrak, & Herikstad, 2003). In this work, the attention was focused on the untargeted analysis of simulants put in contact with re-usable objects made of polycarbonate (PC) allowed for food contact, and in particular with the aim of identifying molecules derived from a possible degradation of the material during ageing. UHPLC coupled to HR-MS has been applied for the identification of oligomers. The study has been carried out by employing two different food simulants. Previous studies reported the identification of some molecules deriving by PC degradation after ageing and thermal decomposition of PC (Collin et al., 2012; Jang & Wilkie, 2004), nevertheless, no literature dealing with the possible migration of such substances in simulants and/or in food products can be actually found. Similarly, there are no studies about possible migration of colouring agents from tableware, although this topic should be of great interest in the safety assessment (Council of Europe, 1989). 2. Materials and methods 2.1. Chemicals All chemicals employed were of analytical reagent grade. Methanol and water used for chromatographic eluent preparation were of UHPLCeMS grade and purchased by Sigma Aldrich (Milan, Italy) as well as ammonium formate, ethanol and isooctane. Calibration solutions Pierce LTQ Velos ESI Positive ion and Pierce LTQ Velos ESI Negative ion from Thermo Fisher Scientific (Rockford, IL, USA) were used to calibrate the mass spectrometer. 2.2. Samples Fifteen polycarbonate tableware objects of different age (between 2 and 14 years old) and color were submitted to analysis. Samples were named accordingly to the age and the color. All samples were purchased from the same producer. Between them, three samples were new and never been used. Migration tests were carried out by employing ethanol 95% (v/v) and isooctane, reported to be substitutes for the food simulants D2 (EU Regulation 10/2011 on plastic materials and articles intended to come into contact with food, European Commission, 2011). The choice of these simulants derived by the scarce interest of previous studies on migration from plastic objects to fats and oils. Containers were filled with the simulant and placed in a climatic chamber at 40 C for 1 h, covering the surface with glass plates in order to prevent solvent evaporation. Experiments were repeated three consecutive times. Aliquots obtained were evaporated and redissolved in 1 mL of ethanol. Samples were filtered and submitted
to UHPLC-MS analysis. The degree of damage of the PC has been evaluated by examination of the surface by a lens (10) and a digital microscope. To each sample, a label corresponding to numbers from 1 to 5 was assigned according to the amount of scratches or cracks occurring on its surface. Label 1 was assigned to new and unused samples, characterized by a perfectly clean surface, while higher numbers were assigned progressively to more damaged surfaces (Bignardi et al., 2015). Some examples of the surface appearance are depicted in Fig. 1, showing different degrees of damage (increasing from A to D). 2.3. UHPLC-HRMS analysis Analyses were performed by UHPLC-HRMS (Thermo Scientific Ultimate 3000 RSLC) nano system operating in capillary-flow mode coupled to a Q Exactive Mass spectrometer (Thermo Scientific, Fremont, CA). Chromatographic separations were achieved on a C18 Acclaim PepMap RSLC (Thermo Scientific, Fremont, CA) column (150 mm 0.3 mm, 2.0 mm particle size) thermostated at 35 C. Injection volume was 1 mL. Eluents, delivered at a flow-rate of 10 mL min1, were as follows: eluent A, ammonium formate 1 mM in 10:90 methanol:water v/v; eluent B, methanol containing 1 mM ammonium formate. Solvent B was set at 40% and delivered by a linear gradient to 99% in 20 min. Q-Exactive instrument was equipped with a pneumatically assisted ESI interface with a stainless steel needle adapted for capillary flow. Conditions of the interface were fixed according to a previous reported method (Bignardi et al., 2014). Mass calibration was performed every three days in order to obtain a mass accuracy lower than 2 ppm. Chromeleon 6.8 and XCalibur 2.2 softwares (Thermo Fisher Scientific MA, USA) were used to control the instrument and for data processing. Signal acquisition was performed by full-MS-data dependent MS/MS experiments with inclusion list in the range 90e1000 m/z, in both positive and negative mode. The extraction from full MS of each parent ion taken into account allowed the visualization of the isotopic pattern and the matching between the exact mass and the accurate mass. MS/MS spectra obtained in the same run were used for confirmatory purpose (Bignardi et al., 2015). 2.4. Statistical analysis Means, standard deviations (SDs) and correlation values were calculated with SPSS (version 19.0, SPSS Inc., Chicago, Illinois, USA) statistical software. 3D plots of degradation products areas monitored were obtained with the Statgraphics Centurion 16.1 software. 3. Results Migration tests were carried out by employing two different food simulants, ethanol 95% (v/v) and isooctane. Since no standard corresponding to the identified molecules could be available, peak areas normalized respect to the surface in contact with simulant were used for a semi-quantitative approach. 3.1. Data on products of degradation The presence of several oligomers derived from PC chain was evidenced in the examined simulants. The molecules identification was carried out by reconstructing the mass fragmentation spectrum of the most representative molecular ions visible in the fullscan chromatograms, as reported in a previous study (Bignardi et al., 2014).
C. Bignardi et al. / Food Control 71 (2017) 329e335
331
Fig. 1. Pictures of sample surfaces, recorded by digital microscope, showing increasing degrees of damage (from A to D).
9.42
100 80
m/z 366.1700
60 40 20 0 100
9.97
80
m/z 486.1911
60 40 20
13.32
0 100
9.64 m/z 481.2020
80 60 40
11.41
20
12.81 14.79 15.89 5.89 17.55 19.72 21.28 1. 22.94 24.59 2
0 100
27.18 .18 28.20 29.80
10.13 m/z 735.2963
80 60 40 20
11.27 12.58 13.93 .93 16.38 17.51 19.19 9 20 20.65 65 22.90 10.68
0 100 80
m/z 989.3906
60 40 20 1 13.00 11.41
0 0
2
4
6
8
10
12
14
16
18 1 20 Time (min)
22
24
26
28
30
32
Fig. 2. Chromatograms and chemical structures of the main PC degradation products identified.
Fig. 2 shows the peaks recorded, and the structures of the molecules corresponding to the main detected signals. Previous literature studies about polycarbonate ageing induced by light identified some of these molecules as PC-degradation products (Collin et al., 2012; Jang & Wilkie, 2004). The molecular weight values, comprised between 227 and 989 g/mol suggest that their ingestion may represent a possible risk for health. From a qualitative point of view, a similar chromatographic
profile was recorded for both simulants employed, although all signals found in isooctane were of lower intensity. An example of analysis, related to the sample named “yellow-2007”, is reported in Fig. 3, showing the comparison of the signal recorded for the molecule characterized by m/z ¼ 486.1911, in ethanol and in isooctane. The peak of interest eluted at a retention time of about 9.50 min, and its presence is clearly visible in all three migration steps performed. The recorded area decreased progressively from
332
C. Bignardi et al. / Food Control 71 (2017) 329e335
Fig. 3. Peaks recorded for the molecule characterized by m/z ¼ 486.1911 during three consecutive migration experiments (panels A, B, C) in the two simulants considered.
the first to the third experiment. From the comparison of the two simulants, it is clear that ethanol acts as a stronger extracting solvent. A similar behavior was recorded for all the other molecules detected. A remarkable interesting point was that the pattern of degradation products released by old and new PC objects was different. A previous study (Bignardi et al., 2015) reported higher content of BPA being released from old samples, while there are no studies exploring a possible correlation between ageing and amount of other degradation products. In order to put in evidence possible differences, the values of peak areas detected in ethanol 95% were considered for all molecules, and evaluated in relation to the age and the degree of damage of each sample. The peak area values of four of the molecules reported in Fig. 2 were correlated with sample age, surface damage degree and BPA; obtained values are reported in Table 1. It can be seen that, obviously, high correlation can be observed between damage degree, age, and BPA amount, according to that described elsewhere (Bignardi et al., 2015). As for larger fragments, it could be seen that
all peak areas considered were well correlated with age: in details, a high positive correlation was found with two molecules (481.2020 and 735.2963 m/z), and a sensible negative one with two other compounds (366.1700 and 486.1911 m/z). In particular, the first couple of molecules (481.2020 and 735.2963 m/z) was also correlated with damage degree and BPA amount. Therefore, the observed correlations allowed to group together two pairs of analytes showing similar features. Three-dimensional graphs reporting the relations observed about the molecules characterized by m/z ¼ 481.2020 (panel A) and m/z ¼ 735.2963 (panel B) are reported in Fig. 4. In both cases, it can be noticed that the recorded peak areas were higher in old samples, and were also well correlated with damage level (see Table 1). By looking at the molecule structures, it can be seen that both analytes are characterized by the presence of -OH groups on the rings, suggesting that they could derive from PC hydrolysis. This hypothesis was also supported by the high amounts of BPA found to migrate from the same samples; BPA can be, in fact, considered as the final product of the hydrolysis process.
Table 1 Correlations between peak areas of the four molecules, ageing, surface damage degree and BPA.
Area peak m/z 366.1710 Area peak m/z 481.2020 Area peak m/z 735.2963 Age Damage BPA
Area peak m/z 486.1911
Area peak m/z 366.1700
Area peak m/z 481.2020
Area peak m/z 735.2963
Age
Surface damage
0,78 0,35 0,29 0,59 0,04 0,24
0,45 0,43 0,58 0,32 0,32
0,82 0,86 0,74 0,88
0,87 0,81 0,86
0,90 0,77
0,71
C. Bignardi et al. / Food Control 71 (2017) 329e335
333
Fig. 4. 3D graph of peak areas of the molecule with m/z ¼ 481.2020 (panel A) and the molecule with m/z ¼ 735.2963 (panel B) plotted against age and surface damage.
Fig. 5. 3D graph of peak areas of the molecule with m/z ¼ 366.1700 (panel A) and the molecule with m/z ¼ 486.1911 (panel B) plotted against to age and surface damage.
A different behavior can be evidenced in Fig. 5 regarding the molecules characterized by m/z ¼ 366.1700 (panel A) and m/ z ¼ 486.1911 (panel B). The areas of the two compounds were highly correlated between each other, as reported in Table 1. The three-dimensional graphs show a negative correlation between them and the age of samples; in fact, higher area values were observed in new samples than in old ones. No correlation with damage level was recorded. By considering the molecule structures, it can be seen that only one -OH group is present on a side ring of the compounds. A possible hypothesis to explain the higher
amounts released by new objects, is that these compounds derive from uncomplete PC polymerization. They might be present on the surface of the material in form of unreacted molecules, and therefore easily released when the sample was put in contact with a solvent. We can conclude that two different populations of PCdegradation products are present: the first one composed by molecules deriving from polycarbonate damage, and the second composed by unreacted molecules, possibly formed as side reaction products during the polymerization process, and unable to react
334
C. Bignardi et al. / Food Control 71 (2017) 329e335
Fig. 6. Peaks recorded for the colouring agent Solvent Yellow 232 in a new and an old sample during three consecutive migration experiments (panels A, B, C) in the simulant ethanol 95%.
further. We can suggest that the ester linkage may be hydrolyzed during ageing, while unreacted chains are found in new samples. 3.2. Identification and migration of colouring agents Traces of colouring agents, not belonging to NIAS, but considered as unwanted migrants, were identified in migration experiments of some analysed samples, precisely in the red (solvent Red 179), the orange (solvent Yellow 184) and the yellow ones (solvent Yellow 232). Molecules found were belonging to the category of solvent dyes. Blue and white samples did not release any dye since they were probably coloured by inorganic pigments. Migration experiments were performed three consecutive times in both ethanol 95% and isoctane. Different peak intensity was obtained for the molecules in the two simulants. The amount of extracted dye was found to decrease during consecutive experiments in both solvents, but was still detectable until the third experiment only when ethanol 95% was employed. Therefore, ethanol was more effective in extracting the dye than isooctane. To evaluate the eventual effect of the ageing on dye release, experiments on two samples of the same shape and color (yellow) but different age (2013 vs. 2007) were performed. In Fig. 6, the chromatograms showing the peak referred to the yellow dye (identified as solvent Yellow 232) found in three consecutive migration experiments performed in ethanol 95% are reported. Signals corresponding to the yellow dye are present in both samples, in all experiments, and were found to decrease from the first to the third migration. From the comparison between the two samples it is evident that the new sample releases higher amount of dye. A similar result was obtained by the analysis of two equal orange samples differing only for the age (2014 vs 2009). These data, although limited to a restricted number of samples, suggest that new plastic objects may act as potential source of dye release. It
would be interesting to investigate in a deeper way such behavior, since according to the national regulation of many countries organic dyes should not migrate into food simulants (Council of Europe, 1989). 4. Conclusion Studies on plastic re-usable containers aimed at storing food are very limited and should be encouraged since normally the tests for checking material suitability are only performed on new samples, and are not repeated after ageing. Since these objects are usually employed during many years, despite ageing and surface damage, they obviously undergo to degradation, and may become a potential source of NIAS. The molecules generated by material decomposition can act as possible contaminants as they can easily reach the food products. It is the first time that PC degradation products have been identified in food simulants; besides, a peculiar behavior has been observed for different groups of compounds, suggesting the presence of two groups of oligomers diversely correlated with age and surface damage. It is also the first time that colouring agents have been found to migrate in simulants, and this behavior should be deeply investigated since they may represent a possible source of food contaminants. Conflict of interest The authors declare no conflict of interest. References Bignardi, C., Cavazza, A., Corradini, C., & Salvadeo, P. (2014). Targeted and untargeted data-dependent experiments for characterization of polycarbonate foodcontact plastics by ultra-high performance chromatography coupled to
C. Bignardi et al. / Food Control 71 (2017) 329e335 quadrupole Orbitrap tandem mass spectrometry. Journal of Chromatography A, 1372, 133e144. , C., Salvadeo, P., & Corradini, C. (2015). UHPLC-highBignardi, C., Cavazza, A., Lagana resolution mass spectrometry determination of bisphenol A and plastic additives released by polycarbonate tableware: Influence of ageing and surface damage. Analytical and Bioanalytical Chemistry, 407, 7917e7924. Brede, C., Fjeldal, P., Skjevrak, I., & Herikstad, H. (2003). Increased migration levels of bisphenol A from polycarbonate baby bottles after dishwashing, boiling and brushing. Food Additives and Contaminants, 20, 684e689. Brüschweiler, B. (2014). Assessment of non-intentionally added substances from food contact materials in food: Which way to go? Toxicology Letters, 229, S34. Canellas, E., Vera, P., & Nerin, C. (2015a). UPLC-ESI-Q-TOF-MS and GC-MS identification and quantification of non-intentionally added substances coming from biodegradable food packaging. Analytical and Bioanalytical Chemistry, 407, 6781e6790. Canellas, E., Vera, P., & Nerín, C. (2015b). Risk assessment derived from migrants identified in several adhesives commonly used in food contact materials. Food and Chemical Toxicology, 75, 79e87. re, P.-O., The rias, S., Lambert, J.-M., Perdereau, J., & Gardette, J.-L. Collin, S., Bussie (2012). Physicochemical and mechanical impacts of photo-ageing on bisphenol a polycarbonate. Polymer Degradation and Stability, 97, 2284e2293. Corradini, C., Alfieri, I., Cavazza, A., Lantano, C., Lorenzi, A., Zucchetto, N., et al. (2013). Antimicrobial films containing lysozyme for active packaging obtained by sol-gel technique. Journal of Food Engineering, 119, 580e587. Council of Europe, Committee of Ministers. (1989). On the use of colourants in plastic materials coming into contact with food. Resolution AP, 1e6. European Commission. (2011). Regulation (EC) No. 10/2011 of the European Parliament and of the Council of 14 January 2011 on plastic materials and articles intended to come into contact with food. Official Journal of European Union, L 12, 1e89. Factor, A. (1996). Mechanisms of thermal and photodegradations of bisphenol A polycarbonate. In: Polymer durability (pp. 59e76). ACS Pub. Gao, Y., Gu, Y., & Wei, Y. (2011). Determination of polymer additives, antioxidants
335
and ultraviolet (UV) absorbers by high-performance liquid chromatography coupled with UV photodiode array detection in food simulants. Journal of Agriculture and Food Chemistry, 59, 12982e12989. Harvey, J. A. (2012). Chemical and physical aging of plastics. In M. Kutz (Ed.), Handbook of environmental degradation of materials (pp. 195e210). Delmar, New York: Myer Kutz Assoc Inc. Hoekstra, E. J., & Simoneau, C. (2013). Release of bisphenol A from polycarbonate e A review. Critical Reviews in Food Science and Nutrition, 53, 386e402. Hollnagel, H. M., van Herwijnen, P., & Sura, P. (2014). Assessing safety of nonintentionally added substances in polymers used for food contact applications. Toxicology Letters, 229(Suppl), S34. Hosseinnejad, M. (2014). Active packaging for food applications e A review. International Journal of Advanced Biological and Biomedical Research, 2, 1174e1180. Jang, B. N., & Wilkie, C. A. (2004). A TGA/FTIR and mass spectral study on the thermal degradation of bisphenol A polycarbonate. Polymer Degradation and. Stability, 86, 419e430. Ketelslegers, H. (2014). Non-intentionally added substances in food contact materials: Eu regulatory requirements and safety assessment. Toxicology Letters, 229(Suppl), S33. Lantano, C., Alfieri, I., Cavazza, A., Corradini, C., Lorenzi, A., Zucchetto, N., et al. (2014). Natamycin based sol-gel antimicrobial coatings on polylactic acid films for food packaging. Food Chemistry, 165, 342e347. Lau, O. W., & Wong, S. K. (2000). Contamination in food from packaging material. Journal of Chromatography A, 882, 255e270. ~ o, C. (2013). The challenge of identifying Nerin, C., Alfaro, P., Aznar, M., & Domen non-intentionally added substances from food packaging materials: A review. Analytica Chimica Acta, 775, 14e24. n García, R., Cooperb, I., Franz, R., & Paseiro Losada, P. (2006). Sanches Silva, A., Sendo Compilation of analytical methods and guidelines for the determination of selected model migrants from plastic packaging. Trends in Food Science & Technology, 17, 535e546. Struik, L. C. E. (1977). Physical aging in plastics and other glassy materials. Polymer Engineering & Science, 17, 165e173.
Contents lists available at ScienceDirect
Food Control journal homepage: www.elsevier.com/locate/foodcont
Release of non-intentionally added substances (NIAS) from food contact polycarbonate: Effect of ageing , Paola Salvadeo, Claudio Corradini Chiara Bignardi, Antonella Cavazza*, Carmen Lagana degli Studi di Parma, Parco Area delle Scienze 17/A, 43124, Italy Dipartimento di Chimica, Universita
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2016 Received in revised form 27 June 2016 Accepted 14 July 2016 Available online 16 July 2016
Non-intentionally added substances (NIAS) released from polycarbonate tableware were evaluated through untargeted analysis by UHPLC-ESI-Orbitrap. Migration experiments from fifteen samples of different age were performed using ethanol 95% and isooctane as food simulants. High-resolution mass spectrometry permitted the identification of oligomers derived from polycarbonate degradation and traces of colouring agents. Data were analysed with the aim of exploring a possible correlation between type and amount of different oligomers and age of the samples. It has been found out for the first time that the pattern of polycarbonate degradation products observed in new samples was different from that of old samples. Colouring agents were found to be released in higher amount from new samples than from old ones. The work shows the high potential of the technique employed and its importance in the field of control related to safety concern. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Polycarbonate Migration Degradation products Colourants Ageing
1. Introduction The migration of substances from materials in contact with food is considered a positive event when active packaging is realized (Hosseinnejad, 2014). In such case, molecules such as antioxidants (Corradini et al., 2013) and antimicrobials (Lantano et al., 2014) are intentionally incorporated in a plastic film put in contact with a food product with the goal to protect it from spoilage. On the other side, migration from plastic is undesired when potential contaminants occurring in the material reach a food product and may act as toxic compounds affecting health (Lau & Wong, 2000). Therefore, to establish the identity of all substances released by plastic, and evaluate the main parameters affecting their migration (time of contact, temperature, type of food, etc.) is of paramount importance (Hoekstra & Simoneau, 2013). This matter does not regard only food, but involves the pharmaceutical and cosmetic fields too, since those products are often packaged in plastic, and many medical devices are made of plastic. Besides, cosmetics regulation does not require the indication of expiring date, therefore products can be stored for many years in plastic and/ or may remain for a long time on shelves before consumption. Furthermore, products such as sunscreens are generally subjected
* Corresponding author. E-mail address: [email protected] (A. Cavazza). http://dx.doi.org/10.1016/j.foodcont.2016.07.013 0956-7135/© 2016 Elsevier Ltd. All rights reserved.
to high temperature when left for several hours under the sun. In addition, it is worth to consider that the process of migration is not only related to storage and packaging, but also to industrial processing and manufacturing; in fact industrial plants often contain plastic parts, and contact at high temperature and for long time may occur. Many molecules representing potential migrants are well known, and some of them, such as bisphenol A (BPA) and additives including antioxidants and UV absorbers have been studied since n García, Cooperb, Franz & Paseiro long time (Sanches Silva, Sendo Losada 2006; Gao, Gu, & Wei, 2011). Besides, in parallel to a target analysis, it is important to take into account the possible release of unknown substances, such as the so-called non-intentionally added substances (NIAS) that sometimes are found in a product without a clear explanation about their origin. They may represent impurities of raw materials employed, or may derive from degradation of the material components (Nerin, Alfaro, Aznar, ~ o, 2013). The identification of these molecules, together & Domen with the estimation of a potential threshold effect for risk assessment is a relevant issue that deserves great attention in terms of safety concern (Brüschweiler, 2014; Hollnagel, van Herwijnen, & Sura, 2014; Ketelslegers, 2014). In this context, the development of innovative analytical methods based on the employment of modern techniques such as high resolution mass spectrometry (HR-MS) has been demonstrated to be very useful and effective for the identification of
330
C. Bignardi et al. / Food Control 71 (2017) 329e335
unknown compounds released by plastics used in food contact materials (Bignardi, Cavazza, Corradini, & Salvadeo, 2014; Canellas, Vera, & Nerin, 2015a; Canellas, Vera, Ner, í, & n, 2015b). An important point that it is worth to be considered, regards the re-usable plastic containers aimed at storing food products, since they are submitted not only to ageing process, but also to mechanical damage due to the repeated use and frequent washing cycles. Factors such as high temperature and ultravioletevisible light exposure, humidity, atmospheric components, contact with liquids are responsible for chemical and physical ageing (Factor, 1996) that provoke the formation of tiny scratches and small cracks on the surface. As a direct consequence, the object becomes opaque and foggy, and formation of micro-cavities occurs. All these phenomena can be believed to be responsible for the release of polymer components (Struik, 1977; Harvey, 2012), and previous studies showed an increment of bisphenol A released from aged and damaged bottles and tableware respect to unused (Bignardi, , Salvadeo, & Corradini, 2015; Brede, Fjeldal, Cavazza, Lagana Skjevrak, & Herikstad, 2003). In this work, the attention was focused on the untargeted analysis of simulants put in contact with re-usable objects made of polycarbonate (PC) allowed for food contact, and in particular with the aim of identifying molecules derived from a possible degradation of the material during ageing. UHPLC coupled to HR-MS has been applied for the identification of oligomers. The study has been carried out by employing two different food simulants. Previous studies reported the identification of some molecules deriving by PC degradation after ageing and thermal decomposition of PC (Collin et al., 2012; Jang & Wilkie, 2004), nevertheless, no literature dealing with the possible migration of such substances in simulants and/or in food products can be actually found. Similarly, there are no studies about possible migration of colouring agents from tableware, although this topic should be of great interest in the safety assessment (Council of Europe, 1989). 2. Materials and methods 2.1. Chemicals All chemicals employed were of analytical reagent grade. Methanol and water used for chromatographic eluent preparation were of UHPLCeMS grade and purchased by Sigma Aldrich (Milan, Italy) as well as ammonium formate, ethanol and isooctane. Calibration solutions Pierce LTQ Velos ESI Positive ion and Pierce LTQ Velos ESI Negative ion from Thermo Fisher Scientific (Rockford, IL, USA) were used to calibrate the mass spectrometer. 2.2. Samples Fifteen polycarbonate tableware objects of different age (between 2 and 14 years old) and color were submitted to analysis. Samples were named accordingly to the age and the color. All samples were purchased from the same producer. Between them, three samples were new and never been used. Migration tests were carried out by employing ethanol 95% (v/v) and isooctane, reported to be substitutes for the food simulants D2 (EU Regulation 10/2011 on plastic materials and articles intended to come into contact with food, European Commission, 2011). The choice of these simulants derived by the scarce interest of previous studies on migration from plastic objects to fats and oils. Containers were filled with the simulant and placed in a climatic chamber at 40 C for 1 h, covering the surface with glass plates in order to prevent solvent evaporation. Experiments were repeated three consecutive times. Aliquots obtained were evaporated and redissolved in 1 mL of ethanol. Samples were filtered and submitted
to UHPLC-MS analysis. The degree of damage of the PC has been evaluated by examination of the surface by a lens (10) and a digital microscope. To each sample, a label corresponding to numbers from 1 to 5 was assigned according to the amount of scratches or cracks occurring on its surface. Label 1 was assigned to new and unused samples, characterized by a perfectly clean surface, while higher numbers were assigned progressively to more damaged surfaces (Bignardi et al., 2015). Some examples of the surface appearance are depicted in Fig. 1, showing different degrees of damage (increasing from A to D). 2.3. UHPLC-HRMS analysis Analyses were performed by UHPLC-HRMS (Thermo Scientific Ultimate 3000 RSLC) nano system operating in capillary-flow mode coupled to a Q Exactive Mass spectrometer (Thermo Scientific, Fremont, CA). Chromatographic separations were achieved on a C18 Acclaim PepMap RSLC (Thermo Scientific, Fremont, CA) column (150 mm 0.3 mm, 2.0 mm particle size) thermostated at 35 C. Injection volume was 1 mL. Eluents, delivered at a flow-rate of 10 mL min1, were as follows: eluent A, ammonium formate 1 mM in 10:90 methanol:water v/v; eluent B, methanol containing 1 mM ammonium formate. Solvent B was set at 40% and delivered by a linear gradient to 99% in 20 min. Q-Exactive instrument was equipped with a pneumatically assisted ESI interface with a stainless steel needle adapted for capillary flow. Conditions of the interface were fixed according to a previous reported method (Bignardi et al., 2014). Mass calibration was performed every three days in order to obtain a mass accuracy lower than 2 ppm. Chromeleon 6.8 and XCalibur 2.2 softwares (Thermo Fisher Scientific MA, USA) were used to control the instrument and for data processing. Signal acquisition was performed by full-MS-data dependent MS/MS experiments with inclusion list in the range 90e1000 m/z, in both positive and negative mode. The extraction from full MS of each parent ion taken into account allowed the visualization of the isotopic pattern and the matching between the exact mass and the accurate mass. MS/MS spectra obtained in the same run were used for confirmatory purpose (Bignardi et al., 2015). 2.4. Statistical analysis Means, standard deviations (SDs) and correlation values were calculated with SPSS (version 19.0, SPSS Inc., Chicago, Illinois, USA) statistical software. 3D plots of degradation products areas monitored were obtained with the Statgraphics Centurion 16.1 software. 3. Results Migration tests were carried out by employing two different food simulants, ethanol 95% (v/v) and isooctane. Since no standard corresponding to the identified molecules could be available, peak areas normalized respect to the surface in contact with simulant were used for a semi-quantitative approach. 3.1. Data on products of degradation The presence of several oligomers derived from PC chain was evidenced in the examined simulants. The molecules identification was carried out by reconstructing the mass fragmentation spectrum of the most representative molecular ions visible in the fullscan chromatograms, as reported in a previous study (Bignardi et al., 2014).
C. Bignardi et al. / Food Control 71 (2017) 329e335
331
Fig. 1. Pictures of sample surfaces, recorded by digital microscope, showing increasing degrees of damage (from A to D).
9.42
100 80
m/z 366.1700
60 40 20 0 100
9.97
80
m/z 486.1911
60 40 20
13.32
0 100
9.64 m/z 481.2020
80 60 40
11.41
20
12.81 14.79 15.89 5.89 17.55 19.72 21.28 1. 22.94 24.59 2
0 100
27.18 .18 28.20 29.80
10.13 m/z 735.2963
80 60 40 20
11.27 12.58 13.93 .93 16.38 17.51 19.19 9 20 20.65 65 22.90 10.68
0 100 80
m/z 989.3906
60 40 20 1 13.00 11.41
0 0
2
4
6
8
10
12
14
16
18 1 20 Time (min)
22
24
26
28
30
32
Fig. 2. Chromatograms and chemical structures of the main PC degradation products identified.
Fig. 2 shows the peaks recorded, and the structures of the molecules corresponding to the main detected signals. Previous literature studies about polycarbonate ageing induced by light identified some of these molecules as PC-degradation products (Collin et al., 2012; Jang & Wilkie, 2004). The molecular weight values, comprised between 227 and 989 g/mol suggest that their ingestion may represent a possible risk for health. From a qualitative point of view, a similar chromatographic
profile was recorded for both simulants employed, although all signals found in isooctane were of lower intensity. An example of analysis, related to the sample named “yellow-2007”, is reported in Fig. 3, showing the comparison of the signal recorded for the molecule characterized by m/z ¼ 486.1911, in ethanol and in isooctane. The peak of interest eluted at a retention time of about 9.50 min, and its presence is clearly visible in all three migration steps performed. The recorded area decreased progressively from
332
C. Bignardi et al. / Food Control 71 (2017) 329e335
Fig. 3. Peaks recorded for the molecule characterized by m/z ¼ 486.1911 during three consecutive migration experiments (panels A, B, C) in the two simulants considered.
the first to the third experiment. From the comparison of the two simulants, it is clear that ethanol acts as a stronger extracting solvent. A similar behavior was recorded for all the other molecules detected. A remarkable interesting point was that the pattern of degradation products released by old and new PC objects was different. A previous study (Bignardi et al., 2015) reported higher content of BPA being released from old samples, while there are no studies exploring a possible correlation between ageing and amount of other degradation products. In order to put in evidence possible differences, the values of peak areas detected in ethanol 95% were considered for all molecules, and evaluated in relation to the age and the degree of damage of each sample. The peak area values of four of the molecules reported in Fig. 2 were correlated with sample age, surface damage degree and BPA; obtained values are reported in Table 1. It can be seen that, obviously, high correlation can be observed between damage degree, age, and BPA amount, according to that described elsewhere (Bignardi et al., 2015). As for larger fragments, it could be seen that
all peak areas considered were well correlated with age: in details, a high positive correlation was found with two molecules (481.2020 and 735.2963 m/z), and a sensible negative one with two other compounds (366.1700 and 486.1911 m/z). In particular, the first couple of molecules (481.2020 and 735.2963 m/z) was also correlated with damage degree and BPA amount. Therefore, the observed correlations allowed to group together two pairs of analytes showing similar features. Three-dimensional graphs reporting the relations observed about the molecules characterized by m/z ¼ 481.2020 (panel A) and m/z ¼ 735.2963 (panel B) are reported in Fig. 4. In both cases, it can be noticed that the recorded peak areas were higher in old samples, and were also well correlated with damage level (see Table 1). By looking at the molecule structures, it can be seen that both analytes are characterized by the presence of -OH groups on the rings, suggesting that they could derive from PC hydrolysis. This hypothesis was also supported by the high amounts of BPA found to migrate from the same samples; BPA can be, in fact, considered as the final product of the hydrolysis process.
Table 1 Correlations between peak areas of the four molecules, ageing, surface damage degree and BPA.
Area peak m/z 366.1710 Area peak m/z 481.2020 Area peak m/z 735.2963 Age Damage BPA
Area peak m/z 486.1911
Area peak m/z 366.1700
Area peak m/z 481.2020
Area peak m/z 735.2963
Age
Surface damage
0,78 0,35 0,29 0,59 0,04 0,24
0,45 0,43 0,58 0,32 0,32
0,82 0,86 0,74 0,88
0,87 0,81 0,86
0,90 0,77
0,71
C. Bignardi et al. / Food Control 71 (2017) 329e335
333
Fig. 4. 3D graph of peak areas of the molecule with m/z ¼ 481.2020 (panel A) and the molecule with m/z ¼ 735.2963 (panel B) plotted against age and surface damage.
Fig. 5. 3D graph of peak areas of the molecule with m/z ¼ 366.1700 (panel A) and the molecule with m/z ¼ 486.1911 (panel B) plotted against to age and surface damage.
A different behavior can be evidenced in Fig. 5 regarding the molecules characterized by m/z ¼ 366.1700 (panel A) and m/ z ¼ 486.1911 (panel B). The areas of the two compounds were highly correlated between each other, as reported in Table 1. The three-dimensional graphs show a negative correlation between them and the age of samples; in fact, higher area values were observed in new samples than in old ones. No correlation with damage level was recorded. By considering the molecule structures, it can be seen that only one -OH group is present on a side ring of the compounds. A possible hypothesis to explain the higher
amounts released by new objects, is that these compounds derive from uncomplete PC polymerization. They might be present on the surface of the material in form of unreacted molecules, and therefore easily released when the sample was put in contact with a solvent. We can conclude that two different populations of PCdegradation products are present: the first one composed by molecules deriving from polycarbonate damage, and the second composed by unreacted molecules, possibly formed as side reaction products during the polymerization process, and unable to react
334
C. Bignardi et al. / Food Control 71 (2017) 329e335
Fig. 6. Peaks recorded for the colouring agent Solvent Yellow 232 in a new and an old sample during three consecutive migration experiments (panels A, B, C) in the simulant ethanol 95%.
further. We can suggest that the ester linkage may be hydrolyzed during ageing, while unreacted chains are found in new samples. 3.2. Identification and migration of colouring agents Traces of colouring agents, not belonging to NIAS, but considered as unwanted migrants, were identified in migration experiments of some analysed samples, precisely in the red (solvent Red 179), the orange (solvent Yellow 184) and the yellow ones (solvent Yellow 232). Molecules found were belonging to the category of solvent dyes. Blue and white samples did not release any dye since they were probably coloured by inorganic pigments. Migration experiments were performed three consecutive times in both ethanol 95% and isoctane. Different peak intensity was obtained for the molecules in the two simulants. The amount of extracted dye was found to decrease during consecutive experiments in both solvents, but was still detectable until the third experiment only when ethanol 95% was employed. Therefore, ethanol was more effective in extracting the dye than isooctane. To evaluate the eventual effect of the ageing on dye release, experiments on two samples of the same shape and color (yellow) but different age (2013 vs. 2007) were performed. In Fig. 6, the chromatograms showing the peak referred to the yellow dye (identified as solvent Yellow 232) found in three consecutive migration experiments performed in ethanol 95% are reported. Signals corresponding to the yellow dye are present in both samples, in all experiments, and were found to decrease from the first to the third migration. From the comparison between the two samples it is evident that the new sample releases higher amount of dye. A similar result was obtained by the analysis of two equal orange samples differing only for the age (2014 vs 2009). These data, although limited to a restricted number of samples, suggest that new plastic objects may act as potential source of dye release. It
would be interesting to investigate in a deeper way such behavior, since according to the national regulation of many countries organic dyes should not migrate into food simulants (Council of Europe, 1989). 4. Conclusion Studies on plastic re-usable containers aimed at storing food are very limited and should be encouraged since normally the tests for checking material suitability are only performed on new samples, and are not repeated after ageing. Since these objects are usually employed during many years, despite ageing and surface damage, they obviously undergo to degradation, and may become a potential source of NIAS. The molecules generated by material decomposition can act as possible contaminants as they can easily reach the food products. It is the first time that PC degradation products have been identified in food simulants; besides, a peculiar behavior has been observed for different groups of compounds, suggesting the presence of two groups of oligomers diversely correlated with age and surface damage. It is also the first time that colouring agents have been found to migrate in simulants, and this behavior should be deeply investigated since they may represent a possible source of food contaminants. Conflict of interest The authors declare no conflict of interest. References Bignardi, C., Cavazza, A., Corradini, C., & Salvadeo, P. (2014). Targeted and untargeted data-dependent experiments for characterization of polycarbonate foodcontact plastics by ultra-high performance chromatography coupled to
C. Bignardi et al. / Food Control 71 (2017) 329e335 quadrupole Orbitrap tandem mass spectrometry. Journal of Chromatography A, 1372, 133e144. , C., Salvadeo, P., & Corradini, C. (2015). UHPLC-highBignardi, C., Cavazza, A., Lagana resolution mass spectrometry determination of bisphenol A and plastic additives released by polycarbonate tableware: Influence of ageing and surface damage. Analytical and Bioanalytical Chemistry, 407, 7917e7924. Brede, C., Fjeldal, P., Skjevrak, I., & Herikstad, H. (2003). Increased migration levels of bisphenol A from polycarbonate baby bottles after dishwashing, boiling and brushing. Food Additives and Contaminants, 20, 684e689. Brüschweiler, B. (2014). Assessment of non-intentionally added substances from food contact materials in food: Which way to go? Toxicology Letters, 229, S34. Canellas, E., Vera, P., & Nerin, C. (2015a). UPLC-ESI-Q-TOF-MS and GC-MS identification and quantification of non-intentionally added substances coming from biodegradable food packaging. Analytical and Bioanalytical Chemistry, 407, 6781e6790. Canellas, E., Vera, P., & Nerín, C. (2015b). Risk assessment derived from migrants identified in several adhesives commonly used in food contact materials. Food and Chemical Toxicology, 75, 79e87. re, P.-O., The rias, S., Lambert, J.-M., Perdereau, J., & Gardette, J.-L. Collin, S., Bussie (2012). Physicochemical and mechanical impacts of photo-ageing on bisphenol a polycarbonate. Polymer Degradation and Stability, 97, 2284e2293. Corradini, C., Alfieri, I., Cavazza, A., Lantano, C., Lorenzi, A., Zucchetto, N., et al. (2013). Antimicrobial films containing lysozyme for active packaging obtained by sol-gel technique. Journal of Food Engineering, 119, 580e587. Council of Europe, Committee of Ministers. (1989). On the use of colourants in plastic materials coming into contact with food. Resolution AP, 1e6. European Commission. (2011). Regulation (EC) No. 10/2011 of the European Parliament and of the Council of 14 January 2011 on plastic materials and articles intended to come into contact with food. Official Journal of European Union, L 12, 1e89. Factor, A. (1996). Mechanisms of thermal and photodegradations of bisphenol A polycarbonate. In: Polymer durability (pp. 59e76). ACS Pub. Gao, Y., Gu, Y., & Wei, Y. (2011). Determination of polymer additives, antioxidants
335
and ultraviolet (UV) absorbers by high-performance liquid chromatography coupled with UV photodiode array detection in food simulants. Journal of Agriculture and Food Chemistry, 59, 12982e12989. Harvey, J. A. (2012). Chemical and physical aging of plastics. In M. Kutz (Ed.), Handbook of environmental degradation of materials (pp. 195e210). Delmar, New York: Myer Kutz Assoc Inc. Hoekstra, E. J., & Simoneau, C. (2013). Release of bisphenol A from polycarbonate e A review. Critical Reviews in Food Science and Nutrition, 53, 386e402. Hollnagel, H. M., van Herwijnen, P., & Sura, P. (2014). Assessing safety of nonintentionally added substances in polymers used for food contact applications. Toxicology Letters, 229(Suppl), S34. Hosseinnejad, M. (2014). Active packaging for food applications e A review. International Journal of Advanced Biological and Biomedical Research, 2, 1174e1180. Jang, B. N., & Wilkie, C. A. (2004). A TGA/FTIR and mass spectral study on the thermal degradation of bisphenol A polycarbonate. Polymer Degradation and. Stability, 86, 419e430. Ketelslegers, H. (2014). Non-intentionally added substances in food contact materials: Eu regulatory requirements and safety assessment. Toxicology Letters, 229(Suppl), S33. Lantano, C., Alfieri, I., Cavazza, A., Corradini, C., Lorenzi, A., Zucchetto, N., et al. (2014). Natamycin based sol-gel antimicrobial coatings on polylactic acid films for food packaging. Food Chemistry, 165, 342e347. Lau, O. W., & Wong, S. K. (2000). Contamination in food from packaging material. Journal of Chromatography A, 882, 255e270. ~ o, C. (2013). The challenge of identifying Nerin, C., Alfaro, P., Aznar, M., & Domen non-intentionally added substances from food packaging materials: A review. Analytica Chimica Acta, 775, 14e24. n García, R., Cooperb, I., Franz, R., & Paseiro Losada, P. (2006). Sanches Silva, A., Sendo Compilation of analytical methods and guidelines for the determination of selected model migrants from plastic packaging. Trends in Food Science & Technology, 17, 535e546. Struik, L. C. E. (1977). Physical aging in plastics and other glassy materials. Polymer Engineering & Science, 17, 165e173.