Journal of Chromatography A, 1372 (2014) 133–144

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Targeted and untargeted data-dependent experiments for characterization of polycarbonate food-contact plastics by ultra high performance chromatography coupled to quadrupole orbitrap tandem mass spectrometry Chiara Bignardi, Antonella Cavazza, Claudio Corradini ∗ , Paola Salvadeo Dipartimento di Chimica, Università degli Studi di Parma, Parco Area delle Scienze 17/A, 43124, Parma, Italy

a r t i c l e

i n f o

Article history: Received 3 September 2014 Received in revised form 24 October 2014 Accepted 27 October 2014 Available online 7 November 2014 Keywords: Polycarbonate Bisphenol A Plastic additives Organic colorants High-resolution tandem mass spectrometry Data-dependent acquisition Food-contact material

a b s t r a c t Materials that come in contact with foods are potential sources of chemical food contamination. Consequently, characterization of their composition is of paramount importance considering the possible occurrence of several unknown molecules such as non-intentionally added substance (NIAS), residual monomers, degradation products, plastic additives and organic colorants. Previous studies concerning the characterization in terms of composition are focalized in the recognition of additives. To the best author’s knowledge there are no scientific data about the composition of a plastic material in terms of colorants. In this work, an analytical method employing capillary ultra high-performance liquid chromatography and electrospray ionization quadrupole orbitrap high-resolution mass spectrometry (UHPLC–ESI Qorbitrap) was exploited for characterization of polycarbonate for food contact material. Data-dependent experiments for targeted and untargeted analysis were employed after a total dissolution of polycarbonate samples and extraction of its components. The presence of common additives such as antioxidants and UV absorbers was confirmed by targeted analysis, while, the untargeted approach combined with the high mass accuracy of orbitrap technology allowed to identify for the first time some polycarbonate degradation products and the organic dyes effectively used for the coloration of plastic objects intended to come in contact with food. The present study shows the high potential of this technique in the field of material characterization aimed at food safety evaluation. © 2014 Published by Elsevier B.V.

1. Introduction Food contact materials are all materials and articles intended to come in contact with food, such as packaging and containers, kitchen equipment, cutlery and dishes. The safety of materials in contact with food must be evaluated since some molecules could migrate from materials into food. Materials should be manufactured in compliance with EU regulation, which also require good manufacturing practices, so that any potential transfer to foods does not raise safety concerns, change the composition of the food in an unacceptable way, or have adverse effects on the taste and odor of foods [1–3]. Plastic materials are composed by monomers and other starting substances transformed through chemical reactions in a polymer, which represents the principal component.

∗ Corresponding author. Tel.: +39 0521 906023; fax: +39 0521 905557. E-mail address: [email protected] (C. Corradini). http://dx.doi.org/10.1016/j.chroma.2014.10.104 0021-9673/© 2014 Published by Elsevier B.V.

Additives such as antioxidants, stabilizers and plasticizers have a major influence in the processing and shelf-life of plastics and are responsible for many properties of these materials. They are present in small amounts in plastics (generally ranging from 0.1% and 1%), dispersed in the polymer matrix, with the aim of avoiding effects such thermo-oxidative deterioration, which initiates scission and cross-linking of the macromolecular chains and consequently leads to polymer deterioration [4]. The polymer has got an inert structure with an high molecular weight that represents a low potential risk for human health since the organism cannot absorb molecules with a molecular weight greater than 1000 Da. On the contrary, as plastic additives and organic colorants have generally low molecular weight, they may migrate from plastics into foods, representing a potential risk for human health. Moreover, in polymer matrices it is possible to find monomers and oligomers that have not reacted in the polymerization reactions and nonintentionally added substances (NIAS) such as impurities of plastic additives [4].

134

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

Polycarbonate (PC) is one of the high-performance heterochain polymeric materials that comprise the family of engineering thermoplastics with a wide variety of applications due to excellent mechanical properties, high impact strength, heat resistance and high modulus of elasticity, as well as excellent toughness, clarity and transparency. These properties make it an ideal choice for tableware, microwave ovenware, reusable bottles, food storage containers and water pipes. Bisphenol A is used as monomer for PC production, and it is considered to be an endocrine disrupter, as first reported in 1993 [5]. Recent studies indicate the potential of BPA to disrupt thyroid hormone action [6], cause proliferation of human prostate cancer cells [7] and block testosterone synthesis [8] at very low part-per-trillion doses. Use of BPA in food contact materials is permitted in the European Union (EU) under Regulation 10/2011/EU [3], although in January 2011, the European Commission adopted Directive 2011/8/EU [9], prohibiting its use for the manufacture of polycarbonate infant feeding bottles. The heightened interest in the safety of BPA used food contact applications has resulted in increased public awareness as well as scientific interest. Consequently, appropriate, reliable methodologies are crucial for both industrial and enforcement testing of compliance with the legislation, and for risk assessment [10,11]. Interest towards BPA gave rise to the development of many analytical methods with LODs low enough to assess the human exposure at low-dose [12], mainly carried out by gas or liquid chromatography coupled to fluorescent or UV detection, and mass spectrometry [13–18] aimed at its quantification in food simulants after migration tests [19–23] and also in real samples [24–26]. Nevertheless, apart from possible BPA release, PC contains additives and other low molecular weight compounds, which may represent potential migrating substances into food. In the last years, some analytical methods have been applied in the field of research on plastic additives. Two basic different approaches are pursued: one involving the extraction of additives from the polymer matrix followed by analysis; the second based upon their direct analysis within the intact polymer by ambient ionization mass spectrometry technologies, as very recently reviewed [27]. A number of papers employing HPLC–MS for the analysis of plastic additives in standard solution [28–33] and in real extracts [34–37] can be found in the recent literature. High attention has also lately been directed towards non-intentionally added substances (NIAS) possibly occurring. On the other hand, there is a lack of studies concerning the identification of organic colorants used in plastic materials intended to come in contact with food. Organic colorants are generally low-molecular weight compounds and, for this reason, represent, like additives, a potential risk for human health. Another important point is the identification of potential BPA-polycarbonate degradation products with molecular weight below 1000 Da, since the works present in literature are principally focused on the polycarbonate monomer and/or on its degradation products like 9 9-dimethylxanthene [14,20]. The present study reports a new analytical method developed by capillary ultra high-performance liquid chromatography (UHPLC) coupled to electrospray ionization high-resolution mass spectrometry (HRMS) employing the Q-exactive, a new benchtop mass spectrometer equipped with orbitrap technology, successfully applied in the field of proteomics [38,39] and drug discovery [40–44]. The high resolution mass spectrometry gives the advantages of unambiguously indentify a molecule thanks to the high mass accuracy, and this is crucial when unknown compounds have to be identified. A polymer extract is a very complex matrix and in this context, high resolution is decisive in the discrimination of very similar compounds, as for example oligomers derived from polymer scissions. The identification of the compounds of interest,

such as plastic additives, NIAS, colorants and BPA-polycarbonate degradation products was achieved by performing targeted and untargeted experiments in both positive and negative ionization mode. This work shows some examples of data obtained by the application of this innovative technology in the detection of several classes of substances occurring in PC plastic materials, and demonstrates the applicability of the method to a possible useful survey for safety assessment. 2. Materials and methods 2.1. Chemicals All chemicals were of analytical reagent grade. Methanol and water used as eluents were of UHPLC–MS grade and were purchased by Sigma Aldrich (Milan, Italy). Ammonium formate used as additive in eluents, chloroform, hexane and acetone was also purchased by Sigma Aldrich (Milan, Italy). Pierce LTQ Velos ESI Positive ion and Pierce LTQ Velos ESI Negative ion calibration solutions from Thermo Fisher Scientific (Rockford, IL, USA) were used to calibrate the mass spectrometer. The following compounds, which can be likely found in polycarbonate as monomers and additives, or degradation compounds, were selected: 2,2-bis(4-hydroxyphenyl)propane (BPA), BPA diglycidylether (BADGE), 2,2-Bis(4-hydroxyphenyl)propaned16 (BPA-d16 ), 2,2 -methylenebis(4-methyl-6-tert-butylphenol) (Cyanox 2246), 2-hydroxy-4-methoxybenzophenone (Cyasorb UV9), 2,4-dihydroxybenzophenone (Uvinul 400), 2,2 dihydroxy-4,4 -dimethoxybenzophenone (Cyasorb UV12), 2,2 dihydroxy-4-methoxybenzophenone (Cyasorb UV24), 2-(2hydroxy-3-tert-butyl-5-methylphenyl)-2H-5-chlorobenzotriazole (Tinuvin 326), 2-hydroxy-4-n-octyloxybenzophenone (Chimassorb 81), 2-(2-hydroxy-3,5-di-tert-butylphenyl)-5-chlorob(Tinuvin 327), 1,3,5-trimethyl-2,4,6-tris(3,5enzotriazole di-tert-butyl-4-hydroxybenzyl)benzene (Irganox 1330), tris(2,4-di-tert-butylphenyl)phosphate (Irgafos 168), 2-(2 -hydroxy5 -tert-octylphenyl) benzotriazole (Cyasorb UV5411), octadecyl 3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate (Irganox 1076), 2,5-bis(5-tert-butyl-benzoxazol-2-yl)thiophene (Uvitex OB), 2(2H-benzotriazol-2-yl)-4,6-bis(1-methyl-1-phenylethyl)phenol (Tinuvin 234), 2-(2H-benzotriazol-2-yl)-4,6-di-tert-pentylphenol (Tinuvin 328), 2,6-di-tert-butyl-4-methylphenol (BHT), butylated hydroxyanisole (BHA), 6,6 -di-tert-butyl-2,2 -thiodi-p-cresol (Irganox 1081), didodecyl 3,3 -thiodipropionate (Advastab and pentaerythritol tetrakis(3-(3,5-di-tert-butyl-4800) hydroxyphenyl)propionate) (Irganox 1010) were purchased from Sigma Aldrich (Milan, Italy). Standard solutions of BPA, BADGE, Cyasorb UV 9, Cyasorb UV 12, Cyasorb UV 24 and Uvinul 400 at the concentration of 500 mg L−1 were prepared in methanol; all the other standard solutions at the same concentration were prepared in acetone. 2.2. Sample treatment and recovery Four samples of PC tableware of different color (red, yellow, pink and orange) were investigated. The best procedure for cutting pieces of the samples to submit to extraction was optimized with the aim to avoid any thermal treatment. A hollow punch having a diameter of 25 mm was used to collect samples. A crucial problem in the analysis of the compounds object of this study is that their molecules are inherently ubiquitous in the laboratory environment, and then could be introduced into the sample during sample treatment. Therefore, to preserve the sample solutions from any contamination due to possible contact with plastic material, all the laboratory procedures were performed employing

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

glassware previously washed with distilled water and then with acetone and hexane and dried in oven at 80 ◦ C, as reported in literature [45]. All the above procedure was also extended to glass syringes and pipettes. Only PTFE filters were employed, which were previously washed with methanol, and then emptied to remove any solvent residue. Polycarbonate extracts were prepared following a method reported in literature [20] and slightly modified as follows: small pieces of PC (1.5 g) were transferred to a 100 mL round flask. 12 mL of chloroform were added, and the sample was shaken with magnetic stirring to avoid adhesion of the polymer to the flask until dissolution was complete, which required about 40 min. Then, 18 mL of methanol were added with vigorous hand shaking to reprecipitate the polymer. This extract was recuperated and transferred into a clean round flask. The flask was rinsed with four additional washes with 10 mL of methanol (×2) and 10 mL of acetone (×2). The extract was evaporated by rotary evaporator and resuspended by 2 mL of methanol or acetone, and, after sonication for 4 min to remove precipitate from the flask walls, it was filtered through 0.45 and 0.22 ␮m PTFE filters. Blank samples were prepared in parallel, following the same procedures. Recovery values were calculated in order to verify the extraction yields of the sample treatment by analyzing in duplicate a polycarbonate sample fortified before dissolution at one concentration levels (5 mg L−1 ) with a standard mix solution containing analytes not detected in that PC sample: Uvitex OB, Tinuvin 326 and BPA-d16 .

2.3. Ultra high performance liquid chromatography–tandem high resolution mass spectrometry analysis (UHPLC-HRMS) UHPLC–HRMS was performed on a Thermo Scientific Ultimate 3000 RSLCnano system operated in capillary-flow mode coupled to a Thermo Scientific Q Exactive Mass spectrometer (Thermo Scientific, Fremont, CA). Individual standards at a concentration of 1 ␮g mL−1 were infused with a syringe at a flow rate of 2 ␮L min−1 through a T piece connected to an LC system with a mobile phase composed by 90:10 methanol:water at flow rate of 8 ␮L min−1 . Using the Q Exactive tune application, the precursor ion was selected in the quadrupole (tMS/MS mode) and product ions were found by increasing the Normalized Collision Energy (NCE).

135

The monoisotopic mass of each compounds was calculated using Qual browser of the XcaliburTM 2.2 Software (Thermo Fisher Scientific MA, USA). UHPLC separation was carried out on a C18 Acclaim PepMap RSLC (150 mm × 0.3 mm, 2.0 ␮m particle size) (Thermo Scientific, Fremont, CA) column thermostated at 35 ◦ C using a gradient solvent elution system composed by: [(A) 1 mM ammonium formate in 10:90 methanol:water (v/v)/(B) 1 mM ammonium formate in methanol]. Gradient elution was as follows: solvent B was initially set at 40%, then delivered by a linear gradient from 40% to 99% in 20 min. Solvent B was maintained at 99% for 30 min before column re-equilibration (10 min). The flow-rate was 10 ␮L min−1 and the injection volume was 1 ␮L. Q-Exactive instrument was equipped with a pneumatically assisted ESI interface with a stainless steel needle adapted for capillary flow. The position of the probe was placed at B. Nitrogen obtained from a nitrogen generator Zefiro (Clan Tecnologica, Seville, Spain) was employed for spray stabilization, as collision gas and as damping gas in the C-trap. The sheath gas (nitrogen, 99.999% purity), the auxiliary gas (nitrogen, 99.998% purity) and the sweep gas were delivered at flow rates of 8, 5 and 1 arbitrary units, respectively, for both positive and negative mode. Conditions of the interface were as follows: ESI spray voltage +3.0/−2.6 kV, capillary temperature, 320 ◦ C; S lens RF level, 50 V. The mass calibration of orbitrap was performed every three days to ensure a working mass accuracy lower than 2 ppm. Chromeleon 6.8 and XCalibur 2.2 softwares were used to control the instrument and for data processing.

2.4. Targeted and untargeted analysis for qualitative characterization of PC extracts Polycarbonate extracts were analyzed from a qualitative point of view in order to know the composition of plastic samples taken into account. To this aim, a targeted and an untargeted approach (in both positive and negative mode) were adopted. In both cases, data dependent Top N acquisition (Full MS/dd MS/MS) was exploited (intensity threshold 5 × 104 ). As reported by Kumar and Rúbies [41], a Full MS/dd MS/MS (Top N) experiment consists of a full scan event followed by MS/MS scan events of Top N precursors in the inclusion list or the N most abundant ions at that time. A Top N approach was optimized such that in a given time, when

Table 1 List of additives selected with molecular formula, detected ion and the two fragments selected for the identification with the collision induced dissociation energy. Compounds BPA BHT BHA Tinuvin 234 Tinuvin 326 Tinuvin 327 Tinuvin 328 Cyasorb UV 9 Cyasorb UV 12 Cyasorb UV 24 Cyasorb UV 5411 Irgafos 168 Advastab 800 Uvinul 400 Cyanox 2246 Chimassorb 81 Uvitex OB BADGE Irganox 1076 Irganox 1010 Irganox 1330 Irganox 1081

Molecular formula C15 H16 O2 C15 H24 O C11 H16 O2 C30 H29 N3 O C17 H18 ClN3 O C20 H24 ClN3 O C22 H29 N3 O C14 H12 O3 C15 H14 O5 C14 H12 O4 C20 H25 N3 O C42 H63 O3 P C30 H58 O4 S C13 H10 O3 C23 H32 O2 C21 H26 O3 C26 H26 N2 O2 S C21 H24 O4 C35 H62 O3 C73 H108 O12 C54 H78 O3 C22 H30 O2 S

Molecular ion

Products ions −1

227.1078 [M − H] 219.1754 [M − H]−1 179.1078 [M − H]−1 448.2383 [M + H]+1 316.1211 [M + H]+1 358.1681 [M + H]+1 352.2383 [M + H]+1 229.0859 [M + H]+1 275.0914 [M + H]+1 245.0808 [M + H]+1 324.2070 [M + H]+1 647.4588 [M + H]+1 532.4394 [M + NH4 ]+1 215.0703 [M + H]+1 339.2330 [M − H]−1 327.1955 [M + H]+1 431.1788 [M + H]+1 358.2013 [M + NH4 ]+1 548.5048 [M + NH4 ]+1 1194.8179 [M + NH4 ]+1 792.6289 [M + NH4 ]+1 357.1894 [M − H]−1

211.0763, 133.0651 (hcd 65) 203.1437, 163.1119 (hcd 70) 164.0834, 149.0599 (hcd 52) 370.1917, 119.0856 (hcd 30) 260.0583, 107.0494 (hcd 48) 302.1057, 246.0431 (hcd 45) 282.1603, 212.0820 (hcd 50) 151.0388, 105.0338 (hcd 50) 169.0492, 151.0388 (hcd 30) 151.0388, 121.0285 (hcd 35) 212.0820, 92.0501 (hcd 50) 347.1769, 291.1143 (hcd 40) 329.2144, 143.0162 (hcd 20) 137.0232, 105.0338 (hcd 45) 163.1119 (hcd 45) 215.0703, 137.0232 (hcd 40) 415.1470, 401.1316 (hcd 62) 229.1221, 191.1065 (hcd 20) 475.4143, 419.3515 (hcd 12) 729.2902, 563.2272 (hcd 20) 569.4344, 219.1745 (hcd 20) 194.0760, 163.1119 (hcd 35)

136

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

N abundant masses in the inclusion list are detected in the survey scan, a dd MS/MS is triggered to select the precursor ion in the inclusion list, fragmented in HCD cell, collected in the C trap and analyzed in the orbitrap. In Top N, N depends on the loop count and multiplex (N = loop count × multiplex). In this study, multiplex for product ion scan was not assessed, and hence N = 8 corresponds directly to loop count (multiplex was set as 1). In particular, for targeted approach, an inclusion list with the exact masses of each standard selected and their collisional dissociation energy was employed; in the case of the untargeted approach, the inclusion list option was disabled. In both cases, a resolving power of 70 K was used for Full MS, and 17.5 K for dd MS/MS events. Full scan data were acquired from 90 to 1200 m/z, the automatic gain control was set at 1 × 106 and the dynamic exclusion was 10 s. Data evaluation was performed by extracting the accurate mass traces (±10 ppm) concerning the protonated/deprotonated ion in full scan.

3. Results and discussion The characterization of PC intended to come in contact with food is of paramount importance since its wide diffusion. Before doing migration tests, a characterization of a plastic sample is, in our opinion, very important in order to find which molecules are present, and therefore may migrate to a food product. In the first part of the work, the most used additives were selected based on a literature overview. Standard solutions of each compound were injected, and collisional induced dissociation were optimized with the aim to create a high-resolution fragmentation spectra database. In Table 1 the standard selected, their molecular formula, parent ion recorded in the full scan, and most representative product ions are reported. From a chromatographic point of view, a capillary-UHPLC set up was employed in order to improve sensitivity; in fact, since electrospray ionization is a concentration-dependent technique, this configuration generates a significantly higher number of

24.71

100 50 2.12

0 100

25.55 28.35 33.30

6.88

25.89

50 16.80

0 100

27.39 26.38

33.72 37.25 39.50

44.60 49.43 51.42 53.86

50 27.56 31.27 33.81 38.91 40.63 44.51 47.19 50.53 53.47 27.09

0 100 50 3.63 5.68

0 100

11.38

16.19

20.43 23.69

28.05 30.92 34.73 38.45 42.14 45.56 48.89 52.13 54.57 58.84 30.45

50 31.27 35.32 37.40 40.64 44.72 47.47 50.65 53.55 32.62

0 100

59.08

50 31.82

14.19

0 100

33.57 37.12 39.83 43.64 46.90 50.18 37.07

50 0

2.14 0

24.77 27.14

6.84 8.34 5

10

15

20

25

33.78 36.08 30 35 Time (min)

38.12 42.27 40

47.81 49.68 51.29 45

50

55

60

Fig. 1. LC–ESI–HRMS chromatograms of the seven plastic additives found in polycarbonate extracts. From above Cyasorb UV5411, Tinuvin 234, Uvitex OB, Tinuvin 327, Advastab 800, Irganox 1076 and Irgafos 168.

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

ions that potentially enters the mass spectrometer. In this context, since the response in ESI–MS is strongly dependent on the mobile phase composition, methanol was used as organic phase instead of acetonitrile because of its lower boiling point, which favored the desolvation of the electrospray droplets and therefore increases the ionization efficiency. Ammonium formate was added to mobile phases to avoid sodium adducts, but no modifiers like ammonia, formic acid or acetic acid were adopted because samples were acquired in both positive and negative mode. 3.1. Extraction efficiency Sample pretreatment was performed by following a method reported in literature [20], based on polymer dissolution and its successive re-precipitation with some modifications as reported in the experimental section. Since no data about recovery in real polymer extracts after standards addition were previously performed, experiments on recovery evaluation were carried out. To this aim, three different standard compounds not contained in the analyzed samples (BPA-d16 , Tinuvin 326, and Uvitex OB) were selected on the basis of their different hydrophobicity, to be representative of the different classes of compounds possibly occurring. Preliminary experiments shown that the effect of the solvent to be employed in the last step of extract resuspension was determinant to assess the extraction yields. Therefore, two aliquots of the same PC sample were spiked with identical amounts of standards and then submitted to the same treatment, but using as final solvent acetone and methanol, respectively. Very good recoveries were obtained for BPA-d16 , 98 ± 6%, and 95 ± 8% by resuspending the sample in

137

methanol and in acetone, respectively. Regarding the other two analytes, Uvitex OB and Tinuvin 326, higher recovery values were obtained by employing acetone (84 ± 3%, 102 ± 7%) rather than methanol (51 ± 4%, 42 ± 3%). Therefore acetone was selected as the best solvent for all subsequent experiments. 3.2. Targeted analysis High resolution tandem mass spectrometry under data dependent acquisition control (Data-Dependent Acquisition, DDA) to identify compounds has become an interesting and valuable approach. DDA coupled to high resolution mass spectrometry (HRMS) is a common feature used for peptide/protein identification in complex mixtures. In the field of plastic additives the utility of this acquisition mode for characterization of food-contact plastics has not yet been evaluated and exploited. As well known, DDA is a product ion scan mode providing automated switching between MS and MS/MS and then returning to MS using data dependent criteria. Tandem mass spectrometry scanning is usually triggered when at least one ion exceeds a preset threshold in MS mode. Full MS–ddMSMS obviates the need to select any target precursor ion a priori, which is particularly useful in the analysis of samples that may contain a large variety of possible unknowns. The advantages of acquisition by high resolution Full MS–ddMSMS are numerous, since it allows to unequivocally identify a molecule at different levels. First, the extraction of parent ion from full scan allows to verify the match between the exact mass of a molecule and the accurate mass measured by the instrument and the corresponding isotopic pattern simulated by

Fig. 2. Panel A: representation of the isotopic pattern simulated by the software by setting the molecular formula of Tinuvin 234 (above) and that one found in the PC extract (below). Panel B: comparison between fragmentation spectrum obtained by injecting a standard solution of Tinuvin 234 (above) and that found in the real sample (below). Insert: molecular structure of Tinuvin 234.

138

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

the software. Second, the identification is also made through the comparison between the fragmentation spectrum of an unknown molecule found in the sample and that one obtained by injection of a standard solution. Additionally, a full scan experiment potentially enables the retrospective data mining in an untargeted test setting for unknown substances. In this study, a qualitative screening of PC samples was conducted in order to know the composition in terms of plastic additives, eventual NIAS, BPA-polycarbonate degradation products and organic colorants. In particular, both targeted and untargeted approaches were carried out. As polycarbonate extracts were very complex samples, an elution program employing a linear gradient of 60 min was necessary in order to guarantee the elution of all molecules occurring in the extract. As reported in the experimental section, targeted analysis was carried out by monitoring several known compounds commonly used as additives (see list of selected standards), employing datadependent acquisition with inclusion list. The obtained results lead to the identification in PC extracts of the following plastic additives: Tinuvin 234, Tinuvin 327, Cyasorb

UV5411, Uvitex OB, Irgafos 168, Irganox 1076 and Advastab 800, as shown in Fig. 1. In particular, Tinuvin 234, Tinuvin 327 and Cyasorb UV5411 belong to the group of UV absorbers having the function of absorb the most dangerous wavelengths, UV-A (320–400 nm) and UV-B (290–320 nm), dissipating them at higher wavelengths. Irgafos 168 is a processing stabilizer for polymers, Uvitex OB represents a whitening agent, while the others belong to the category of antioxidants which prevent the yellowing of plastic. It has to be underlined that all of them are approved for the use in food-contact plastics. In Fig. 2, an example of the criteria used for the molecules recognition is shown. Apart from the exact correspondence of retention times, in detail, panels A shows the comparison between the isotopic pattern simulated by the software in the case of Tinuvin 234 and that one found in the sample obtained by extracting the molecular ion from Full MS; panel B shows the comparison between the fragmentation spectra. In the insert the molecular structure is reported. For all identified compounds, the comparison between the isotopic ratios simulated by the software and those found in the sample shown a very good match (see Supplementary material).

Fig. 3. Panel A: colorant Solvent Yellow 184 identified in the PC orange sample. Panel B: colorant Solvent Yellow 232 identified in the PC yellow sample. Panel C: colorant Solvent Red 179 identified in the PC red sample. Panel D: colorant Solvent Red 135 found in the PC pink and orange sample. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

139

Similarly, the fragmentation spectra obtained in the PC extracts perfectly overlap those obtained by injecting the standard solutions (see Supplementary material).

Results about molecules identified by this approach are reported in the following sections.

3.3. Untargeted analysis

3.3.1. Colorants An important step of untargeted analysis was represented by the identification of organic colorants found in PC extracts and responsible for brilliant colors. An accurate analysis of the mass spectra associated to each peak of the full-scan chromatogram, allowed to reveal the presence of mass peaks possibly belonging to unknown compounds of interest. The attribution of the identity of each molecule has been achieved through three consecutive steps: first, the corresponding molecular formula was searched by software elaboration starting from the accurate mass recorded. Then, since more than one formula was often obtained, the comparison of all isotopic patterns provided by a simulator allowed to reach the unequivocal identity. Finally, a meticulous investigation of the molecular structures possibly associated to the found formula was performed with the support of available chemical databases. In Fig. 3, the details of identification of four molecules of colorants found in the extracts of orange (panel A: solvent yellow 184), yellow (panel B: solvent yellow 232), red (panel C: solvent red 179) and pink (panel D: solvent red 135) PC

This part of the work was properly finalized to the identification of unknown molecules possibly occurring in the extracts (BPApolycarbonate degradation products, organic colorants, NIAS). The results could be obtained thanks to a careful observation of each section of the full scan chromatograms, paying particular attention to the possible presence of characteristic isotopic patterns, as in the case of molecules containing chlorine. As reported in the experimental section, the untargeted analysis was performed through data dependent acquisition without inclusion list. In this case, the data-dependent files contained a list of precursor ions with their fragmentation spectra. Since the precursor ions were unknown molecules, the interpretation of each single fragmentation spectrum was necessary. Starting from an accurate mass present in a full scan or in a fragmentation spectrum, the software can generate a list of molecular formula (with a maximum mass deviation of ±2 ppm) suggesting the possible formula and giving also the possibility to verify the match of the isotopic ratios.

20.17

100 80

m/z 369.1000

60 40 20

21.41 22.87 25.99 30.42 33.10

12.06 14.92 19.64

0 100

21.53

80

m/z 271.0866

60 40 20 2.09

0 100

6.32 8.74 12.02 13.74 18.89 20.87

22.66 25.22 29.05 25.05

80

34.23

m/z 321.1022

60 40 25.72 26.79

20 0.78 3.72

0 100

17.01

20.92

28.46 31.25 3

27.48

80

m/z 406.9307

60 40 20 0

2.21 4.84 8.17 9.55 0

5

10

13.96 15

26.57 20

25

28.78 30.42 33. 30 35 Time (min)

Fig. 4. LC–ESI–HRMS chromatograms of the four organic colorants identified in the orange, yellow, pink and red PC samples, respectively. From above: Solvent Yellow 232, Solvent Yellow 184, Solvent Red 179, Solvent Red 135. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

140

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

samples, respectively, are presented. In Fig. 4, the corresponding chromatograms are reported. All the colorants identified are included in the list of the dyes commonly used in the production of food contact materials and belong to the category of Solvent colorants, that are, in fact, dyes soluble in organic solvents. These molecules are in general apolar or partially polar compounds insoluble in water, and are commercialized with the name solvent followed by their color. Red and yellow dyes are often azocompounds, while green and blue ones are in

generally anthraquinone-based compounds. These molecules are very common for the use in plastic coloration thanks to their high heat and light resistance and easy coloring. However, traceability of colorants is often lost upstream because plastic manufactures often buy plastic pellets already colored, without receiving any information about the identity of the dye employed as dyes. This aspect is particularly problematic considering that often these plastic pellets derived from foreign countries having different regulations than European Union. In this context,

Fig. 5. Panel A: fragmentation spectrum of molecular ion at m/z 481.2022 detected in negative mode that provides as product ions bisphenol A (m/z 227.1078) and its fragments (m/z 211.0763, m/z 133.0651 and m/z 93.0335). Panel B: fragmentation pathway of molecular ion at m/z 486.1913 detected in positive mode.

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

the identification of colorants in plastic to come in contact with food is of paramount importance in order to verify in future migration studies their behavior in contact with simulants and food.

141

As it can be seen in Table 2, all signals found in the mass spectra represent pieces of the polycarbonate structure, as often the same units are repeated. These molecules were probably generated during sample preparation procedure, since during that step polycarbonate was completely dissolved. In Fig. 6, the chromatograms of the most representative degradation products detected in positive and in negative mode are visualized. Literature studies about polycarbonate ageing [46,47] induced by light, identified some of these molecules representing PCdegradation products. Our results reinforce the possibility for such molecules to be a useful tool for studies on photo-Fries reaction and polycarbonate yellowing. For example, a correlation between yellowness effect on PC and content of BPA trimer has been reported [48], as well as other PC by-products and the pro-degradation effect caused from phenolic end groups. Moreover, in perspective, it will be interesting to assess in future studies whether the molecules found could be present in simulants after migration tests also in the form of not reacted-molecules during polycarbonate synthesis.

3.3.2. BPA-polycarbonate dissolution products The accurate analysis of the mass fragmentation spectra related to most intense signals recorded in the full-scan chromatogram (acquired in both positive and negative mode), revealed the presence of mass peaks possibly connected to the structure of bisphenol A-polycarbonate chain. Furthermore, the presence of signals deriving from the typical fragmentation of PC chain allowed the recognition of the presence of several oligomers derived from the chain breakage in the examined extracts. In particular, by following the approach described in the previous section, and by reconstructing the mass fragmentation spectra, the identification of the oligomers was carried out. Fig. 5 reports two examples of molecules found in an extract, acquired in negative (panel A) and in positive mode (panel B). In both of them, some characteristic ions belonging to the fragmentation pattern of bisphenol A, and of its polycarbonate chain, are present.

19.16

100

m/z 366.1700

50 2.04 3.77

0 100

8.20

18.33

12.06

19.95 21.74 27.81 20.29 m/z 424.1755

33.58 35.

50 1.47 4.76 8.22 9.56

0 100

21.32 23.26 27.48 32.33 34

16.57 18.64

21.80

m/z 486.1911

50 2.09

0 100

5.88 8.87 10.54 13.00

22.64 24.49 19.70 20.95 m/z 481.2020

31.91 34.3

50 17.54

0 100

19.22

22.01 24.98 27.95 31.26 22.65 m/z 735.2963

3

50 2.03

0 100

6.83 10.72 14.41 17.98 19.62

23.69 23.76

30.29 32.92 m/z 989.3906

50 0

2.03 0

17.10 18.57 21.21 5

10

15

20

24.39 27.48 25

33.94

30 35 Time (min)

Fig. 6. LC–ESI–HRMS chromatograms of the most representative potential BPA-polycarbonate degradation products found in all the samples analyzed. Molecular formula and molecular structures are reported in Table 2.

142

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

Table 2 List of potential BPA-polycarbonate degradation products detected in PC samples. Chemical structure

4. Conclusions and perspectives This work presents an analytical method based on an innovative UHPLC-HRMS system allowing the detection of several substances occurring in PC objects destined to come in contact with food. A

Molecular formula

Ion detected

C15 H16 O2 bisphenol A

227.1078 [M − H]−1

C17 H18 O4

304.1543 [M + NH4 ]+1

C21 H26 O3

344.2221 [M + NH4 ]+1

C22 H20 O4

366.1700 [M + NH4 ]+1

C19 H20 O6

362.1601 [M + NH4 ]+1

C29 H24 O6

486.1911 [M + NH4 ]+1

C24 H22 O6

424.1755 [M + NH4 ]+1

C33 O6 H32

542.2537 [M + NH4 ]+1

C38 O7 H34

620.2642 [M + NH4 ]+1

C40 O9 H36

678.2698 [M + NH4 ]+1

C45 O9 H38

740.2854 [M + NH4 ]+1

C48 O9 H42

780.3167 [M + NH4 ]+1

C31 O5 H30

481.2020 [M − H]−1

C32 O5 H32

495.2177 [M − H]−1

C33 O7 H32

539.2075 [M − H]−1

C47 O8 H44

735.2963 [M − H]−1

C63 O11 H58

989.3906 [M − H]−1

targeted and an untargeted analytical approaches have been followed. The first one was conducted with the aim of searching for additives commonly employed in PC production to protect the material from degradation, and was revealed to be very efficient. The second approach was aimed at identify NIAS such as unknown

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

molecules possibly deriving from PC degradation, and allowed to recognize the presence of several compounds never found before. Furthermore, for the first time, the identification of the molecules used as coloring agents in PC objects has been performed with success, demonstrating that this technique has many advantages over the traditional systems. The combination between the enhanced sensitivity offered by the UHPLC system and the high selectivity provided by the HR–MS, can be very promising, and represents a convenient system to be further developed in this and in many other fields of application. The shown approach allows to set up further studies finalized to verify the compliance of food-contact plastics; in fact, beyond the identification of the molecules occurring in a material, migration studies can be directly focused in the search of target compounds. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.chroma.2014.10.104. References [1] European Commission, Regulation (EC) No. 1935/2004 of the European Parliament and of the Council of 27 October 2004 on materials and articles intended to come into contact with food and repealing Directives 80/590/EEC and 89/109/EEC, Off. J. Eur. Union L. 338 (2004) 4–17. [2] European Commission, Regulation (EC) No. 2023/2006 of 22 December 2006 on good manufacturing practice for materials and articles intended to come into contact with food, Off. J. Eur. Union L. 384 (2006) 75–78. [3] European Commission, Regulation (EC) No. 10-2011 of 14 January 2011 for plastics intended to come into contact with food, Off. J. Eur. Union L. 12 (2011) 1–89. [4] A. Sanches Silva, R. Sendón García, I. Cooperb, R. Franz, P. Paseiro Losada, Compilation of analytical methods and guidelines for the determination of selected model migrants from plastic packaging, Trends Food Sci. Technol. 17 (2006) 535–546. [5] A.V. Krishnan, P. Stathis, S.F. Permuth, L. Tokes, D. Feldman, Bisphenol-A: an estrogenic substance is released from polycarbonate flasks during autoclaving, Endocrinology 132 (1993) 2279–2286. [6] R.T. Zoeller, R. Bansal, C. Parris, Bisphenol-A, an environmental contaminant that acts as a thyroid hormone receptor antagonist in vitro, increases serum thyroxine, and alters RC3/neurogranin expression in the developing rat brain, Endocrinology 146 (2005) 607–612. [7] Y.B. Wetherill, C.E. Petre, K.R. Monk, A. Puga, K.E. Knudsen, The xenoestrogen bisphenol A induces inappropriate androgen receptor activation and mitogenesis in prostatic adenocarcinoma cells, Mol. Cancer Ther. 7 (2002) 515–524. [8] B.T. Akingbemi, C.M. Sottas, A.I. Koulova, G.R. Klinefelter, M.P. Hardy, Inhibition of testicular steroidogenesis by the xenoestrogen bisphenol A is associated with reduced pituitary luteinizing hormone secretion and decreased steroidogenic enzyme gene expression in rat leydig cells, Endocrinology 145 (2004) 592–603. [9] European Commission, Commission Directive 2011/8/EU of 28 January 2011 amending Directive 2002/72/EC as regards the restriction of use of bisphenol A in plastic infant feeding bottles, Off. J. Eur. Union L. 26 (2011) 11–14. ˜ [10] O. Núnez, H. Gallart-Ayala, C.P.B. Martins, P. Lucci, New trends in fast liquid chromatography for food and environmental analysis, J. Chromatogr. A 1228 (2012) 298–323. ˜ [11] H. Gallart-Ayala, O. Núnez, P. Lucci, Recent advances in LC–MS analysis of foodpackaging contaminants, Trends Anal. Chem. 42 (2013) 99–124. [12] J.G. Teeguarden, S. Hanson-Drury, A systematic review of bisphenol A “Low Dose” studies in the context of human exposure: a case for establishing standards for reporting “Low-Dose” effects of chemicals, Food Chem. Toxicol. 62 (2013) 935–948. [13] M.L. Oca, M.C. Ortiz, A. Herrero, L.A. Sarabia, Optimization of a GC/MS procedure that uses parallel factor analysis for the determination of bisphenols and their diglycidyl ethers after migration from polycarbonate tableware, Talanta 106 (2013) 266–280. ˜ J. Salafranca, Determination of poten[14] C. Nerín, C. Fernández, C. Domeno, tial migrants in polycarbonate containers used for microwave ovens by high-performance liquid chromatography with ultraviolet and fluorescence detection, J. Agric. Food Chem. 51 (2003) 5647–5653. [15] R. Braunrath, D. Podlipna, S. Padlesak, M. Cichna-Markl, Determination of bisphenol A in canned foods by immunoaffinity chromatography, HPLC, and fluorescence detection, J. Agric. Food Chem. 53 (2005) 8911–8917. [16] S¸. Sungur, M. Köro˘glu, A. Özkan, Determinatıon of bisphenol a migrating from canned food and beverages in markets, Food Chem. 142 (2014) 87–91. [17] C. Simoneau, S. Valzacchi, V. Morkunas, L. Van den Eede, Comparison of migration from polyethersulphone and polycarbonate baby bottles, Food Addit. Contam. 28 (2011) 1763–1768.

143

[18] A.G. Asimakopoulos, N.S. Thomaidis, M.A. Koupparis, Recent trends in biomonitoring of bisphenol A, 4-t-octylphenol, and 4-nonylphenol, Toxicol. Lett. 210 (2012) 141–154. [19] E.J. Hoekstra, C. Simoneau, Release of bisphenol A from polycarbonate—a review, Crit. Rev. Food Sci. Nutr. 53 (2013) 386–402. [20] J. Alin, M. Hakkarainen, Migration from polycarbonate packaging to food simulants during microwave heating, Polym. Degrad. Stab. 97 (2012) 1387–1395. [21] N.C. Maragou, A. Makri, E.N. Lampi, N.S. Thomaidis, M.A. Koupparis, Migration of bisphenol A from polycarbonate baby bottles under real use conditions, Food Addit. Contam. 25 (2008) 373–383. [22] E. Fasano, F. Bono-Blay, T. Cirillo, P. Montuori, S. Lacorte, Migration of phthalates, alkylphenols, bisphenol A and di(2-ethylhexyl)adipate from food packaging, Food Control 27 (2012) 132–138. [23] N.C. Maragou, E.N. Lampi, N.S. Thomaidis, M.A. Koupparis, Determination of bisphenol A in milk by solid phase extraction and liquid chromatography–mass spectrometry, J. Chromatogr. A 1129 (2006) 165–173. [24] A. Ballesteros-Gómez, D. Soledad Rubio, Pérez-Bendito, Analytical methods for the determination of bisphenol A in food, J. Chromatogr. A 1216 (2009) 449–469. [25] S. Sungur, M. Koro˘glu, A. Özkan, Determinatıon of bisphenol a migrating from canned food and beverages in markets, Food Chem. 142 (2014) 87–91. [26] A. Alabi, N. Caballero-Casero, S. Rubio, Quick and simple sample treatment for multiresidue analysis of bisphenols, bisphenol diglycidyl ethers and their derivatives in canned food prior to liquid chromatography and fluorescence detection, J. Chromatogr. A 1336 (2014) 23–33. [27] C.W. Klampfl, Mass spectrometry as a useful tool for the analysis of stabilizers in polymer materials, Trends Anal. Chem. 50 (2013) 53–64. [28] D. Louden, A. Handley, E. Lenz, I. Sinclair, S. Taylor, I.D. Wilson, Reversedphase HPLC of polymer additives with multiple on-line spectroscopic analysis (UV, IR, (1)H NMR and MS), Anal. Bioanal. Chem. 373 (2002) 508–515. ˜ G. Fernández-Martínez, M.V. González[29] R. Noguerol-Cal, J.M. López-Vilarino, Rodríguez, L.F. Barral-Losada, Liquid chromatographic methods to analyze hindered amines light stabilizers (HALS) levels to improve safety in polyolefines, J. Sep. Sci. 33 (2010) 2698–2706. [30] R.E. Duderstadt, S.M. Fischer, Effect of organic mobile phase composition on signal responses for selected polyalkene additive compounds by liquid chromatography–mass spectrometry, J. Chromatogr. A 1193 (2008) 70–78. [31] T. Andersen, I.L. Skuland, A. Holm, R. Trones, T. Greibrokk, Temperature programmed packed capillary liquid chromatography coupled to evaporative light-scattering detection and electrospray ionization time-of-flight mass spectrometry for characterization of high-molecular-mass hindered amine light stabilizers, J. Chromatogr. A 1029 (2004) 49–56. [32] M. Himmelsbach, W. Buchberger, E. Reingruber, Determination of polymer additives by liquid chromatography coupled with mass spectrometry. A comparison of atmospheric pressure photoionization (APPI), atmospheric pressure chemical ionization (APCI), and electrospray ionization (ESI), Polym. Degrad. Stabil. 94 (2009) 1213–1219. [33] C. Block, L. Wynants, M. Kelchtermans, R. De Boer, F. Compernolle, Identification of polymer additives by liquid chromatography–mass spectrometry, Polym. Degrad. Stabil. 91 (2006) 3163–3173. [34] E. Reingruber, M. Himmelsbach, C. Sauer, W. Buchberger, Identification of degradation products of antioxidants in polyolefins by liquid chromatography combined with atmospheric pressure photoionisation mass spectrometry, Polym. Degrad. Stabil. 95 (2010) 740–745. [35] S.S. Choi, J.H. Jang, Analysis of UV absorbers and stabilizers in polypropylene by liquid chromatography/atmospheric pressure chemical ionization–mass spectrometry, Polym. Test. 30 (2011) 673–677. [36] M. Stiftinger, W. Buchberger, C.W. Klampfl, Miniaturised method for the quantitation of stabilisers in microtome cuts of polymer materials by HPLC with UV, MS or MS2 detection, Anal. Bioanal. Chem. 405 (2013) 3177–3184. [37] I. Park, K.H. Yoon, H. Namgoong, Determination of Tinuvin 292 in acrylic resins by reversed-phase high performance liquid chromatography electrospray ionization-ion trap mass spectrometry, Chem. Lett. 41 (2012) 301–303. [38] A. Michalski, E. Damoc, J. Hauschild, O. Lange, A. Wieghaus, A. Makarov, N. Nagaraj, J. Cox, M. Mann, S. Horning, Mass spectrometry-based proteomics using Q Exactive, a high performance benchtop quadrupole orbitrap mass spectrometer, Mol. Cell. Proteomics 10 (2011) 011015–11021. [39] S. Gallien, E. Duriez, K. Demeure, B. Domon, Selectivity of LC–MS/MS analysis: implication for proteomics experiments, J. Proteomics 81 (2013) 148–158. [40] M. Concheiro, D. Lee, E. Lendoiro, M.A. Huestis, Simultaneous quantification of 9 -tetrahydrocannabinol, 11-nor-9-carboxy tetrahydrocannabinol, cannabidiol and cannabinol in oral fluid by microflow liquid chromatography–high resolution mass spectrometry, J. Chromatogr. A 1297 (2013) 123–130. [41] P. Kumar, A. Rúbies, F. Centrich, M. Granados, N. Cortés-Francisco, J. Caixach, R. Companyó, Targeted analysis with benchtop quadrupole–orbitrap hybrid mass spectrometer: application to determination of synthetic hormones in animal urine, Anal. Chim. Acta 780 (2013) 65–73. [42] A. Thomas, K. Walpurgis, O. Krug, W. Schänzer, M. Thevis, Determination of prohibited, small peptides in urine for sports drug testing by means of nano-liquid chromatography/benchtop quadrupole orbitrap tandem–mass spectrometry, J. Chromatogr. A 1259 (2012) 251–257. [43] W. Jia, X. Chu, Y. Ling, J. Huang, J. Chang, High-throughput screening of pesticide and veterinary drug residues in baby food by liquid chromatography coupled to quadrupole orbitrap mass spectrometry, J. Chromatogr. A 1347 (2014) 122–128.

144

C. Bignardi et al. / J. Chromatogr. A 1372 (2014) 133–144

[44] F. Shi, C. Guo, L. Gong, J. Li, P. Dong, J. Zhang, P. Cui, S. Jiang, Y. Zhao, S. Zeng, Application of a high resolution benchtop quadrupole–orbitrap mass spectrometry for the rapid screening, confirmation and quantification of illegal adulterated phosphodiesterase-5 inhibitors in herbal medicines and dietary supplements, J. Chromatogr. A 1344 (2014) 91–98. ˜ [45] J.I. Cacho, N. Campillo, P. Vinas, M. Hernández-Córdoba, Determination of alkylphenols and phthalate esters in vegetables and migration studies from their packages by means of stir bar sorptive extraction coupled to gas chromatography–mass spectrometry, J. Chromatogr. A 1241 (2012) 21–27.

[46] B.N. Jang, C.A. Wilkie, A TGA/FTIR and mass spectral study on the thermal degradation of bisphenol A polycarbonate, Polym. Degrad. Stabil. 86 (2004) 419–430. [47] S. Collin, P.O. Bussière, S. Thérias, J.M. Lambert, J. Perdereau, J.L. Gardette, Physicochemical and mechanical impacts of photo-ageing on bisphenol a polycarbonate, Polym. Degrad. Stabil. 97 (2012) 2284–2293. [48] J.E. Pickett, Influence of photo-Fries reaction products on the photodegradation of bisphenol-A polycarbonate, Polym. Degrad. Stabil. 96 (2011) 2253–2265.