Green Sorptive Sampling Coupled With UHPLC–IMS–MS and GC×GC–TOFMS for Characterization of Migrating Substances From Plastic Food Contact Materials

ABSTRACT

Comprehensive analysis of migrating compounds from plastic food contact materials (FCMs) is analytically demanding due to the structural diversity and broad physicochemical properties of intentionally and non‐intentionally added substances (IAS and NIAS). This necessitates the development of advanced, multidimensional separation techniques coupled with high‐resolution detection for trace‐level characterization. Five million tones of chemical additives are incorporated into plastic FCMs annually, which is concerning considering the health risks associated with human exposure to migrating chemicals. This study provides a simplified green sorptive sampling loop for detecting and quantifying IAS and NIAS that migrate from FCMs during everyday use. A polydimethylsiloxane (PDMS) sorptive sampler loop was used as the sampling device, allowing migrated compounds to be enriched directly from the sample matrix. After extraction, the same sampler format was compatible with two analytical workflows: thermal desorption directly into the GC inlet for GC×GC–TOFMS analysis, or liquid back‐extraction into solvent for subsequent UHPLC‐IMS‐MS analysis. Complementary chromatography and ion mobility high‐resolution mass spectrometry (UHPLC–IMS–MS and GC×GC–TOFMS) enabled the detection and quantification of a range of chemical compounds at trace levels, including phthalates, UV absorbers, stabilizers, surfactants, and hydrocarbons. Over 30 compounds were detected as migrating from FCMs using GC–MS, and 26 compounds were identified using LC–MS. Unknown target identification using LC–MS was aided by the use of collision cross section (CCS) values, thus strengthening putative identification. Limits of detection of the eight target analytes ranged from 0.023 to 97 ng L⁻¹ for LC–MS and IAS were detected at concentrations ranging from 0.0065 to ∼25 µg L^− 1. Our study exemplifies the wide range of chemical compounds that migrate from FCMs during everyday FCM use. The green sampling technique presented here allowed the quantification of daily exposure to these chemicals and has potential application to human health and environmental risk assessments.

A green PDMS sorptive sampling loop coupled with complementary liquid and gas chromatography–mass spectrometry enabled detection of migrating chemicals from plastic food contact materials under everyday‐use scenarios. The approach revealed diverse additives and non‐intentionally added substances, supporting improved exposure assessment of plastic‐derived migrants.

1. Introduction

Plastic materials are widely used in food contact applications because of their low cost, durability, low mass, and functional versatility. However, plastics are not chemically inert. They contain intentionally added substances (IAS), such as plasticizers, stabilizers, antioxidants, colorants, and processing aids, as well as non‐intentionally added substances (NIAS), including impurities, degradation products, reaction by‐products, and compounds formed during manufacture, use, recycling, or environmental exposure. Such substances may originate from the original use of the material, the plastic formulation, labeling, transportation, misuse, additives, and/or the polymer matrix itself, as well as from environmental exposure or processing into post‐consumer resins (PCRs) [1].

In 2019, 368 million metric tons of newly made plastic were produced, with single‐use plastic items constituting half of the total production [2, 3]. Single‐use items are commonly manufactured from six main types of plastics, namely polyethylene terephthalate (PET) found in plastic bottles; high‐density polyethylene (HDPE) found in shampoo bottles, milk bottles, and ice‐cream containers; low‐density polyethylene (LDPE) found in shopping bags and bread bags; polystyrene (PS) found in food takeaway packaging; and expanded polystyrene (EPS) found in hot drink cups. The sixth and most abundant single‐use plastic type is polypropylene (PP), which is found in food packaging containers, drinking straws, and bottle caps [2, 3]. Chemical additives such as plasticizers are added to food contact materials (FCM) to increase the plastic service life, such as by improving flexibility [4, 5]. These additives can compose between 10% and 70% w/w of a plastic item, depending on the item's flexibility requirements. Plasticizers are not covalently bonded to the plastic polymer matrix but are closely associated with it, thereby facilitating the migration of plasticizers to the plastic or microplastic surface. Subsequent desorption from the plastic surface and absorption onto the external medium results in plasticizers being released over the plastic's lifetime [4]. Plasticizers can be classified as both extractables and leachables (ELs). Leachables are defined as compounds that migrate out of plastic under ambient conditions as a result of direct contact between the plastic and another medium, such as food, drink, soil, or water. Conversely, extractables are defined as compounds that require input energy, such as elevated temperatures, strong solvents, or exposure to ultraviolet (UV) radiation, to instigate their liberation from plastic [6].

IAS and NIAS occur at trace concentrations, rendering their detection and regulation a challenge in food, water, or the environment. The World Health Organization confirms that these trace levels are sufficient to compromise human health [2]. Consumer practices such as heating, microwaving, prolonged storage, or exposure to sunlight may further increase the release of plastic additives and degradation products [7, 8]. Migration tests are therefore essential for the detection and quantification of compounds released from FCMs, especially when the aim is to assess consumer exposure to known and emerging contaminants. However, these tests are challenging because many additives, degradation products, and contaminants are unknown, and because oligomers and other polymer‐derived compounds may further increase the chemical complexity of the migrated fraction [1, 6].

The challenge of detecting potential migrants from plastic food packaging at trace levels in complex matrices has been addressed to varying degrees by several authors through the use of advanced analytical instrumentation, such as gas and liquid chromatography, ion mobility spectrometry (IMS), and high‐resolution spectroscopy [5, 7, 8, 9, 10]. Recent developments in collision cross section (CCS) databases, such as METLIN‐CCS, have expanded the analytical chemist's toolkit for compound identification [5, 11, 12]. The use of high‐resolution mass spectrometry (HRMS), combined with ion mobility‐derived CCS values, increases confidence in the identification of unknown or tentatively assigned compounds during non‐target screening. However, the main analytical limitation is not only the instrumental platform used, but also the sample preparation method, which determines which compounds are extracted, concentrated, transferred, and ultimately detected.

Solid phase extraction (SPE) and liquid–liquid extractions (LLE) are often used [7, 8]. These methods can provide enrichment of low‐concentration compounds, but they also have important disadvantages. SPE may introduce high blank values due to contamination from sorbents, cartridges, frits, and plastic components used during extraction. It may also show variation between cartridge batches and reduced recovery for compounds that do not match the selectivity of the sorbent. LLE generally requires larger volumes of organic solvents, is less aligned with green analytical chemistry principles, and may be inefficient for broad non‐target screening when analytes differ strongly in polarity. In addition, both SPE and LLE may require solvent evaporation or reconstitution steps, which can lead to analyte loss, contamination, and reduced method reproducibility. Solid‐phase microextraction (SPME) has been explored as a greener alternative to conventional solvent‐based extraction methods, as it combines sampling, extraction, and preconcentration in a single step while reducing solvent consumption [13, 14, 15, 16, 17]. SPME–GC–MS methods have shown good analytical performance for the determination of trace‐level plastic additives in water, including phthalate esters, adipates, and bisphenol analogues, with reported detection limits often in the low ng L⁻¹ to µg L⁻¹ range and strong linearity during method validation [14, 15, 16, 17]. These methods have also been applied to different water matrices, including bottled water, drinking water, lake water, seawater, wastewater, and environmental waters, and have been used to assess the leaching of phthalates from plastic containers and packaging materials [14, 15, 16, 17]. More recent developments, such as derivatization‐SPME, direct‐immersion SPME, and arrow‐SPME coupled to GC–MS or GC–MS/MS, have further improved the sensitivity and scope of SPME‐based analysis for multi‐class plastic additives [13, 17, 18]. In addition to GC‐based workflows, SPME has also been coupled to liquid chromatography‐mass spectrometry. For example, on‐line capillary SPME‐LC‐ESI‐IT‐MS has been used for the sensitive determination of phthalate esters in plastic bottled water [13]. These studies show that SPME can be adapted to both GC‐ and LC‐based platforms, depending on the volatility, polarity, and analytical objective of the target compounds.

Despite these advantages, conventional SPME fibers are most often applied to volatile and semi‐volatile analytes and have small sorptive phase volumes, which can limit sensitivity and extraction capacity for broad non‐target screening. Fiber fragility, limited coating volume, and restricted extraction surface area may also reduce robustness when complex migrant mixtures must be investigated. Therefore, although SPME provides an important green analytical approach, further development of larger‐volume sorptive extraction devices is useful for improving extraction capacity, reducing background contamination, and expanding compatibility with complementary chromatographic workflows.

To address these limitations, the present study applied an in‐house‐developed polydimethylsiloxane (PDMS) sorptive sampler loop for the investigation of chemical migration from single‐use FCMs under realistic consumer‐use scenarios. Compared with SPE and LLE, the PDMS loop reduces solvent consumption, avoids cartridge‐related background contamination, simplifies sample handling, and provides a larger sorptive phase volume than conventional SPME fibers. The sampler can also be used in a complementary workflow: either thermally desorbed directly in the inlet liner of a gas chromatograph for analysis with comprehensive two‐dimensional gas chromatography coupled to time‐of‐flight mass spectrometry (GC×GC–TOFMS), or liquid back‐extracted directly into an LC vial for ultra‐high‐performance liquid chromatography coupled to ion mobility high‐resolution mass spectrometry (UHPLC–IMS‐HRMS). This allows volatile, semi‐volatile, and less volatile migrants to be investigated, thereby expanding the chemical space covered by the method.

In this study, chemical migration from common single‐use FCMs was investigated under scenarios that reflect realistic consumer behavior, such as purchasing bread and leaving it in a hot car, or microwaving takeaway food in PP or PS packaging. The combined use of GC×GC–TOFMS and UHPLC–IMS–HRMS enabled both target analysis of known migrants and non‐target screening of emerging or unexpected compounds. The inclusion of ion mobility and CCS information further improved the confidence of compound identification in complex migrant profiles. The main aim of this work was therefore to provide a greener, low‐background, and more comprehensive analytical approach for assessing consumer exposure to known and unknown chemicals migrating from everyday plastic FCMs.

2. Materials and Methods

2.1. Chemicals and Reagents

Methanol (MeOH), acetonitrile (ACN), and water ROMIL‐UpS ultra‐high purity solvents were purchased from Microsep (Johannesburg, South Africa). Formic acid (99+% purity) was purchased from Thermo Scientific. An 18‐constituent certified 1.5 mL ELs screening standard (0.1–1 µg mL⁻¹ in MeOH) was purchased from Waters Incorporated (Milford, Massachusetts, USA). The mixture contained Irganox 1010, Irganox 1076, Ethanox 330/Irganox 1330, Tinuvin 327, Tinuvin P, and Tinuvin 360 at 1 µg mL⁻¹ ± 1%, while Irganox 245, Irganox 1098, Cyasorb 2908, Uvinul 3030, propylparaben, diethyl phthalate, diphenyl phthalate, tris(p‐cresyl) phosphate, 2‐hydroxy‐4‐octyloxy benzophenone, Uvitex OB, dibutyl sebacate, and methylparaben were present at 0.1 µg mL⁻¹ ± 1%.

2.2. PDMS Sorptive Sampler

A PDMS sorptive sampler loop (Figure S1) was manufactured in‐house, presenting a mass of 0.03 ± 0.002 g and a 26 µL sorption volume. The loop was fashioned from 10.5 cm (0.03 g) of silicone elastomer medical‐grade tubing (0.64 mm OD x 0.3 mm ID, Sil‐Tec, Technical Products, Georgia, USA). The ends were connected by 1 cm of uncoated silica capillary column (250 µm ID) (SGE Analytical Science, Separation Scientific (Pty) Ltd, Roodepoort, South Africa). The continuous loop arrangement hinders water uptake by the PDMS tubing, thereby promoting sensitivity [19]. The PDMS loops were cleaned before sampling by sonicating in a 1:1 MeOH:Acetone (v/v) mixture three times for five minutes each. Subsequently, the loops were conditioned in a 17.8 cm long glass desorption tube (4 mm ID, 6 mm OD) obtained from Gerstel (Chemetrix, Midrand, South Africa) at 250°C overnight, using a hydrogen gas flow rate of 100 mL min⁻¹. Thereafter, the PDMS loops were again sonicated three times for 5 min each in ACN. After cleaning and conditioning, the PDMS loops were wrapped in aluminum foil and stored in cleaned glass bottles until use. For reused loops, the cleaning and conditioning procedure was repeated after each analysis and before subsequent sampling. Loop blanks were analyzed prior to each sample analysis to verify loop cleanliness and to monitor possible background contamination from the sampler.

2.3. FCM Samples and Everyday Scenarios

All single‐use plastic FCMs analyzed, as listed in Table 1, were purchased and transported to the laboratory in a material bag to avoid plastic contamination from retail plastic bags. All Schott bottles, measuring cylinders, funnels, and stainless‐steel forceps used were pre‐rinsed with MeOH and acetone to remove any plasticizers from their surface. The laboratory counter was wiped with a clean wet cloth and lined with a layer of paper towels.

TABLE 1.

Single‐use food contact materials (FCMs) analyzed in this study with their respective constituent plastic type.

Single‐use FCM	Plastic composition a
Samples analyzed using UHPLC–IMS–HRMS
Drinking straw	Polypropylene (PP)
Water bottle	Polyethylene terephthalate (PET)
Clear beverage cup	Polyethylene terephthalate (PET)
Polystyrene hot drink cup	Expanded polystyrene (EPS)
Samples analyzed using GC×GC–TOFMS
Bread bag	Low‐density polyethylene (LDPE)
Clear food packaging container	Polypropylene (PP)
Polystyrene take‐away container	Polystyrene (PS)

^a Plastic composition listed is as indicated on the label or watermark of the purchased FCM.

2.3.1. UHPLC–IMS–HRMS

Bottled water from three commercial 500 mL PET water bottles was sampled as follows: (1) after purchase to determine plasticizers that migrated out of the bottle and into the water during its shelf‐life, (2) after placing bottled water in the trunk of a car for 22 h, and (3) placed outdoors for a 22 h duration of UV/Vis radiation exposure. A stainless‐steel kettle was used to boil the ultra‐pure water to 100°C, which was subsequently added to two 250 mL EPS cups (WestPack, South Africa), filling the cups to the brim. One cup was left to stand at room temperature for 1 h, whereas the other was left to stand at room temperature for 2 h. One clear 350 mL PET cup (WestPack, South Africa) was filled to the brim with room temperature ultra‐pure water and left to stand at room temperature for 1 h. A single 1.3 g PP drinking straw was cut in half and both halves were completely submerged in room temperature ultra‐pure water inside a stainless‐steel cup and left to stand at room temperature for 2 h. The water sample was either shaken or stirred, using a stainless‐steel stirrer, prior to transferring 100 mL of water from each FCM to separate Schott bottles. These samples (n = 1) were subsequently extracted and analyzed by UHPLC–IMS–HRMS (refer to Sections 2.5.1 and 2.6.1).

2.4. Method Validation

2.4.1. UHPLC–IMS–HRMS

The matrix‐matched method of calibration was used for quantification. Six glass Schott bottles each containing 100 mL ultra‐pure MS grade water, were each spiked with one of the following volumes of EL standard mixture: 10 µL (0.01–0.1 µg L⁻¹), 25 µL (0.025–0.25 µg L⁻¹), 100 µL (0.1–1 µg L⁻¹), 200 µL (0.2–2 µg L⁻¹), 300 µL (0.3–3 µg L⁻¹), and 500 µL (0.5‐5 µg L⁻¹). Two milliliters of MeOH were added to prevent irreversible adhesion of plasticizers to the Schott bottle's inner glass surface. The resulting solution was left to stand overnight to equilibrate prior to extraction (refer to Section 2.5.1). Three Schott bottles each, containing 100 mL deionized water, were spiked with 200 µL of the EL standard, giving a concentration range of 0.2–2 µg L⁻¹ for the determination of the precision (% relative standard deviation [%RSD]) and accuracy (% recovery) for each of the eight targets. Limits of detection (LODs) and quantification (LOQs) were determined as the analyte concentrations corresponding to signal‐to‐noise ratios (S/N) of 3 and 10, respectively (Table 2). Method blanks (unspiked) were performed using the same conditions. To compensate for the ubiquity of plasticizers and phthalates, method blanks were repeated throughout the study period (n = 3; between days). For quantified LC–MS target analytes, the blank response from the corresponding analytical day was included in the matrix‐ and method‐matched calibration curve and was therefore compensated for during quantification. For LC–MS non‐target compounds annotated using the Waters ELs library, the blank peak area from the method blank analyzed on the same day as the sample was subtracted once from the corresponding sample peak area. After blank subtraction, zero or negative values were not reported. The variability between repeated method blanks was expressed as %RSD and is provided in Table S2.

TABLE 2.

Matrix‐matched UHPLC–IMS–HRMS calibration parameters, LODs, and LOQs for target analytes.

				Linearity
Compound	Quantification ion (m/z)	Adduct	TWCCSN2 (Å2) a	Range (µg L−1)	Number of points	Regression equation b (y = m x + c)	Slope %RSE c	R 2	LOD d (µg L−1)	LOQ e (µg L−1)
Dibutyl sebacate	337.2347	[M+Na]+	192.8	0.025–0.50	5	y = (228 ± 9.6)x + (25.2 ± 2.7)	4.2	0.995	0.0020	0.0066
Irganox 245	609.3426	[M+Na]+	238.3	0.010–0.50	5	y = (1190 ± 39)x + (23.1 ± 9.9)	3.3	0.996	0.0054	0.018
Irganox 1076	553.4605	[M+Na]+	246.6	0.10–5.0	6	y = (44.3 ± 1.6)x + (330 ± 4.0)	3.6	0.995	0.00039	0.0013
Irganox 1098	637.4945	[M+H]+	268.7	0.010–0.50	6	y = (2060 ± 69)x + (45.8 ± 17)	3.3	0.996	0.00063	0.0021
TCP	369.1250	[M+H]+	181.7	0.010–0.50	6	y = (6710 ± 540)x + (556 ± 130)	7.9	0.995	0.00063	0.0021
Tinuvin 360	659.4076	[M+H]+	276.7	0.25–5.0	5	y = (72.7 ± 1.0)x + (13.6 ± 2.8)	1.4	0.999	0.068	0.55
Tinuvin P	226.0982	[M+H]+	147.1	0.10–5.0	6	y = (31.7 ± 0.94)x + (11.9 ± 2.4)	3.0	0.997	0.097	0.32
Octabenzone	325.1792	[M‐H]−	194.4	0.010–0.50	6	y = (30700 ± 1500)x + (2710 ± 380)	4.8	0.991	0.000023	0.000078
Cyasorb 2908	473.3966	[M‐H]−	225.6	—	1	—	—	—	—	—
Irganox 1330	773.5922	[M‐H]−	300.9	—	1	—	—	—	—	—

^a Travelling wave collision cross section (CCS) experimental values.

^b y = compound peak area; x = compound concentration (µg L⁻¹).

^c Slope % relative standard error.

^d LOD calculated as the concentration that gives a signal‐to‐noise ratio (S/N) of 3.

^e LOQ calculated as the concentration that gives an S/N of 10.

2.5. Sorptive Extraction

2.5.1. UHPLC–IMS–HRMS

The PDMS sampler was secured to a glass‐coated magnetic stirrer bar using stainless steel wire and lowered into the water from the FCMs sampled (refer to Section 2.3) or the matrix‐matched calibration samples using stainless steel forceps [19]. The Schott bottle's plastic lid was lined with a layer of aluminum foil to prevent any contaminants from leaching into the water samples. The water was stirred for 1 h at a speed of 300 rpm, after which the PDMS sampler was removed using stainless steel tweezers, dried with a lint‐free tissue and inserted into a 200 µL conical insert. A 100 µL of 1:1 MeOH:ACN solution was added into the conical insert to cover the PDMS sorptive sampler for solvent BE. The conical insert was placed inside an LC vial, which was sealed and sonicated for 15 mins. It is important to note that 1 mL of water was added to the LC vial prior to sonication to promote energy transfer during sonication for the effective release of sorbed compounds from the PDMS sampler. Finally, the LC vial was opened, and the sampler was removed from the solvent extract using forceps, and the vial was resealed before analysis with UPLC–IMS–MS.

2.5.2. GC×GC–TOFMS

Two LDPE bread bags (6.3 g), food items removed, were sampled by inserting a PDMS sampler in each bag and a knot was subsequently tied in each bag. One bread bag was left to stand in a kitchen for 24 h, while the other was left in the trunk of a car for 24 h. A PP container (1 L; 4.8 g; WestPack, South Africa) and a PS container (1.75 L; 24 g) were microwaved for 3 mins at 750 W. Thereafter, each container was briefly opened to a minimal degree to place a PDMS loop inside for a sorption time of 20 mins, chosen to simulate the average duration in which takeaway food is consumed out of its PP packaging. After the sampling period, the PDMS loop was removed from the FCM with stainless steel tweezers and subjected to GC analysis (refer to Section 2.6.2). The PDMS sampler was thermally desorbed by inserting it into a splitless glass inlet liner (Agilent Chemetrix, Midrand, South Africa) and desorbed at 250°C in a GC inlet. Prior to desorption, the hot inlet liner was manually removed from the GC inlet using a pair of tweezers. The PDMS sampler loop was then inserted into the GC inlet liner, the liner was placed back into the GC inlet, and the run was started.

2.6. Instrumentation

2.6.1. UHPLC‐IMS‐MS

A Waters Synapt G2 HRMS system was used to perform compound separation and detection (Waters Inc., Milford, Massachusetts, USA). The system consisted of a Waters Acquity UPLC linked to a quadrupole‐time‐of‐flight (QTOF) mass spectrometer. MassLynx (version 4.1) software (Waters Inc., Milford, Massachusetts, USA) was used to operate the system. The source conditions were as follows: the capillary voltage for ESI was 2.6 and 2.0 kV for positive and negative mode ionization. The following source conditions were set for both positive and negative ionization modes (EIs): 120°C source temperature, 40 V sampling cone voltage, 4 V extraction cone voltage and 10 L Hr⁻¹ nitrogen cone gas flow. The desolvation temperature and desolvation gas flow were 400°C and 600 L Hr⁻¹, respectively. The acquisition rate of mass spectral scans was 0.3 s in the acquisition mass range of 50–1200 m/z. High‐energy collision‐induced dissociation (CID) was used for tandem fragmentation, where the collision trap energy and collision energy were set at 4 and 6 V, respectively. The voltage ramping scanned from 4 to 15 V for the trap energy and 15 to 45 V for the collision energy. IMS adopted a nitrogen drift gas and a gas flow rate of 90 mL min⁻¹. Within the helium cell, a 180 mL min⁻¹ helium flow rate was used. The mobility t‐Wave was operated at variable IMS wave velocity, which was set at 650 m s⁻¹ together with a 1000–300 m s⁻¹ wave velocity ramp. A 224 m s⁻¹ IMS transfer wave velocity was employed in conjunction with a wave height and transfer wave height of 40.0 and 4.0 V, respectively. Trap DC bias and helium cell DC voltages of 45.0 and 35.0 V were used, respectively. A mobility separation delay was employed, where the IMS wave delay was set at 1000 µs. Finally, to determine experimental travelling wave ^TWCCS values (Ω), the IMS was calibrated using Waters Major Mix Calibration Sample with Driftscope (version 2.8), which produced a ^TWCCS error of < 3.7% (1.07 ± 0.8%, with a range of 0.092%–3.7%).

Separation was achieved using a reverse‐phase step gradient elution scheme from 97% H₂O (0.1% formic acid) to 100% MeOH (0.1% formic acid). A constant column temperature was maintained at 50°C, and a constant flow rate of 0.3 mL min⁻¹ produced a 20 min total run time with a gradient run time of 13 min. Injection volumes were set at 5 µL. A Waters ACQUITY UPLC Ethylene Bridged Hybrid (BEH) C18 was used, which contains 1.7 µm‐sized packing particles and dimensions of 2.1 mm ID x 100 mm length.

2.6.2. GC×GC–TOFMS

Compound separation and detection were achieved using a LECO Pegasus 4D Time‐of‐Flight Mass Spectrometer. The system comprised a modified Agilent 7890 GC that included both a dual‐stage thermal modulator and a secondary oven. Nitrogen gas (produced by a nitrogen gas generator, Peak Scientific Africa, South Africa) cooled by liquid nitrogen was used for the cold jets, and uncooled nitrogen gas for the hot jets. The instrument was operated using ChromaTOF software (version 4.50.8.0 optimized for Pegasus), enabling data capture and processing. The column set used included a Rxi‐1MS primary column with 30 m x 0.25 mm ID x 0.25 um film thickness joined to a Rxi‐17Sil MS secondary column with 0.76 m x 0.25 mm ID x 0.25 um film thickness (Restek, USA). The primary oven's temperature was 40°C for 3 min and subsequently ramped up to 310°C at 10°C/min and held for 5 min. A 10°C temperature offset was adopted for Column 2 relative to Column 1. The modulator temperature was offset by 25°C relative to the secondary oven, with a 4 s modulation period and a 1.005 s hot pulse time. A splitless glass inlet liner (Agilent Chemetrix, South Africa) was used together with a 30 s splitless time and a 250°C desorption temperature. Helium (helium 5.0, Afrox, South Africa) was employed as a carrier gas in constant flow mode at a flow rate of 1.4 mL min⁻¹. The MS transfer line temperature was 280°C, whereas the ion source temperature was 230°C. The ion source was operated in electron EI with an electron energy of 70 eV. An 1800 V detector voltage was used in conjunction with a 100 spectra s⁻¹ data acquisition rate and a 35–550 Daltons (Da) mass acquisition range. Finally, a 5 min solvent delay was adopted when analyzing the spiked PDMS loop.

2.7. Data Analysis

2.7.1. UHPLC–IMS–HRMS

Target compounds were unambiguously identified using retention times, CCS values, accurate masses, and fragmentation patterns that matched those obtained using the certified EL screening standard. An automated data analysis method was established using UNIFI (Waters) in conjunction with the Waters ELs library. Target compounds present within the EL standard but absent in the library were manually added to the library. The mass tolerance criterion was set at 10 ppm, retention time tolerance was set at 0.2 min, and the CCS tolerance was set at 5.0%. The multiple adduct finder and cross adduct combinations were enabled, where the maximum allowed absolute charge for adduct combinations was set at 2. All data were lock‐mass‐corrected with UNIFI before processing. Background laboratory contaminants from the method blank were subtracted from the peak areas of the water samples.

2.7.2. GC×GC–TOFMS

For untargeted analysis, compounds were tentatively identified by comparing full‐range mass spectra of compounds with those in the NIST14 mass spectral library (Version 2.2). The list of detected compounds was narrowed down using a spectral match‐quality criterion of ≥ 75%. In addition, experimental linear retention index (RI) values were calculated based on a method by Van Den Dool and Kratz [20]. Untargeted compounds with experimental RI values within ±35 RI units of the corresponding literature RI values are reported in Table S1. Unequivocal identification of target compounds was achieved by comparing retention times and the full‐range mass spectra of these compounds in the sampled FCMs with those reported in the certified reference standards and NIST14 mass spectral library. The retention time window criterion was set to 3 and 0.1 s for the first‐ and second‐dimension retention times, respectively.

3. Results and Discussion

3.1. Method Performance

The performance of the UHPLC–IMS–HRMS method was evaluated by plotting multilevel calibration curves (matrix‐matched) using at least five concentration levels of the target analytes. The linearity of the matrix‐matched calibrations using the PDMS sampling loop with solvent BE and UHPLC–IMS–HRMS is given in Table 2. The final calibration range for each analyte was selected based on the certified concentration in the standard mixture, the observed instrumental response, and the concentration range over which acceptable linearity was obtained. The linear regression fit (R ²) for all cases was greater than 0.99. Cyasorb 2908 and Irganox 1330 showed no linear response due to their ubiquitous nature; these compounds are thus qualifiable but not quantifiable. CCS matching of the standards to values found in the Waters EL library was done; CCS delta% error ranged from 0.01% for Irganox 245 to 2.6% for TCP. CCS values are provided in Table 3 as these might prove useful for others. The LODs and LOQs ranged from 0.023 ng L⁻¹ for octabenzone to 97 ng L⁻¹ for Tinuvin 360 and from 0.078 ng L⁻¹ for octabenzone to 550 ng L⁻¹ for Tinuvin 360, respectively. A %RSD ranged from 5.0 (Tinuvin P) to 19 (Tinuvin 360)%, excluding Irganox 1076 (26%) and dibutyl sebacate (22%). However, the %RSDs for Irganox 1076 and dibutyl sebacate meet the acceptable criterion of < 30%. Percentage recoveries (accuracy) ranged from 87% for Irganox 1098 to 122 % for both Irganox 245 and Tinuvin 360, excluding dibutyl sebacate (140%) and tris(2‐methylphenyl) phosphate (TCP) (139%) (Table S2).

TABLE 3.

Target migrants detected in drinking water after exposure to a single‐use food contact material (FCM) using UHPLC–IMS–HRMS.

		Concentration (µg L−1) detected in the 100 mL water sampled (n = 1) a
Compound	Comments/uses	PET‐bottle	PET‐bottle car trunk	PET‐bottle UV/Vis	EPS cup hot H2O (2 h)	EPS cup hot H2O (1 h)	PET cup cold H2O	PP straw
Dibutyl sebacate	Plasticizer	0.45	0.35	0.42	0.42	0.46	0.48	0.43
Irganox 245	Antioxidant/thermo‐stabilizer	0.10	0.024	0.087	0.064	0.059	0.16	0.061
Irganox 1076	Antioxidant/non‐discoloring stabilizer	∼19 b	∼24 b	∼14 b	∼25 b	∼14 b	∼14 b	∼19 b
Irganox 1098	Antioxidant/non‐discoloring stabilizer	0.039	0.0065	0.026	0.030	0.0090	0.041	0.024
TCP	Plasticizer/flame retardant	0.025	0.053	0.032	0.037	0.022	0.017	0.026
Tinuvin 360	UV absorber/anti‐weathering agent	2.1	2.6	2.5	2.3	2.7	1.9	2.3
Tinuvin P	UV absorber	1.1	0.33	< LOQ c	< LOQ	0.52	1.2	0.42
Octabenzone	UV absorber	0.13	0.18	∼0.69 b	0.5	∼0.61 b	∼0.63 b	0.5

Abbreviations: EPS, expanded polystyrene; PET, polyethylene terephthalate; PP, polypropylene.

^a 100 mL water sampled; representing a typical drinking amount, n = 1.

^b Approximate value; concentration exceeds the highest calibration level and should therefore be interpreted as an estimated value.

^c Detected, but less than limits of quantification.

Dibutyl sebacate and TCP showed elevated recoveries of 140% and 139%, respectively, exceeding the commonly accepted 70%–120% range for trace organic analysis. This is likely due to matrix enhancement and/or variable low‐level background contamination, as the method blanks showed between‐day %RSD values of 61% for dibutyl sebacate and 56% for TCP over three consecutive days (Table S2). Since matrix‐ and method‐matched calibration curves were prepared and analyzed on the same day as the samples, day‐specific blank responses and matrix effects were partly compensated for during quantification. However, the reported concentrations for these two compounds should be interpreted with caution, as they are likely overestimated.

3.2. Targeted Analyses

3.2.1. UHPLC–IMS–HRMS

The PDMS sampler was used here for the first time to quantify IAS and NIAS migrating from FCMs into drinking water while investigating real‐life scenarios. The results for the targeted analysis using UHPLC–IMS–HRMS are provided in Table 3. Dibutyl sebacate, Tinuvin 360, and octabenzone were present at quantifiable concentrations in all FCMs analyzed (Figure 1). Therefore, these three compounds can be classified as leachables and extractables, as they were liberated from plastic under ambient conditions, as well as under elevated conditions. Therefore, regardless of whether the FCMs analyzed are exposed to elevated temperatures or UV/Vis radiation, human exposure to dibutyl sebacate, Tinuvin 360, and octabenzone is unavoidable due to their prevalence in food or drink media under ambient conditions. Tinuvin P was detected at higher concentrations in the PET‐bottled water obtained directly from the shelf of a grocery store and in the PET cup filled with cold water than under the elevated‐temperature and UV/Vis exposure conditions investigated. These results suggest that, under the conditions evaluated, Tinuvin P (226.0982 m/z) migrated to a greater extent into water under ambient conditions than under elevated conditions. The lower aqueous concentrations observed following elevated‐temperature or UV/Vis exposure may indicate alternative partitioning or loss processes, although volatilization was not directly investigated in this study. Taking into consideration the mass of FCM sampled, dibutyl sebacate, Tinuvin 360, and octabenzone migrated out of the PET cup filled with cold water in far higher quantities than for the remaining samples. For example, 0.048 µg of dibutyl sebacate leached out of the PET cup, whereas 0.045 µg of dibutyl sebacate migrated from the PET bottle analyzed directly from the grocery store shelf (100 mL water sampled). This is quite disconcerting, as lower exposure rates are expected for single‐use FCMs exposed to ambient conditions, such as the PET cup, compared to those exposed to elevated temperatures and UV/Vis radiation. Thus, exposure levels to target ELs are concluded to be highest for PET cups.

FIGURE 1.

Overlay of base peak ion (BPI) chromatograms (UHPLC–IMS–HRMS), using a simplified passive PDMS sampler and solvent back extraction. The purple trace (bottom) shows the extractable and leachable standard reference mixture, and the red trace (top) shows an extract of boiling water after exposure to a polystyrene cup for 2 h.

Irganox 1076 was detected in all investigated FCM exposure scenarios, with estimated concentrations ranging from approximately 14 to 25 µg L⁻¹. The highest estimated concentrations were observed for the PET bottle left in the car trunk and the EPS cup exposed to hot water for 2 h, at approximately 24 and 25 µg L⁻¹, respectively. Lower estimated concentrations were observed for the PET bottle analyzed directly from the grocery store shelf, the PET bottle exposed to UV/Vis light, the EPS cup exposed to hot water for 1 h, the PET cup containing cold water and the PP straw. These results indicate that Irganox 1076 migration was not restricted to elevated‐temperature conditions, but that heat exposure may increase the extent of migration. This is consistent with its use as an antioxidant and non‐discoloring stabilizer in plastic materials. Commission Regulation (EU) No 10/2011 specifies an SML of 6 mg kg^{− 1} food for Irganox 1076 [21]. The highest estimated concentration observed in the present study was approximately 25 µg L^{− 1}. Although this concentration is substantially below the applicable SML on a mass‐concentration basis for an aqueous matrix, the present experiments were not conducted as formal regulatory migration tests; therefore, this comparison should be regarded as contextual rather than as evidence of regulatory compliance. Although Irganox 1076 is neither an endocrine disrupting chemical (EDC) nor carcinogenic, repetitive exposure was found to cause liver damage [22]. Therefore, while the estimated concentrations were below the regulatory limit, the results suggest that consumer practices such as leaving PET bottles in hot vehicles or prolonged hot‐water contact with EPS cups may increase exposure to this additive.

The safety of plastic FCMs is regulated using both specific and overall migration limits (OMLs). Under Commission Regulation (EU) No 10/2011, specific migration limits (SMLs) for individual substances are expressed as the amount of a substance migrating into food or food simulant (mg kg^{− 1}), whereas the OML is 10 mg of total constituents released per dm² of food‐contact surface [21] The experiments performed in the present study were designed to investigate chemical migration under selected consumer‐use scenarios rather than to constitute formal regulatory compliance testing. Consequently, the concentrations measured in the water samples cannot be directly compared with the OML. Nevertheless, only eight target analytes were quantified, while numerous additional non‐target migrants were detected, demonstrating that targeted analysis represents only a portion of the overall chemical migration profile. This highlights the importance of broad screening and identification approaches for improving the characterization of consumer exposure to migrating chemicals. Recent advancements in modeling approaches that characterize the interactions of the FCM with the food matrix have aided in understanding and predicting migration of chemicals from FCM to the food substance [23]. However, these computational models are dependent on an identified chemical compound, thus amplifying the need to identify emerging migrant chemicals by analytical techniques.

3.2.2. GC×GC–TOFMS

GC×GC–TOFMS analysis was used as a complementary technique to UHPLC–IMS–HRMS to maximize the volatility range of detected compounds. The GC×GC–TOFMS workflow was used for screening‐level semi‐quantification only and was not intended to provide fully validated matrix‐matched concentrations. A semi‐quantification was performed by spiking 10 µL of the Waters EL standard directly onto the PDMS loop sampler: 1–10 ng of target analytes. Three compounds were amenable to detection by GC when analyzing the spiked PDMS sampler, namely diethyl phthalate, Tinuvin P, and Irganox 1076. The estimated semi‐quantitative (non‐matrix‐matched) results are provided in Table 4. A limitation of the single‐point, non‐matrix‐matched external semi‐quantification approach is that it does not account for potential differences in thermal desorption transfer efficiency between spiked and field‐sampled PDMS resulting from FCM matrix effects.

TABLE 4.

Compounds detected during targeted GC×GC–TOFMS analysis with their respective chromatographic and mass spectral information and estimated semi‐quantitative amounts detected for food contact materials (FCMs).

							Concentration (ng per FCM) c
Compound	Comments/uses	Molecular weight Da	Formula	CAS	1D RT a (s)	2D RT b (s)	LDPE bread bag car	LDPE bread bag kitchen	Microwaved PP container	Microwaved PS container
Diethyl phthalate	Phthalate	222	C12H14O4	84‐66‐2	996	1.83	170	43	3.4	0.30
Tinuvin P	UV absorber	225	C13H11N3O	2440‐22‐4	1296	2.13	0.55	n.d.	n.d.	n.d.
Irganox 1076	Antioxidant/stabilizer	530	C35H62O3	2082‐79‐3	2008	1.25	65	57	n.d.	n.d.

Abbreviations: LDPE, low‐density polyethylene; n.d., not detected; PP, polypropylene; PS, polystyrene.

^a First dimension retention time.

^b Second dimension retention time.

^c Semi‐quantification = peak area of target analyte in FCM/ peak area of analyte in standard x concentration of standard in ng.

Diethyl phthalate was detected using GC×GC–TOFMS, but not by UPLC‐HRMS. The compound is classified as an EDC; the EU (Directive 2007/19/EU) set the maximum migration limit of diethyl phthalate at 0.3 mg kg⁻¹ food. The exposure to this harmful chemical when leaving a loaf of bread in the car trunk was calculated as 0.24 µg kg⁻¹ of bread, safely below the set limit, assuming that all diethyl phthalate migrated to the 700 g loaf of bread. Heat aggravated the liberation of diethyl phthalate out of the LDPE bread bag where the level of compound released from the LDPE bread bag when exposed to elevated temperatures in the car boot increased by approximately 130 ng when compared to ambient kitchen temperatures (Table 4, Figure 2). Phthalates are a class of plasticizers added to plastic to increase flexibility. The findings support this, as the highly flexible bread bag would contain a larger concentration of phthalates compared to the fairly rigid PP and PS takeaway containers.

FIGURE 2.

Contour plots (GC×GC–TOFMS) of TICs, using a simplified passive PDMS sampler and thermal desorption. The plot on the left gives the chemical profile of a low‐density polyethylene (LDPE) bread bag left in the kitchen, and the plot on the right shows the profile of an LDPE bread bag left in the car trunk for 24 h. Diethyl phthalate—a known endocrine disrupting chemical—is highlighted.

3.2.3. Daily Exposure

It is worth explicitly stating that dibutyl sebacate, Irganox 245, Tinuvin 360, Tinuvin P, and octabenzone are ELs that migrate out of PET bottles during their shelf life and are present in the drink media upon purchase. Interestingly, these five targets are the same compounds released from the analyzed PP straw, except for Tinuvin P. This observation is relevant as a straw is often used to drink beverages out of PET bottles, thereby presenting double exposure to the same EL compounds through a single uptake scenario. Consider a scenario in which an individual purchases a bread and PET bottle filled with water in the morning on their way to work and leaves the FCMs in their car. During teatime, they drink hot coffee out of an EPS cup (250 mL) within 1 h, and during lunch, they drink a cool drink (250 mL) out of a PET cup using a PP straw and eat lunch out of a PP container. On their way home from work, they drink the PET‐bottled water (1 L) that they left in their car. For dinner, they eat out of a PS container, and the following morning have toast for breakfast. Based on the scenario‐specific concentrations and estimated amounts measured in this study, the combined daily exposure to the quantified target migrants was estimated at approximately 42.8 µg of target ELs. Repeating these actions for 7 days a week means that the corresponding hypothetical exposure is 300 µg ELs, whereas continuing for a month or a year results in 1.3 and 16 mg of EL consumption, respectively. Using an average 65‐years life expectancy, by repetition of the above‐mentioned scenarios, a person could consume up to 1.0 g of ELs within their lifetime for the eight quantifiable targets; hence, this is an underestimation of the total ELs that the person is exposed to during their lifetime. This extrapolation is illustrative only because each FCM–scenario combination was analyzed once, the GC results were semi‐quantitative, and the assumption that all compounds detected from the packaging would be transferred to and ingested with the food was not experimentally verified.

3.3. Untargeted Analysis

3.3.1. UHPLC–IMS–HRMS

UHPLC–IMS–HRMS analysis of the various FCMs produced an extensive list of detected non‐targeted compounds (Table 5). Reported non‐targets exhibit a large variety of compounds, namely plasticizers, phthalates, antioxidant stabilizers, UV absorbers, surfactants, and dyes, which exemplifies the trapping efficiency of the in‐house developed PDMS sorptive sampler for a diversity of compound classes at trace levels. Three compounds were detected as being unique to PET bottles, namely bis(2‐ethylhexyl) maleate, a plasticizer; as well as diallyl pentaerythritol and dilauryl thiodipropionate, two stabilizers. Furthermore, 4,4‐bis(2‐benzoxazolyl)stilbene is an optical brightener and methyl palmitate is a plasticizer used exclusively as an ingredient in PP plastic production, which is supported by experimental data, as these two non‐targets were exclusively detected in the PET cup samples [2]. Conversely, 4‐(dodecyloxy)benzoic acid is a major chemical feedstock in the synthesis of PS and was uniquely detected in the 1 and 2 h exposure EPS cup samples [2, 4].

TABLE 5.

Chemical features detected using UHPLC–IMS–HRMS and an in‐house developed PDMS sorptive sampler during an untargeted screening analysis of several FCMs and putatively identified compounds.

									Detector counts d
Compound	Comments/uses	Formula	CAS number	m/z (mass error [ppm])	RT (min)	TWCCSN2 a (Å2) (%CCS error) b	Adduct observed	Fragments (m/z)	PET‐bottle	PET‐bottle car trunk	PET‐bottle UV/Vis	EPS cup hot H2O (2 h)	EPS cup hot H2O (1 h)	PET cup cold H2O	PP straw
1,2‐Dipalmitin	Glycerol	C35H68O5	40290‐32‐2	591.4957 (−0.4)	16.91	262.73 (−0.1)	[M+Na]+	295.1904 523.4721 537.4877 551.5034	4697	n.d.	n.d.	1521	331	1701	1251
4‐(Dodecyloxy)benzoic acid	PS ingredient	C19H30O3	2312‐15‐4	307.2279 (3.8)	13.81	192.29 (−4.3)	[M+H]+	147.0441	n.d.	n.d.	n.d.	56	87	n.d.	n.d.
4,4'‐Bis(2‐benzoxazolyl)stilbene	PP optical brightener	C28H18N2O2	1533‐45‐5	415.1454 (3.2)	12.93	205.34 (−2.0)	[M+H]+	—	n.d.	n.d.	n.d.	n.d.	n.d.	31	n.d.
4,4'‐Bis(diethylamino) benzophenone	UV absorber	C21H28N2O	90‐93‐7	325.2258 (−4.9)	13.42	189.06 (1.1)	[M+H]+	—	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	17
Acetyl tributyl citrate	Plasticizer	C20H34O8	77‐90‐7	403.2341 (3.6)	12.88	218.02 (−4.2)	[M+H]+	—	38	n.d.	22	n.d.	n.d.	n.d.	21
Triallyl pentaerythritol	Antioxidant	C14H24O4	1471‐17‐6	279.1558 (−3.1)	13.68	163.34 (1.6)	[M+Na]+	—	n.d.	n.d.	19	n.d.	n.d.	n.d.	n.d.
Bis(2‐ethylhexyl) adipate	Plasticizer c	C22H42O4	103−23‐1	393.2964 (2.8)	14.84	217.30 (−0.7)	[M+Na]+	—	140	1183	1155	244	18743	n.d.	n.d.
Bis(2‐ethylhexyl) maleate	Plasticizer	C20H36O4	142‐16‐5	363.2502 (−1.1)	14.14	205.81 (−0.5)	[M+Na]+	—	118	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
Bis(2‐ethylhexyl) sebacate	Plasticizer	C26H50O4	122‐62‐3	449.3643 (9.3)	14.03	228.12 (−0.6)	[M+Na]+	—	552	n.d.	n.d.	n.d.	n.d.	n.d.	542
Butylparaben	Paraben preservative c	C11H14O3	94‐26‐8	195.1035 (9.6)	9.77	151.58 (2.4)	[M+H]+	121.0284 139.0390	6114	n.d.	1297	n.d.	n.d.	n.d.	3753
Curcumine	Dye/aroma compound	C21H20O6	458‐37‐7	369.1325 (−2)	12.95	198.08 (−1.7)	[M+H]+	351.1227	157	n.d.	402	497	307	349	n.d.
Diallyl pentaerythritol	PVC stabilizer	C11H20O4	2590‐16‐1	239.1233 (−8.9)	12.57	158.81 (4.6)	[M+Na]+	—	43	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
Diisobutyl phthalate	Phthalate c	C16H22O4	84‐69‐5	301.1388 (−7.2)	7.23	178.60 (−2.0)	[M+Na]+	—	n.d.	n.d.	44	n.d.	80	n.d.	38
Diisodecyl phthalate	Phthalate	C28H46O4	68515‐49‐1	469.3277 (−2.5)	15.51	231.88 (−0.1)	[M+Na]+	57.06988 147.04406 221.08084	5262	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
Diisononyl phthalate	Phthalate c	C26H42O4	28553‐12‐0	441.2935 (−9.2)	14.97	224.09 (−0.6)	[M+Na]+	147.04406 221.08084	43 984	13 581	n.d.	n.d.	n.d.	n.d.	15 438
Dilauryl thiodipropionate	Antioxidant/stabilizer	C30H58O4S	123‐28‐4	537.3941 (−1.3)	16.23	251.96 (−0.1)	[M+Na]+	—	493	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
Glyceryl monostearate	Surfactant	C21H42O4	31566‐31‐1	381.2964 (−2.9)	14.78	205.34 (0.8)	[M+Na]+	—	n.d.	84	84	n.d.	n.d.	52	n.d.
Glycol ricinoleate	Surfactant	C20H38O4	106‐17‐2	365.2691 (7.9)	11.58	201.13 (0.3)	[M+Na]+	307.26316 311.29446	140	n.d.	72	n.d.	n.d.	n.d.	n.d.
Irgafos 168	Antioxidant/stabilizer	C42H63O3P	31570‐04‐4	647.4584 (−0.6)	14.24	278.90 (−1.6)	[M+H]+	133.06479 441.2917	849	394	100	474	484	277	424
Irgafos 38	Antioxidant/stabilizer	C32H51O3P	145650‐60‐8	537.3467 (−0.1)	14.09	238.52 (−2.5)	[M+Na]+	—	n.d.	n.d.	n.d.	n.d.	95	n.d.	48
Methyl palmitate	PP plasticizer	C17H34O2	112‐39‐0	293.2465 (4.6)	12.00	181.09 (−2.5)	[M+Na]+	—	n.d.	n.d.	n.d.	n.d.	n.d.	28	n.d.
NX 8000	Clarifier	C29H40O6	882073‐43‐0	485.2892 (−1.1)	15.24	231.23 (1.1)	[M+H]+	327.19547 329.17474	67	n.d.	n.d.	509	132	n.d.	37
Topanol CA	Antioxidant/PVC additive	C37H52O3	1843‐03‐4	567.3824 (2.7)	14.08	242.21 (0.3)	[M+Na]+	—	95	n.d.	87	n.d.	n.d.	n.d.	335
Tributyl citrate	Thickening agent	C18H32O7	77‐94‐1	361.2247 (7.1)	12.68	194.51 (−0.7)	[M+H]+	—	n.d.	n.d.	68	n.d.	173	n.d.	47
Triton X‐100	Surfactant	C14H22O(C2H4O)7	9002‐93‐1	515.3582 (0.7)	13.76	230.89 (−0.2)	[M+H]+	281.15948 327.20134	104	n.d.	n.d.	201	45	134	198
TXIB	Plasticizer c	C16H30O4	6846‐50‐0	309.2011 (−8.2)	14.52	175.76 (−0.3)	[M+Na]+	73.02841	699	n.d.	799	n.d.	n.d.	n.d.	n.d.

Abbreviations: EPS, expanded polystyrene; m/z, mass‐to‐charge ratio; n.d., not detected; PET, polyethylene terephthalate; PP, polypropylene; RT, retention time.

^a Travelling wave collision cross section.

^b CCS error between experimental and literature CCS values recorded in the Waters Extractables and Leachables library.

^c Classified EDC on the TEDX list of potential endocrine disruptors [22].

^d All values were background‐subtracted using same‐day method blank detector counts.

Increasing the duration for which EPS cups were exposed to boiling water, 2 h exposure compared to 1 h, resulted in higher responses for four non‐targets, namely 1,2‐dipalmitin, curcumin, NX 8000, and triton X‐100. Glyceryl monostearate is a surfactant that was only detected for the PET bottles exposed to elevated temperatures and UV/Vis radiation and, therefore, is delineated as an extractable. UV/Vis‐catalyzed release from PET bottles was uniquely observed for allyl pentaerythritol, tributyl citrate, and diisobutyl phthalate. Of concern is the fact that diisobutyl phthalate is classified as an EDC and, thus, it is recommended to avoid drinking beverages out of PET bottles that have been exposed to sunlight to prevent human exposure to this EDC [22]. Furthermore, bis(2‐ethylhexyl) adipate is an endocrine disrupting plasticizer [22] that was detected in all samples except the PET cup and the PP drinking straw. Thus, manufacturing FCMs from PP avoids human exposure to bis(2‐ethylhexyl) adipate, as well as environmental exposure to the EDC upon disposal of PP plastic items. The detector counts for bis(2‐ethylhexyl) adipate in the shelved PET bottle were very low (140) compared to those for the PET bottle left in the car trunk (1183) and the sun (1155). Therefore, leaving PET‐bottled beverages in a hot car or in the sun should be avoided to prevent high exposure levels to this EDC.

Butylparaben and diisononyl phthalate are two endocrine disrupting compounds that each leached out of the PET bottle analyzed directly from the shelf, as well from the PP straw, as reported in Table 5. Finally, TXIB is an endocrine disrupting plasticizer that was only detected for PET bottles, whereas diisobutyl phthalate is an EDC that was absent from the PET cup yet detected in the PET UV/Vis sample, EPS 1 h sample, and PP straws. Overall, five EDCs were reported among the non‐targets that produced matching CCS values with those in the Waters EL library. Irgafos 168 is a non‐endocrine disrupting antioxidant stabilizer that was detected for all samples and, thus, is a related ingredient in all three PET, PP, and EPS plastic types.

3.3.2. GC×GC–TOFMS

The GC×GC–TOFMS untargeted screening analysis detected over 3000 features per FCM, owing to the high resolution and sensitivity of GC×GC–TOFMS. The list of non‐target compounds per FCM was narrowed down by removing background compounds arising from the PDMS sampler (e.g., siloxanes) and selecting compounds with a spectral match‐quality with the NIST14 mass spectral library of ≥ 75%, which produced a list of approximately 300 detected compounds per FCM analyzed. The list was further refined by selecting specific compounds of interest, whose experimental RI values were within ±35 RI units, or better, of the corresponding literature RI values (Table S1).

Responses for plasticizers and phthalates were substantially lower (Table S1) for the microwaved PP and PS containers (Figure 3) compared to those for the LDPE bread bag. This finding is not unexpected as plasticizers and phthalates are compounds added to plastic to promote flexibility. Of note is that three out of the six plasticizers and phthalates detected are EDCs, namely dibutyl phthalate, diisobutyl phthalate, and dimethyl phthalate. Therefore, while the responses for plasticizers and phthalates are comparatively lower for PP and PS relative to LDPE, these plasticizers and phthalates are still liberated under microwave conditions. Styrene and styrene oxide were found to migrate from the microwaved PS container. Styrene is a chemical feedstock used for the manufacturing of PS, whereas styrene oxide is a PS‐specific plasticizer; thus, their presence is not unanticipated. Nevertheless, their detection in the microwaved PS container is concerning because both compounds are highly toxic to the human body and are classified as EDCs [22]. Microwaving food in its PP or PS takeaway packaging should be avoided to prevent the migration of endocrine disrupting plasticizers and phthalates and other toxic compounds into the food with subsequent consumption.

FIGURE 3.

Contour plots (GC×GC–TOFMS) of total ion chromatograms (TICs), using a simplified passive PDMS sampler and thermal desorption. The plot on the left gives the chemical profile of a polystyrene container after microwave heating, and the plot on the right shows the profile of a polypropylene food container after microwave heating.

The predominant chemical class among the non‐target compounds detected exclusively for the LDPE bread bag left in the car trunk for 22 h consisted of was hydrocarbons, including four straight‐chain alkenes. These hydrocarbons are consistent with exposure to traffic‐related volatile or semi‐volatile organic compounds (Table S1) supposedly from vehicle exhaust fumes. In addition, naphthalene, a polycyclic aromatic hydrocarbon (PAH), was present at a considerably higher level (peak area counts) when compared to the level of naphthalene detected for the LDPE bread bag stored in the kitchen. This establishes a further potential chemical route of exposure when transporting food items in a vehicle. The conclusion is that whilst the LDPE bread bag was left in the car trunk, it was exposed to car exhaust fumes while driving the car around. Therefore, by extension, driving around with groceries in the car results in likely exposure of the groceries to hydrocarbons from the exhaust fumes, which are subsequently ingested when groceries are consumed. Furthermore, an ink solvent, triethylene glycol, was detected for the LDPE bread bag left in the car trunk, pointing to the liberation of ink‐related material used in the bread bag labels from the surface of the bread bag when exposed to elevated temperatures.

4. Conclusion

Our study provides a simplified, green sampling method for the detection and quantification of a range of migrant chemicals from FCMs. The sampling approach overcomes various analytical challenges when analyzing IAS and NIAS in FCM, namely low concentration levels of target analytes, complex sample matrices, and the identification of unknown migrants from recycled plastics that are continuously introduced into FCMs during the plastic manufacturing chain. The sampling method is solvent‐free when using thermal desorption combined with GC×GC–TOFMS, or requires minimal solvent (100 µL) when using back‐extraction combined with UHPLC–IMS–HRMS. Results shed light on the vast range of chemicals consumers are exposed to through use of plastic FCMs. This characterization is vital in order to accurately determine overall exposure limits and to comply with EU legislation (EU No10/2011). The health risks associated with exposure to these migrant compounds emphasize a need to quantify maximum allowable exposure limits during daily use of plastic items. In addition, the health risks associated with many of these compounds remain unknown and are furthermore compounded by the identification of various untargeted compounds also found migrating from the FCMs sampled in this study. Putative identification of untargeted compounds was supported by experimental CCS value matching to known library values, thus strengthening the level of identification. A limitation of the present study is that each FCM–scenario combination was analyzed as a single experimental sample (n = 1). Although the LC–MS method was validated and analytical uncertainty associated with the calibration model is reported in Table 2, the study design does not capture sample‐to‐sample variability within each exposure scenario. The reported concentrations should therefore be interpreted as scenario‐specific values rather than statistically replicated exposure estimates. Future studies should include duplicate or triplicate measurements, particularly for key exposure scenarios, to provide a more comprehensive uncertainty envelope for exposure assessments. The identification and quantification of known and unknown chemical migrants from FCMs provide useful information toward mitigating the risks associated with plastic chemicals, not only regarding human health exposure, but it also has relevance to the environmental risks associated with plastic pollution.

Author Contributions

Gizelle van Niekerk: formal analysis, investigation, writing – original draft preparation. Yvette Naudé: methodology, resources, writing – review and editing. Madelien Wooding: conceptualization, supervision, project administration, validation, writing – review and editing.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: jssc70549‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data supporting this study's findings are available from the corresponding author upon reasonable request.