Development of a Long-term Migration Test Method for Plastic Food Utensils, Containers, and Packaging

Abstract

To ensure the safety of plastic food utensils, containers, and packaging, migration testing is essential for the qualitative and quantitative analysis of chemical substances that migrate from these materials into foods. In the context of foods with shelf lives ranging from several months to several years, conducting actual long-term migration tests is particularly challenging. It is therefore necessary to establish accelerated test conditions that yield equivalent migration levels. In order to establish such accelerated test conditions, results obtained from long-term migration tests using food-simulating solvents are required. However, when conducting long-term migration tests, concerns arise regarding the spoilage of food-simulating solvents and the adsorption of migrated substances onto the test container. To address these problems, model samples were prepared by incorporating ten substances with a wide range of Log Pow values into eight types of general-purpose synthetic resins. Using four types of food-simulating solvents (water, 4% acetic acid, 20% ethanol, and olive oil), potential methods for long-term migration testing were examined. An analytical approach based on liquid chromatography–tandem mass spectrometry (LC-MS/MS) was evaluated and confirmed to be applicable for use with the various food-simulating solvents. More specifically, in the long-term migration test using water, a decrease in the migration amount of dimethyl isophthalate was observed in high-impact polystyrene and polyamide due to the influence of microorganisms proliferating within the migration solution. It was also demonstrated that the addition of sodium azide is effective in preventing spoilage. Furthermore, it was confirmed that the adsorption of substances with Log Pow values of <6 onto glass containers could be considered negligible. Using the LC-MS/MS-based long-term migration test protocol established in this study, it becomes possible to examine conditions for setting accelerated test parameters.

1. Introduction

Synthetic resin utensils, containers, and packaging materials are essential in the distribution, display, and storage of food. However, they can potentially contain monomers, additives, degradation products, or impurities, which could pose a risk to human health¹⁾. To prevent these chemical substances from migrating into food and causing health hazards, Japan implemented the Food Sanitation Act to ensure the safety of such food-contact materials. Traditionally, regulations were based on the Negative List (NL) system (Standards and Specifications for Apparatus, Containers and Packages; Ministry of Health and Welfare (MHLW) Notification No. 370, 1959²⁾), which restricted only substances with safety concerns. However, with recent amendments to the Food Sanitation Act, a Positive List (PL) system was introduced for utensils, containers, and packaging (MHLW Notification No.196, 2020³⁾, issued on May 1, 2020). Subsequently, related notifications were revised (MHLW Notification No.324, 2023⁴⁾), and the PL system became fully applicable on June 1, 2025. Under this framework, only substances that have undergone safety evaluations are to be used as raw materials for synthetic resins. Consequently, when adding new raw materials to the PL, a risk assessment is required. The method for this assessment is outlined in the “Guidelines for the Risk Assessment of Food Apparatus, Containers, and Packages” (revised by the Food Safety Commission of Japan in April, 2024)⁵⁾.

To conduct such risk assessments, migration tests are required to determine the types and amounts of substances migrating from utensils, containers, and packaging into foods. However, since a wide variety of foods come into contact with these materials, conducting migration tests using actual foods is complex and difficult due to the cumbersome procedures involved^6,7). It is therefore necessary to establish simple and broadly applicable testing methods for various products and food types. Generally, migration tests are conducted using various food-simulating solvents instead of actual foods, and the migration amounts obtained are used to estimate exposure levels and concentrations in the diet⁵⁾.

The types of food-simulating solvents and migration test conditions are generally determined based on the contact conditions (e.g., temperature and duration) and the types of food that come into contact with the materials. For example, in the United States, 10% ethanol is used to represent aqueous and acidic foods; 50% ethanol is used for alcoholic foods and dairy products; and vegetable oil is used to represent fatty foods⁸⁾. Meanwhile, in the EU, 10% ethanol is used to represent aqueous foods, 3% acetic acid is used for acidic foods, 20% ethanol is employed for alcoholic foods, 50% ethanol is used to represent dairy products, vegetable oil is used for fatty foods, and poly(2,6-diphenyl-p-phenylene oxide) is employed to represent dry foods⁹⁾.

In terms of the migration temperature and duration, the United States classifies food contact conditions into eight categories, ranging from temperatures of 20 to 121°C and durations of 2 hours to 10 days⁸⁾. In the EU, food contact temperatures are classified into eleven categories ranging from 5 to 225 °C, and contact durations are classified into nine categories ranging from 5 minutes to 10 days⁹⁾, thereby covering a broader range than in the United States. On the other hand, under the NL system in Japan, specific migration conditions are defined, including 60 °C for 30 min and 90 °C for 30 min. However, at the time of this study, no standardized migration test method existed for risk assessment purposes.

In recent years, advancements in food processing and packaging technologies have extended the shelf lives of foods and beverages, contributing to disaster preparedness and a reduction in food waste. As a result, some food and beverages can now be stored for several months or years. However, as the shelf life increases, the migration of chemical substances from utensils, containers, and packaging into foods is expected to rise, thereby highlighting the necessity to establish suitable migration test conditions. Considering the difficulties associated with performing actual long-term migration tests, it is essential to establish accelerated test conditions that can be performed over a short period of time while yielding migration amounts equivalent to those observed in long-term tests. Thus, to determine appropriate accelerated test conditions, actual long-term migration tests must be conducted to compare the migration amounts. However, very few long-term migration tests have been conducted to date^10,11,12,13). Although these studies conducted long-term migration tests over periods such as 1.5 or 2 years, the test materials were largely restricted to relatively clean products, including polyethylene terephthalate (PET) bottles and polymer-coated cans. To determine appropriate accelerated test conditions for various resins, long-term migration testing using material-specific test specimens is required; however, this approach involves several methodological challenges. These include the potential for microbial contamination during the preparation of test specimens and the adsorption of migrated substances onto the test container, both of which may lead to discrepancies in migration amounts.

The aim of this study is therefore to establish a long-term migration test method as a basis for determining appropriate accelerated test conditions for various resins and food types. To address issues specific to prolonged testing—namely, the spoiling of food simulants and the adsorption of migrating substances onto the test container—we examined countermeasures using model samples composed of eight types of general-propose resins blended with ten additives of varying physical properties. Through these investigations, we aimed to provide fundamental insights necessary for the development of a robust long-term migration testing method.

2. Materials and Methods

2.1 Samples

The materials used for the model samples in the migration tests were selected based on their general applicability and physical properties. Eight types of resins were chosen, namely high-density polyethylene (HDPE), polypropylene (PP), PET, high-impact polystyrene (HIPS), polyamide (PA), soft polyvinyl chloride (sPVC), hard polyvinyl chloride (hPVC), and polyvinylidene chloride (PVDC)¹⁴⁾. As additives, ten substances with a wide range of partition coefficients (Log Pow: 1.7–13.4) were selected, as listed in Table 1. These additives were incorporated at concentrations of either 0.5 or 1.0% to produce model samples in sheets approximately 1 mm-thick. These levels were selected as reasonable additive concentrations for preparing representative model materials. Notably, the tributyl acetylcitrate (ATBC) content in sPVC was set at 20%, referencing typical plasticizer concentrations. Additionally, due to boiling point and decomposition concerns, octocrylene was not incorporated into sPVC, and both dimethyl isophthalate (DMP) and octocrylene were excluded from PVDC. Migration tests were conducted using test specimens of the prepared sheets, which were cut into pieces measuring 2 × 5 cm.

Table 1.  Ten substances used for spiking the model samples.

Substances		1	2	3	4	5	6	7	8	9	10
		DMP	DPS	BZP	ATBC	TBPS	Octocrylene	DEHA	Santonox	BNX 1035	Irganox 1076
Log Pow*		1.7	2.6	3.2	4.3	5.7	6.9	8.1	8.2	10.4	13.4
Amount of substance (%)	HDPE	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
	PP	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
	PET	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
	HIPS	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
	PA	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
	sPVC	1.0	1.0	1.0	20.0	1.0	-	1.0	1.0	1.0	1.0
	hPVC	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0	1.0
	PVDC	-	1.0	1.0	1.0	1.0	-	1.0	1.0	1.0	1.0

DMP: dimethyl isophthalate, DPS: diphenyl sulfone, BZP: benzophenone, ATBC: tributyl acetylcitrate, TBPS: 4-tert-butylphenyl salicylate, Octocrylene: 2-cyano-3,3-diphenylacrylic acid 2-ethylhexyl ester, DEHA: bis(2-ethylhexyl) adipate, Santonox: 4,4'-thiobis(6-tert-butyl-m-cresol), BNX 1035: 2,2'-thiodiethylene bis[3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate], Irganox 1076: stearyl 3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate

HDPE: high density polyethylene, PP: polypropylene, PET: polyethylene terephthalate, HIPS: high impact polystyrene, PA: polyamide, sPVC: soft polyvinyl chloride, hPVC: hard polyvinyl chloride, PVDC: polyvinylidene chloride

*: KOWWIN v1.68 estimate, -: No spiking

2.2 Reagents

2.2.1 Standards

Dimethyl isophthalate (>99.0%), diphenyl sulfone (>99.0%), benzophenone (>99.0%), tributyl o-acetylcitrate (>97.0%), 4-tert-butylphenyl salicylate (>98.0%), 2-ethylhexyl 2-cyano-3,3-diphenylacrylate (>98.0%), bis(2-ethylhexyl) adipate (>98.0%), 4,4’-thiobis(6-tert-butyl-m-cresol) (>98.0%), 2,2’-thiodiethylene bis[3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate] (>98.0%), and stearyl 3-(3,5-di-tert-butyl-4-hydroxyphenyl)propionate (>98.0%) were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Dimethyl isophthalate-2,4,5,6-d₄ (>99.9%), tri-n-butyl 2-acetyl-d₃-citrate (>97%), and bis[(±)-2-ethylhexyl] hexane-d₈-dioate (>98%) were purchased from C/D/N Isotopes Inc. (Quebec, Canada).

2.2.2 Reagents

Acetic acid (guaranteed reagent grade), and ethanol (HPLC grade) were purchased from Kanto Chemical Corporation (Tokyo, Japan). Methanol (LC/MS grade), sodium azide (guaranteed reagent grade), formic acid (LC/MS grade), 2-propanol (LC or LC/MS grade), and olive oil were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). A 1 mol/L aqueous solution of ammonium formate was purchased from NACALAI TESQUE Inc. (Kyoto, Japan). Distilled water (18.2 MΩ) was obtained from an Auto Still WG203 apparatus (Yamato Scientific Co., Ltd., Tokyo, Japan). Notably, HPLC grade distilled water (Kanto Chemical Co., Ltd., Tokyo, Japan) was used as a food simulant in the migration tests. To measure the adenosine triphosphate (ATP) levels, a Lucifell HS Set and a Lucifell ATP Standard Reagent Set were employed (Kikkoman Biochemifa Co., Ltd., Tokyo, Japan).

2.3 Apparatus

The determination of additives migrated into food-simulating solvents was conducted using liquid chromatography–tandem mass spectrometry (LC-MS/MS), as will be described later. When using water, 4% acetic acid, and 20% ethanol as food-simulating solvents, an ExionLC^TM series instrument coupled to a Triple Quad 4500 system (SCIEX, Framingham, MA, USA) was employed. When using olive oil as a food-simulating solvent, a Xevo TQD triple quadrupole mass spectrometer (Waters, Milford, USA) was used. ATP luminescence measurements were performed using a Lumitester C-110 instrument (Kikkoman Biochemifa, Tokyo, Japan).

2.4 Migration Procedures

The migration solution was prepared using the immersion method, with the appropriate volume of the food-simulating solvent (40 mL) being applied at a ratio of 2 mL/cm², relative to the surface are of the test specimen (totaling 20 cm² for both sides). A 50 mL glass screw-cap bottle was used for the purpose of the migration tests. The migration conditions, including the temperature, duration, and food-simulating solvent, were selected based on the migration test conditions for long shelf-life foods in the United States⁸⁾ and in the EU⁹⁾.

2.5 Preparation of the LC-MS/MS Solution

When using water, 4% acetic acid, and 20% ethanol as the food-simulating solvent, an aliquot (0.2 mL) of the migration solution was sampled and diluted 5-, 50-, 100- or 500-fold using methanol containing 0.1% formic acid to prepare the measurement solution. Using olive oil, an aliquot (1 mL) of the migration solution was sampled and diluted either 10- or 100-fold using 2-propanol.

2.6 LC-MS/MS Method and Conditions

When using water, 4% acetic acid, and 20% ethanol as the food-simulating solvent, LC-MS/MS analysis was conducted using an Acquity UPLC BEH C8 column (2.1 mm i.d. × 100 mm, 1.7 μm, Waters, Japan) with a column temperature of 30 °C. Mobile phase A consisted of water containing 0.5% formic acid and 1 mM ammonium formate, while mobile phase B consisted of methanol (LC-MS grade) containing 0.5% formic acid and 1 mM ammonium formate. The gradient conditions were set as follows: 65% B (0 min) → 80% B (1.5–4 min) → 95% B (5–9 min) → 100% B (9.5–12.5 min) → 65% B (12.6–18 min). The flow rate was set to 0.2 mL/min, and the injection volume was 4 μL. The solutions were diluted with 20% methanol containing 0.1% formic acid to obtain 60% methanol solutions, which were then used for LC-MS/MS injection. The electrospray ionization (ESI) method was used in positive mode, and the corresponding MS/MS conditions are listed in Table S1. As internal standards, based on retention time proximity, dimethyl isophthalate-2,4,5,6-d₄ was used for substances 1–3, tri-n-butyl 2-acetyl-d₃-citrate was used for substances 4–6 and 8, and bis[(±)-2-ethylhexyl] hexane-d₈-dioate was used for substances 7, 9, and 10 (Table S1).

Using olive oil as the food-simulating solvent, analysis was conducted using an Acquity UPLC BEH C18 column (2.1 mm i.d. × 100 mm, 1.7 μm, Waters, Japan) with a column temperature of 40 °C. Mobile phase A consisted of water containing 0.1% formic acid and 1 mM ammonium formate, while mobile phase B consisted of methanol (LC-MS grade) containing 0.1% formic acid and 1 mM ammonium formate. The gradient conditions were set as follows: 75% B (0–5 min) → 75–100% B (5–10 min) → 100% B (10–20 min). The flow rate was set to 0.25 mL/min. The ESI(+) method was used, and the MS/MS conditions are listed in Table S2. Quantification was performed using the absolute calibration curve method.

2.7 Assessment of the Method Performance

A standard solution of the 10 additives was added to the food-simulating solvents to achieve a concentration of 1 µg/mL, followed by thorough mixing. Dilution procedures were then performed as described in Section 2.5, and a spike test was conducted. For olive oil, verification was also carried out using solutions prepared by adding the standard solution to a concentration of 0.1 µg/mL. The method performance was evaluated using two parallel trials per day over 5 days, and the obtained analytical values were used to determine trueness, repeatability (RSDr), and intermediate precision (RSDi) of the method. The target values were set with accuracies between 80 and 110%, an RSDr of ≤15%, and an RSDi of ≤23%, referenced to the Codex Alimentarius Commission Procedural Manual¹⁵⁾ and the Guidelines on Good Laboratory Practice in Pesticide Residue Analysis CAC/GL 40-1993¹⁶⁾.

2.8 Measurement of the ATP Levels

To confirm the presence of microbial growth in the migration solution, the concentration of ATP derived from microorganisms was measured using the Lucifell HS Set¹⁷⁾. Initially, an ATP calibration curve was created using the Lucifell ATP Standard Reagent Set. A calibration curve was prepared with ATP concentrations from 2.0×10⁻¹² to 2.0×10⁻⁹ (M). Each aliquot (0.1 mL) of the appropriately diluted ATP standard reagent was mixed with the ATP extraction reagent (0.1 mL), followed by the addition of the luminescent reagent (0.1 mL). After stirring for a few seconds, the ATP luminescence was measured using the Lumitester C-110 instrument. Notably, all measurements were performed within 20 s of adding the luminescent reagent, and a calibration curve was established. Subsequently, the ATP luminescence in the migration solution was measured. More specifically, following addition of the ATP elimination reagent (50 μL) to the migration solution (0.5 mL), the mixture was thoroughly stirred and allowed to react at room temperature for 30 min. After this time, an aliquot (0.1 mL) of the resulting solution was transferred to another tube, and the ATP extraction reagent (0.1 mL) was added. After 20 s, the luminescent reagent was added (0.1 mL), followed by stirring for a few seconds. The ATP luminescence was then measured using the Lumitester C-110 instrument within 10 seconds of adding the luminescent reagent. Using the prepared calibration curve, the ATP concentration in the migration solution was determined.

2.9 Statistical Analysis

A Welch’s t-test was used for statistical analysis, considering situations where equal variance could not be guaranteed. The P level was set at 0.05. Microsoft Excel was used to conduct Welch’s t-test and calculate the correlation coefficient of the calibration curve.

3. Results and Discussion

3.1 Selection of Food-simulating Solvents for Long Shelf-life Foods

When selecting appropriate food-simulating solvents, the main focus was placed on long shelf-life foods, excluding dried foods. In the EU, migration tests using poly(2,6-diphenyl-p-phenylene oxide) as a simulant are established for dried foods⁹⁾. Since the testing method is markedly different, dried foods were excluded from this study. The food categories were determined in accordance with the standards and specifications for utensils, containers, and packaging under the Food Sanitation Act of Japan²⁾. For acidic foods, alcoholic beverages, and general foods that do not fall into these categories, 4% acetic acid, 20% ethanol, and distilled water were adopted as food-simulating solvents, respectively. For oils and fatty foods (oily foods), the Food Sanitation Act specifies heptane as the food-simulating solvent. In this study, olive oil was selected as the food-simulating solvent for oily foods considering that this medium has already been adopted in both the United States⁸⁾ and the EU⁹⁾.

3.2 Assessment of the LC-MS/MS Method Performance

The performance of the LC-MS/MS method was evaluated for the ten additives incorporated into the model samples using the four food-simulating solvents employed in the long-term migration tests. Using water, 4% acetic acid, and 20% ethanol as the food-simulating solvents, a calibration curve was created. Since as much numerical data as possible were needed to study the accelerated test conditions, the limit of quantification (LOQ) was set at a concentration that satisfied a signal-to-noise ratio ≥3. Consequently, the LOQs for the ten substances ranged from 0.0002 to 0.001 µg/mL, corresponding to the lowest point of calibration curve (Table S1). The calibration curves had coefficients of determination ≥0.998 (data not shown). A performance evaluation was also conducted through a spike test, which showed a trueness of 84–103%, indicating favorable results for all substances (Table 2). Furthermore, both RSDr and RSDi met the criteria of ≤15 and ≤23%, respectively, demonstrating satisfactory results for all substances.

Table 2.  Results of the spike tests.

Food-simulating solvent	Spiked level	Parameters	Substances
			1	2	3	4	5	6	7	8	9	10
			DMP	DPS	BZP	ATBC	TBPS	Octocrylene	DEHA	Santonox	BNX 1035	Irganox 1076
Water	1 µg/mL	Trueness (%)	95	98	91	86	85	95	97	87	87	86
		RSDr (%)	4.7	4.9	3.5	3.8	4.0	4.2	3.7	6.4	5.1	3.6
		RSDi (%)	4.7	7.5	7.2	11.6	11.4	16.8	6.7	14.2	8.2	12.3
4% Acetic acid	1 µg/mL	Trueness (%)	93	100	93	87	85	95	93	86	86	85
		RSDr (%)	4.5	5.9	6.9	5.6	4.2	3.1	4.7	6.4	2.9	3.2
		RSDi (%)	4.5	7.9	8.1	10.6	10.6	16.8	11.3	17.4	9.1	14.2
20% Ethanol	1 µg/mL	Trueness (%)	98	103	97	92	86	94	96	84	89	87
		RSDr (%)	3.0	3.8	4.1	4.7	4.1	3.8	2.6	6.1	3.7	4.4
		RSDi (%)	4.0	8.3	9.7	11.7	12.1	19.5	9.4	11.8	6.2	9.7
Olive oil	0.1 µg/mL	Trueness (%)	107	108	107	95	107	104	101	107	93	95
		RSDr (%)	4.0	2.4	4.7	2.9	9.0	2.4	7.6	5.1	7.1	15.3*
		RSDi (%)	5.1	2.4	6.1	2.9	14.8	4.1	10.2	5.6	8.1	26.0*
	1 µg/mL	Trueness (%)	103	102	102	101	98	102	90	100	99	104
		RSDr (%)	3.4	2.0	2.9	3.9	17.8*	3.0	6.8	4.6	1.8	11.6
		RSDi (%)	4.3	2.7	4.9	5.2	17.8	3.2	11.0	10.4	8.5	11.6

RSDr:Repeatability, RSDi:Intermediate precision

*: Not meeting the target value

A preliminary study was conducted to evaluate the applicability of GC-MS as an analytical method using olive oil as a food-simulating solvent. Although the pretreatment methods involving liquid-liquid extraction with various solvents were investigated, satisfactory recoveries were not obtained. Therefore, LC-MS/MS was selected with the expectation that trace amounts of olive oil could be eliminated in the ionization chamber. To simplify the procedure as much as possible, the migration solutions were diluted with an organic solvent and directly introduced into the LC-MS/MS system. Considering the solvent compatibility with olive oil and its suitability for application in the LC column, 2-propanol was selected for dilution. Upon preparing a standard solution using 2-propanol containing 10% olive oil, the coefficient of determination for the calibration curve was determined to be ≥0.995 (data not shown). The LOQ ranged from 0.002 to 0.005 µg/mL, corresponding to the lowest point of calibration curve (Table S2). Subsequently, a test solution was prepared by adding ten different additives to the olive oil at concentrations of 0.1 and 1 µg/mL. These solutions were then diluted 10- and 100-fold with 2-propanol, and analyzed using LC-MS/MS to evaluate trueness. As a result, trueness values ranging from 90 to 108% were obtained for all substances, representing a good method accuracy. Although the RSDi for substance 10 (Irganox 1076) exceeded the target value, reaching up to 26.0%, the RSDi values for all other substances met the criterion of ≤23% (Table 2). In addition, it was found that the RSDr slightly exceeded 15% for substance 5 (p-tert-butylphenyl salicylate: TBPS) and substance 10 (Irganox 1076), while all other substances gave satisfactory results. Although some repeatability results exceeded the target criteria, the primary objective of this study was to collect a large amount of data to establish conditions for the accelerated tests, and achieve migration behaviors approximately equivalent to those of long-term migration tests. Therefore, in order to achieve the objective, the results for substances that did not meet the target values should serve ​​as reference values in future studies. Based on these findings, this method was deemed suitable for use in the current study, as it not only required only a simple dilution of olive oil with 2-propanol, but it also allowed rapid and straightforward analysis to be performed. Consequently, in the subsequent investigations, this method was used to measure the migration levels of various substances into the food-simulating solvents.

3.3 Prevention of Spoilage of Migration Solutions in Long-term Migration Tests

When conducting long-term migration tests at room temperature, it was considered that the migration solution may spoil when using water as the food-simulating solvent. The resulting proliferation of microorganisms could lead to degradation of the migrated substances, or their adsorption onto the biomass, leading to inaccurate migration quantification. To address this issue, the effectiveness of sodium azide, a commonly used preservative, was evaluated. Additionally, the potential effects of sodium azide on the migration levels and the analytical performance were examined.

Migration tests were conducted at room temperature (25 °C) for periods ranging from 90 days to 1 year using model samples composed of eight different resins and two types of food-simulating solvents (i.e., distilled water with 0.02% sodium azide and without) and the migration levels observed under both conditions were compared. As outlined in Table 3, among the ten substances evaluated in the current study, substance 1 (DMP) exhibited a phenomenon in HIPS and PA, wherein the migration level initially increased in distilled water (without sodium azide), but then gradually decreased over time. In contrast, in the presence of sodium azide, the migration level increased initially but later remained constant. As a result, the difference in migration levels between the two conditions became more pronounced as the migration period was extended. In HIPS results, the presence of sodium azide resulted in a significant difference in the migration levels of substance 1 (DMP) (Welch’s t-test: t(2) = 4.30, p = 0.015). Although there was a difference in migration levels between the two conditions in PA results, no significant difference was observed (Welch’s t-test: t(2) = 4.30, p = 0.16). For the other substances, no significant differences due to sodium azide addition were observed within the initial 90 days. However, after a period of 0.5 to 1 year, migration levels were generally higher in the presence of sodium azide, particularly for substances with relatively high Log Pow values. Tsochatzis et al.¹⁸⁾ reported that the presence of dissolved sodium chloride (2–10%) during migration experiments in 10% ethanol increased the migration of caprolactam and oligomers from PA films, suggesting that changes in ionic strength may influence the migration of polar substances. In our study, however, the substances showing increased migration in the presence of sodium azide were those with relatively high Log Pow values, and the concentration of sodium azide added to the aqueous simulant was only 0.02%, far lower than the levels examined in previous report. Therefore, the mechanism by which sodium azide contributed to the increased migration observed in our long-term tests remains unclear based on the currently available data. In addition, no obvious spoilage of the migration solution was observed visually.

Table 3.  Impact of sodium azide addition on the amount of migration in a long-term migration test conducted at room temperature (25 °C).

Material	Migration period	Substances
		1	2	3	4	5	6	7	8	9	10
		DMP	DPS	BZP	ATBC	TBPS	Octocrylene	DEHA	Santonox	BNX 1035	Irganox 1076
HDPE	90 days	0.95	0.96	0.93	0.92	0.87	-	-	1.57	-	-
	0.5 year	1.06	1.05	1.05	0.92	0.85	0.30	0.16	0.80	-	-
	1 year	1.18	0.96	1.00	0.77	0.61	0.12	0.08	0.39	-	-
PP	90 days	0.81	0.88	0.98	0.88	0.90	-	-	1.55	-	-
	0.5 year	0.86	0.93	0.96	0.91	0.92	0.39	-	0.89	-	-
	1 year	1.03	1.04	1.00	1.13	1.55	0.49	-	0.64	-	-
PET	90 days	-	0.67	0.89	-	-	-	-	-	-	-
	0.5 year	-	0.82	1.00	-	-	-	-	-	-	-
	1 year	-	0.85	0.94	-	-	-	-	-	-	-
HIPS	90 days	0.16	0.78	0.79	0.68	-	-	-	-	-	-
	0.5 year	0.06	0.83	0.83	0.73	-	-	-	-	-	-
	1 year	0.03	0.80	0.85	0.66	-	-	-	-	-	-
PA	90 days	0.55	1.00	1.00	-	-	0.74	0.79	-	1.62	-
	0.5 year	0.52	1.05	1.00	-	-	0.37	0.33	0.35	2.69	-
	1 year	0.09	1.04	1.06	-	-	0.18	0.21	0.23	-	-
sPVC	90 days	1.00	0.99	1.04	0.96	0.93	-	-	0.62	-	-
	0.5 year	1.00	0.98	1.05	1.09	0.93	-	-	0.66	-	-
	1 year	1.00	1.05	1.04	0.83	0.65	-	-	0.29	-	-
hPVC	90 days	-	0.80	0.82	0.89	-	-	-	-	-	-
	0.5 year	-	0.76	0.83	1.18	-	-	-	-	-	-
	1 year	-	0.96	0.92	1.53	-	-	-	-	-	-
PVDC	90 days	-	1.10	1.11	0.89	0.91	-	-	0.56	-	-
	0.5 year	-	1.00	1.05	0.96	0.88	-	0.09	0.82	-	-
	1 year	-	1.00	1.13	0.69	0.54	-	0.11	0.32	-	-

Ratio of migration without sodium azide addition/migration with sodium azide addition

-:Not calculable because the amount of migration is ND

However, since a decrease in migration levels was noted in the absence of sodium azide, ATP concentration measurements were conducted to determine whether microbial growth had occurred in the migration solution. The ATP concentrations of the migration solutions from HDPE, PP, HIPS, and PA after 1.5 years were measured, and the results are listed in Table 4. In HDPE and PP, where no differences in the migration levels of substance 1 (DMP) were observed, no differences in the ATP concentrations were detected. Conversely, in HIPS and PA, where remarkable differences in the migration levels of substance 1 (DMP) were observed, clear differences in the ATP concentrations were also detected, indicating the presence of microorganisms in the migration solutions containing no sodium azide. In the PA results, a marked difference was observed between the first test solution and the remaining two. In the first test solution prepared with water without sodium azide, the ATP concentration was low and no microbial growth was observed. Consequently, the migration level of substance 1 (DMP) did not decrease in this test solution, which likely contributed to the variability observed in the PA migration results. This interpretation was supported by the relationship between substance 1 (DMP) levels and ATP concentrations in HIPS and PA migration solutions (Fig. 1). The obtained correlation coefficients were determined to be −0.758 and −0.957, respectively, indicating a negative correlation. This suggests that substance 1 (DMP) was likely influenced, and possibly degraded, by microorganisms proliferating in the migration solution. Numerous studies have shown that the phthalates can be degraded by various bacteria in the natural environments^19,20), and also biodegradation of substance 1 (DMP) by specific bacteria has been reported²¹⁾. These results clearly indicated that the addition of sodium azide to the migration solution was effective in inhibiting microorganism proliferation to obtain accurate migration levels of substance 1 (DMP) during long-term migration tests.

Table 4.  Amount of migration of substance 1 (DMP) and the ATP concentration derived from the microorganisms in the migration solution during the long-term migration test (25 °C, 1.5 years).

		Water		Water containing sodium azide
		Amount of DMP migrated (µg/mL)	ATP concentration (x10−12 M)	Amount of DMP migrated (µg/mL)	ATP concentration (x10−12 M)
HDPE	n = 1	39.6	0.33	34.0	0.37
	n = 2	41.5	0.71	34.0	0.29
	n = 3	41.0	0.61	32.7	0.20
PP	n = 1	33.0	0.29	31.9	0.06
	n = 2	33.7	0.10	31.5	0.14
	n = 3	33.2	0.11	31.7	0.07
HIPS	n = 1	0.002	3.67	0.4	0.11
	n = 2	0.007	6.03	0.4	0.06
	n = 3	0.005	1.16	0.3	0.12
PA	n = 1	27.8	0.43	31.2	0.39
	n = 2	0.9	6.73	26.5	0.27
	n = 3	1.0	5.72	31.5	0.23

Figure 1.

 . Plot showing the amount of migrated DMP against the ATP concentration in HIPS and PA

On the other hand, no clear correlation was observed between the ATP concentration and the migration levels of substances other than substance 1 (DMP) (data not shown). Furthermore, for substances with relatively high Log Pow values, a significant increase in the migration levels was observed in the sodium azide-containing aqueous solutions, exceeding those detected in 20% ethanol, and indicating a clear migration-enhancing effect (data not shown). Thus, to verify whether sodium azide promotes migration, migration tests were conducted at four different sodium azide concentrations (i.e., 0, 0.01, 0.02, and 0.1%) using the conditions commonly employed in the EU and the United States for long-term accelerated storage tests at room temperature (i.e., 40 °C for 10 days)^8,9). Upon subsequent comparison of the migration levels in HDPE and PA, it was found that the migration levels remained unchanged regardless of the sodium azide concentration, indicating that no migration-promoting effect occurred during the 40 °C 10 days migration test (data not shown). However, during long-term tests, it was suspected that, for substances with relatively high Log Pow values, sodium azide may interact with the resin and increase the migration levels. Therefore, it was deemed appropriate to adopt the migration level from the sodium azide-containing solution for substance 1 (DMP), and the migration levels from the sodium azide-free solution for the other substances.

3.4 Examination of Glass Container Adsorption during Long-term Migration Tests

In long-term migration tests, accurate migration levels cannot be obtained if the substances are adsorbed onto the migration container (glass bottle). To evaluate the above possibility, the occurrence of adsorption was investigated by comparing the quantified values in two cases: (1) when the migration solution (obtained by exposing HDPE to water and 20% ethanol at room temperature (25 °C) for one year) was measured directly, and (2) when both the migration solution and the methanol rinse solution (obtained by rinsing the glass bottle after removal of the test specimen and the migration solution) were measured together. The same process was conducted for the migration test performed at 40 °C for 10 days. As a result, in long-term migration tests using 20% ethanol as the food-simulating solvent, no adsorption onto the glass bottle was observed. However, when using water, substances with a Log Pow of ≥6 were adsorbed onto the glass (Table 5). This tendency is consistent with previous findings showing that hydrophobic compounds are prone to adsorb onto glassware; for example, it has been reported that pesticides in aqueous solution exhibit increased adsorption to glass beakers as their Log Pow values increase²²⁾. Therefore, when conducting long-term migration tests using water, adsorption onto the glass bottle can be ignored for substances with a Log Pow of <6. On the other hand, using migration conditions of 40 °C for 10 days, no adsorption was observed, thereby confirming that adsorption onto the glass surface can be disregarded. This is likely attributable to the increased solubility of the substances at 40 °C relative to 25 °C, which reduces their propensity to partition to the glass surface. Since substances with a Log Pow of ≥6 are highly hydrophobic, migration amounts to water are expected to be minimal. The most reliable approach for such substances is to rinse the glass migration container with an appropriate organic solvent such as methanol to recover adsorbed compounds. However, this procedure is difficult to apply in long-term experiments designed to monitor temporal changes by repeated sampling of the migration solution. Therefore, when migration of substances with a Log Pow of ≥6 is observed under aqueous, long-term test conditions, the measured amounts should be interpreted with caution because of the potential for underestimation.

Table 5.  Adherence of compounds to the glass bottle during the migration test (HDPE).

Migration condition	Food-simulating solvent	Washing	Substances
			1	2	3	4	5	6	7	8	9	10
			DMP	DPS	BZP	ATBC	TBPS	Octocrylene	DEHA	Santonox	BNX 1035	Irganox 1076
25 °C, 1 year	Water	Not washed	43	29	12	0.37	0.046	0.0048	0.0027	0.010	<0.0010	<0.0010
		Washed	45	30	12	0.39	0.051	0.012	0.011	0.015	<0.0010	<0.0010
		Ratio	0.96	0.97	1.0	0.95	0.90	0.40	0.25	0.66	-	-
	20% EtOH	Not washed	60	53	37	2.0	0.33	0.075	0.0051	0.23	<0.0010	<0.0010
		Washed	61	52	35	2.1	0.30	0.074	0.0054	0.21	<0.0010	<0.0010
		Ratio	0.98	1.0	1.1	0.95	1.1	1.0	0.94	1.1	-	-
40 °C, 10 days	Water	Not washed	36	23	11	0.38	0.032	<0.005	<0.005	0.021	<0.0010	<0.0025
		Washed	36	23	11	0.37	0.033	<0.005	<0.005	0.021	<0.0010	<0.0025
		Ratio	1.0	1.0	1.0	1.0	0.97	-	-	1.0	-	-
	20% EtOH	Not washed	48	28	34	2.7	0.31	0.16	0.016	0.51	0.0021	<0.0025
		Washed	47	28	34	3.0	0.29	0.15	0.015	0.46	0.0019	<0.0025
		Ratio	1.0	1.0	1.0	0.90	1.1	1.1	1.1	1.1	1.1	-

Not washed: quantitative value (µg/mL) without methanol wash, Washed: quantitative value with methanol wash (µg/mL), Ratio = Not washed / Washed

4. Conclusion

In this study, a model system was prepared by adding ten types of additives to eight types of synthetic resins that are frequently used in utensils, containers, and packaging. A long-term migration test method was developed using four types of food-simulating solvents, namely water, 4% acetic acid, 20% ethanol, and olive oil. Initially, the performance of the liquid chromatography–tandem mass spectrometry method was evaluated for the four food-simulating solvents. The target values were met for almost all substances, and it was confirmed that the analytical method was suitable for achieving the objectives of this study. To address concerns related to the potential spoilage of the migration solution when using water as the food-simulating solvent during long-term migration tests, the effectiveness of adding sodium azide as a preservative was investigated. Some substances were highly likely to be affected by microbial activity, and in such cases, the addition of sodium azide to the migration solution was effective in ensuring accurate measurement during the long-term migration tests. Furthermore, the adsorption of migrated substances onto glass container surfaces was also examined. Substances with Log Pow values of ≥6 tended to adsorb onto glass surfaces in the long-term migration tests using water. Although adsorption was negligible for most substances, these results indicate that adsorption may need to be considered when evaluating highly hydrophobic compounds, as it could lead to a slight underestimation of migration levels under specific conditions.

Moving forward, the long-term migration testing method established in this study will be used to examine the conditions for accelerated testing. In deriving these conditions, the results of the long-term migration tests, particularly the time-dependent migration profiles and temperature-related differences, will be used to evaluate the relationship between real-time and accelerated conditions. The establishment of an accelerated testing method will enable efficient and reliable risk assessment of compounds migrating from utensils, containers, and packaging intended for long-term food storage.

Supplementary materials.

foodsafetyfscj-14-1-25-s001.pdf ^{(378.4KB, pdf)}