European Food Research and Technology

, Volume 221, Issue 3, pp 416–422

Epoxidized soy bean oil (ESBO) migrating from the gaskets of lids into food packed in glass jars


  • Anja Fankhauser-Noti
    • Official Food Control Authority of the Canton of Zürich
  • Katell Fiselier
    • Official Food Control Authority of the Canton of Zürich
  • Sandra Biedermann
    • Official Food Control Authority of the Canton of Zürich
  • Maurus Biedermann
    • Official Food Control Authority of the Canton of Zürich
    • Official Food Control Authority of the Canton of Zürich
  • Franz Armellini
    • Institute for food investigation of the State Vorarlberg
  • Karl Rieger
    • Institute for food investigation of the State Vorarlberg
  • Ingun Skjevrak
    • Norwegian Food Safety Authority
Original Paper

DOI: 10.1007/s00217-005-1194-4

Cite this article as:
Fankhauser-Noti, A., Fiselier, K., Biedermann, S. et al. Eur Food Res Technol (2005) 221: 416. doi:10.1007/s00217-005-1194-4


Epoxidized soy bean oil (ESBO) was determined in foods packed in glass jars closed by lids with a poly(vinyl chloride) (PVC) gasket. The methyl ester of a diepoxy linoleic acid isomer was measured, using transesterification directly in the homogenized food and on-line HPLC-GC-FID analysis. Infant foods from the Swiss market consisting of vegetables, potato and rice or muesli with fruits and berries contained less than 7 mg/kg ESBO, but meat (its fat?) strongly increased ESBO migration up to 86 mg/kg. Some 12% of the products exceeded 15 mg/kg. Austrian and Norwegian samples gave similar results.

Edible oil strongly extracts the ESBO from the gasket in food contact within a few weeks. Since this part of the gasket on average contained 91 mg ESBO, the legal limit is likely to be far exceeded whenever the food contains free oil contacting the gasket, such as oily sauces or vegetables and fish in oil. In fact, the mean ESBO concentration in 86 samples was 166 mg/kg, with a maximum of 580 mg/kg.


Epoxidized soy bean oil (ESBO)On-line LC-GCGasketsLids for glass jarsInfant foodOily food


Epoxidized soy bean oil (ESBO) is used as additive to poly(vinyl chloride) (PVC). It serves as plasticizer and as scavenger for hydrochloric acid liberated from PVC during heat treatment for curing. PVC plasticized with ESBO is widely used as gasket for sealing metal lids on glass jars and for the manufacture of cling films.

The gaskets in lids usually consist of 0.5–1.5 g plastisol, i.e. plasticized PVC. Plastisols usually contain 35–45% plasticizer, often ESBO. Further, some 0.2–1% of a stabilizer is added, such as zinc and calcium soaps (frequently of the non-listed 2-ethyl hexanoic acid [1]), 1–3% slip agents (oleamide and/or erucamide) and silicone oil. Some plastisols contain a blowing agent (so far azodicarbonamide) to reduce the amount of plastisol needed, but this resulted in the migration of semicarbazide [2]. Safe tightness at minimal force (easy opening) is the main technical requirement.

The tolerable daily intake (TDI) of ESBO defined by the Scientific Committee on Food (SCF) of the EU is 1 mg/kg body weight (bw) [3]. This value is based on a toxicological assessment performed by the British Industrial Biological Research Association (BIBRA) in the late 1980s [4]. Toxicity varied with the specification of the ESBO sample. Repeated oral administration affected the liver, kidney, testis and uterus. According to the conventional European rules for food packaging materials, the TDI results in a maximum tolerated migration of 60 mg/kg, but no specific migration limit (SML) was listed in EU directive 72/2002 EC since it corresponds to the overall migration limit (OML).

In 1998 it was published that ESBO migrated into infant foods sold in glass jars with metal lids in amounts sometimes exceeding the TDI [5]. The authors suggested a lower SML for these products, since infants regularly consume a larger amount per body weight than assumed for calculating normal SMLs. Subsequent surveys performed in 1999 and 2002 showed that the ESBO concentrations did not decrease [6, 7].

In 2003, the EFSA calculated the risks of semicarbazide released by the same type of lids into infant foods assuming a worst case consumption from glass jars of 700 g/day for an infant weighing 7.5 kg [8]. Using the same assumptions, the SML for ESBO in infant food would be 10.7 mg/kg. In 2004, the EFSA issued an opinion on ESBO in infant food summarizing average daily intake as 134–234 g [9]. The 95th and 97.5th percentiles were between 375 and 460 g/day. Assuming 450 g being eaten by an infant of 8.5 kg, the TDI is reached with a concentration of 18.9 mg/kg. In fact, “menus” are usually in portions of 180–220 g, the “desert” in quantities of around 120 g. Two menus and a desert consumed in a day add up to about 500–550 g.

In 2002, the SCF accepted a proposal from industry to estimate the mean daily intake of fatty food on the basis of a maximum consumption of 200 g fat or oil per day [10]. The elaborated Fat consumption Reduction Factor (FRF) is applicable to concentrations of migrants in fatty food before the value is compared to an SML. It is calculated from the percent fat or oil in a food divided by 20, i.e. it is 5 for pure fat or oil and 1 for a food containing up to 20 g/100 g fat. The FRF has not yet been transposed into legislation (2004). The fat content in infant food is mostly below 3 g/100 g, i.e. the FRF is not applicable. That of general food often reaches 30–60%, but since the SML is equal to the OML and the FRF is not applicable to the OML, this has no effect on the legal acceptance.

Castle et al. [11, 12] described a method for ESBO analysis based on determining methyl esters of diepoxy linoleic acid, from which the ESBO concentration is calculated. The lipids are extracted from food and transesterified. Then the dioxolanes of cyclopentanone are formed, which enables the detection of the diepoxy linoleic acid by mass spectrometry (MS) without prior preseparation of the methyl esters. We used a method [13] involving transesterification directly in the homogenated food under conditions optimized for determining the fat content and the fatty acid composition [14]. Apart from being faster (90 s at 25 °C), this includes fatty acids of ESBO triglycerides bonded to foods by an epoxy group of one fatty acid. The diepoxy fatty acids were isolated from the transesterified extract by normal phase high performance liquid chromatography (NPLC) and transferred on-line into gas chromatography (GC) with flame ionization detection (FID).

This paper reports first results on ESBO contents in foods packed in jars, starting with considerations of the ESBO content in the gasket, its migration into edible oil and modes of food contact. Results are divided into the two categories of infant products and general food. They show that for certain products the European legal limits are systematically exceeded.


The method for ESBO analysis was described in [11]. Since migration of ESBO may substantially differ between jars, usually the content of 3–5 jars was mixed, including food adhering to the lid. To 100–500 mg of homogenated food (100 mg for oil, 200 mg for sauces and 500 mg for infant food), 4 ml dioxane and internal standard (diepoxy methyl eicosanoate) were admixed, dissolving or suspending the sample. For transesterification, 5 ml methoxide/methanol 5% were added. The reaction was stopped after 90 s (ambient temperature).

The diastereomer of methyl diepoxy linoleate first eluted from GC was measured using the second diastereomer of the internal standard. These methyl esters were isolated from the transesterified extract using a 25 cm × 0.2 mm i.d. NPLC column packed with a cyano phase and 20% methyl tert. butyl ether (MTBE)/pentane as mobile phase. The fraction window was checked by the diastereomers eluted before and after the two key components.

The NPLC fraction of 300 μl volume was transferred on-line to GC using a Dualchrom 3000 instrument (Fisons, today ThermoElectron, Milano, Italy). Eluent evaporation occurred in the concurrent evaporation mode using the on-column interface, a 50 cm × 0.53 mm i.d. coated precolumn and a solvent vapor exit. GC analysis involved a 30 m × 0.25 mm i.d. separation column coated with a 0.2 μm film of PS-255, a methyl polysiloxane, and FID.

The detection limit of the method for infant food was 2 mg/kg ESBO, 6 mg/kg being determined quantitatively. For other foods, the detection limit was similar, with occasional interferences rising the limit to 10 mg/kg. From the validation tests it was concluded that the uncertainty of the measurement was less than 10%. The content of methyl diepoxy linoleate analyzed in the ESBO extracted from 73 lids varied with a relative standard deviation of 5.8%. If the ESBO composition in the lid was not determined, the uncertainty of the measurement was around 15%.

The qualitative analysis of plasticizers in the gaskets was performed on a dioxane extract from the lids (1 h at RT) transesterified as described above during 1 min. As only a small portion of the phthalates is transesterified under these conditions, analysis by GC-FID on a 15 m × 0.25 mm i.d. column coated with PS-255 showed the presence of ESBO, ELO and the phthalates.

ESBO migration

Gasket material in food contact

The lid shown at the left in Fig. 1 is typical for infant products (PT closure = Push on/Twist off), with a gasket positioned by a metal punch and with a thread in the side wall formed by the gasket. The lid at the right (twist closure) is typical for jars used for general food, with metal lugs engaging on the thread of the glass jar and a gasket deposited by a pistol. In the lower right quarter, the lid was cleaned, pointing out the colored (gray) part of the gasket in food contact.
Fig. 1

Inside view of typical lids for infant foods (left) and general food (right) with the bottom right quarter cleaned to better show the gasket. The position of the rim of the jar neck imprinting in the gasket is highlighted on the left

Table 1 lists amounts of gasket material in a broad variety of lids. The gasket is pressed against the rim of the glass jar, which leaves a clearly visible imprint (Fig. 1). Only the material oriented to the center is in food contact. The gasket was cut at the internal edge of the imprint; the inside and the outside part were removed separately and weighed.
Table 1

Amounts of gasket material in some lids, distinguishing between internal zone in contact with food and the external zone; percent of the gasket material and amount of ESBO in food contact assuming 40% ESBO in the gasket


Lid diameter (cm)

Gasket (mg)

In food contact

Food contact



Percentage of gasket

ESBO (mg)

Infant food






















Normal food

























































Mean values







The total amount of gasket material varied between 730 and 1190 mg (average 1021 mg). There was no obvious relationship between the weight of the gasket and the diameter of the lid. Between 155 and 375 mg (average 228 mg) plastisol was in contact with the food. Assuming 40% ESBO in the plastisol, 64–150 mg ESBO were in food contact.

Lids no. 5 and 6 were from the same lot of a product and the difference (160 towards 250 mg plastisol in food contact) shows that the deposition of the gasket is not necessarily well reproduced. The difference is reflected by the ESBO transfer into the food: in the product with lid 5, the ESBO content was 250 mg/kg, in that closed by lid 6 it was 480 mg/kg.

Rate of ESBO migration into oil

ESBO migration into olive oil was determined with lids of the same batch as lid 7 of Table 1. According to the producer, the plastisol contained 44% ESBO, i.e. 134 mg ESBO was exposed to the jar content. The same result was obtained by dissolving the plastisol in warm tetrahydrofuran (THF), precipitating the PVC with methanol and determining ESBO in the supernatant.

For migration testing, the jar was filled with 10 g olive oil, closed with the lid and turned on its head. Table 2 shows that after a heat treatment at 80 °C for 1 h (thermostatted bath of polyethylene glycol 600), 7 mg ESBO was transferred into the oil or 5% of the ESBO in food contact. The duration of 20 or 60 min at 125 °C caused 13 or 19 mg ESBO, respectively, to migrate (10 or 14% of the total). After a day at ambient temperature (no initial heating), 7 mg ESBO migrated into the oil, i.e. 5% of the content in food contact, the same as after 1 h at 80 °C. After 28 days it was 46 mg or 34% of the total. Since the shelf life of such products is far longer, this suggests that heat treatment of the product may not be relevant, since migration at ambient temperature alone may cause a transfer exceeding 50% of the content in the plastisol.
Table 2

Migration of ESBO into olive oil in absolute amounts per lid (mg) as well as in percent of the amount in the plastisol in contact with the jar content



Migration (mg/lid)

Percentage of total

80 °C

1 h



125 °C

20 min



1 h



25 °C

1 day



3 day



6 day



28 day



Combining the data of Tables 1 and 2 it is concluded that from the 64–150 mg ESBO in food contact (average 91 mg) the majority may be transferred into food provided there is sufficient contact to edible oil. It also suggests that many lids have a potential to release an amount of ESBO to oily food which exceeds the TDI. Calculated for a small jar (100 g), the concentrations could reach 500 mg/kg (100 mg ESBO in food contact, 50% migration); for a 250 g jar, the corresponding estimate would be 200 mg/kg.

Food contact

Migration strongly depends on the mode of contact of the food with the gasket. From products immersed in oil, oil squirts to the lid many times. After the first contact, an oil layer remains on the gasket and extracts ESBO. During the next contact (perhaps days later), the oil adhering to the gasket is replaced and the extracted ESBO transferred into the food. The fresh oil picks up further ESBO and the process is repeated.

Some foods (e.g. mustard and mayonnaise) are usually of sufficiently firm consistence that even shaking, e.g. during transport, does not result in contact. This is easily recognized by the lid being clean and the top surface of the food usually reflecting the cone resulting from the filling process. No ESBO has been found in such samples. For certain sauces, such as milled herbs in oil, temperature may be critical: at rather low temperatures the product remains firm (and free of ESBO), but summer temperatures are sufficient to separate the oil from the solids and to cause high ESBO migration. An example is shown in Fig. 2.
Fig. 2

Herbs in oil. Right, sample kept below 20 °C with a firm structure not contacting the lid; ESBO, <3 mg/kg. Left, the same product kept at ambient summer temperature with oil separated and containing 430 mg/kg ESBO

For foods scarcely dissolving ESBO, i.e. containing at most a few percent fat or oil, migration primarily occurs during heat treatment. Infant foods are sterilized during up to 1 h. The jars are rotated in order to mix the product and accelerate heat transfer, which brings the product into intense contact with the lid. Afterwards, some of the paste remains hanging at the lid.

For a sample of infant food prepared from potato, vegetables and meat, the distribution of the ESBO within the jar was determined. At ambient temperature, the paste was sufficiently firm that even a shock did not move it, i.e. it preserved the distribution within the jar as it resulted from sterilization. Samples from the top and the bottom of the filling (totally 180 g) contained 46 mg/kg and 45 mg/kg ESBO (Table 3), respectively, which is indicative of intense mixing during the sterilization process and that no relevant further migration occurred afterwards. The approximately 2 g of food adhering to the lid contained 133 mg/kg ESBO, i.e. merely 3.2% of the migrated ESBO, confirming that migration was relevant only at the high temperature. For practical use this suggests that it makes little difference whether the paste adhering to the lid is consumed by the infants—or recovered for analysis.
Table 3

Distribution of ESBO in a sample of infant food



Concentration (mg/kg)

Amount (mg)

Distribution (%)

Adhering to lid




Top of jar content




Bottom of jar content


The conclusions for the above sample should be taken with care, however. For rather firm pastes it was repeatedly found that thorough mixing of the entire jar content was essential: picking a sample from the upper part resulted in too high ESBO concentrations.

It is concluded that the transfer from the gasket to the food is a complex process depending on numerous parameters, such as the food contact, the heating process accompanying filling and the extractive properties of the food (oil content, availability of the oil).

Variation between jars

As a consequence of variable amounts of gasket material in food contact and variable transfer between the gasket and the filling, ESBO contents in different jars of the same product and even of the same lot may vary substantially. There were products with a high jar to jar reproducibility, such as no. 3 in Table 4, and others, where differences exceeded a factor of two (e.g. products 6, 9–11). For this reason it might be adequate to analyze several samples of a product or to combine the content of several jars and determine a concentration for an averaged sample.
Table 4

Variation of the ESBO contents in different jars of the same lot


Content (g)

ESBO (mg/kg)

Infant food


Vegetables, rice, veal


14, 22


Potato, vegetables, veal


42, 56


Potato, veal


81, 83, 84, 85, 88, 89


Fish, carrot, potato, rice


18, 26, 31


Pasta, beef


24, 27, 30


Turkey, vegetables, rice


9, 11, 28


Spaghetti Bolognese


68, 70


Rice, vegetables, chicken


13, 21, 23, 24

Oily products


Olive paste


108, 266


Peppers in oil


87, 188


Herb sauce


153, 285, 430


Peanut preparation


186, 210


Tomato sauce


101, 110


Herb sauce


80, 91, 161


Rucola/tomato sauce


92, 120

ESBO in infant foods

Figure 3 shows the results from 56 products of infant food surveyed on the Swiss market. They are not fully representative, since later in the campaign the products never found to contain relevant amounts of ESBO (vegetables, products with fruits and berries free of fat) were no longer collected. Therefore the total number of the samples expected to contain less than 5 mg/kg ESBO should be increased by about 20.
Fig. 3

ESBO contents in 56 infant foods from the Swiss market

The median of the 56 products analyzed was 3 mg/kg. Less than 5 mg/kg ESBO was found in 28 out of the 56 products, and this number should be increased by about 20 because of the products not even collected. Eleven samples contained 5–10 mg/kg ESBO, 9 samples 11–20 mg/kg and 8 more than 20 mg/kg.

The products were from four brands. The 4 highest results (32, 49, 65 and 86 mg/kg) were from jars of the same producer. Although the same lids were used and all products were sterilized, 12 of the totally 21 products of this producer contained 2 mg/kg ESBO or less. The highest concentration (86 mg/kg) was in a mixture of potato and 30% veal, the second (65 mg/kg) in a product containing 40% beef next to flour and starch, the third (49 mg/kg) in potato with 10% veal and some vegetable and the fourth (32 mg/kg) in vegetables, corn flour and 10% meat. Also the next five samples of this producer (22, 19, 6, 6 and 4 mg/kg) contained meat. Conversely, in all 12 products without meat, 3 mg/kg or less ESBO was found. This comparison suggests that the presence of meat increased the migration up to 40 times, i.e. that the food composition had a strong influence on the uptake of ESBO. The labeled fat content was below 2 g/100 g in all samples.

The 15 products of brand 2 included samples with 22 and 19 mg/kg ESBO, both with veal. The sample free of meat with the highest ESBO content (13 mg/kg) was prepared with noodles and cream with 2.8 g/100 g fat. Two fruit preparations contained 6 and 7 mg/kg ESBO despite a fat content below 1 g/100 g. Turnips with 1.6 g/100 g fat (corn oil) extracted 6 mg/kg ESBO from the lid.

Of one producer, all 6 products contained less than 2 mg/kg ESBO, but there were no menus containing meat or noodles and the fat content was always far below 1 g/100 g.

From the Austrian market, 31 samples from 5 brands were analyzed. Products were selected among those known to be critical: 24 contained meat, 12 of which with less than 10 mg/kg ESBO, 7 with 10–19 mg/kg ESBO and 5 with 24, 54, 63, 69 and 79 mg/kg ESBO. The highest migration was determined (in sequence of decreasing concentration) in two products of spaghetti Bolognese, potato/carrots/beef, noodles/ham and rice/vegetables/chicken. Among the 7 products free of meat, the highest ESBO concentrations were 12 and 13 mg/kg, both noodle preparations, but from different producers.

Of the 11 products from the Norwegian market, 8 and 2 contained meat and fish, respectively. The mean ESBO concentration was 15 mg/kg, with the top four concentrations being 56, 27, 25, and 16 mg/kg. The lowest result (3 mg/kg) was for the single sample free of meat and fish.

Table 5 summarizes the ESBO content by product group, i.e. products containing meat, those with fish, those mainly consisting of vegetables, noodle preparations and products with fruits and berries. All products with high ESBO contents (above 13 mg/kg) contained meat or fish.
Table 5

ESBO content of infant food grouped by main ingredients



Mean (mg/kg)

Minima/maxima (mg/kg)

















Fruits and berries




ESBO in oily general food

When 1–2% fat in infant food may cause the limit for ESBO to be exceeded, it should be expected that oily general food are even worse—even though only few are as intensely sterilized as infant foods.

ESBO only migrates into oily foods: none of the analyzed jams, honeys, tomato juices, tomato purées and other pastes virtually free of oil or fat contained more than 5 mg/kg ESBO (about 30 samples tested). Sauces with well emulsified oil (no free oil visible) contained ESBO at variable, but modest concentrations and were not considered for the statistics.

Of 48 sauces with free oil on the surface and ESBO as the main or exclusive plasticizer in the gasket, merely two complied with the legal limit of 60 mg/kg. ESBO concentrations varied between 47 and 580 mg/kg, with an average concentration of 183 mg/kg (Table 6). The products primarily consisted of herbs mixed into oil, such as “pesto alla Genovese”, olive pastes with oil and oily tomato preparations sold as spaghetti sauces. The highest concentration (580 mg/kg) was in a herb preparation sold as a 50 g portion, the lowest in a spaghetti sauce with little oil on the surface, on the borderline of being included into this group of products.
Table 6

ESBO concentrations in oily sauces and products in oil



ESBO (mg/kg)



Sauces with oil




Products in oil




The 38 products immersed in oil comprised dry tomato, garlic, peppers, artichokes, olives mixed with cheese, beans, mushrooms, fish and more complex delicacies. The ESBO concentrations referring to the total of the jar content (including the oil) varied between 85 and 350 mg/kg, i.e. not a single product complied with the legal limit. With 145 mg/kg, the mean ESBO concentration was lower than in the sauces, which is primarily the result of the larger package size (a mean weight of 264 g compared to 167 g).


For infant foods in jars it is concluded that there are still products on the market which easily cause the TDI for ESBO to be exceeded. Of several products, the infant cannot even eat up the content of a jar without reaching the TDI (a 8.5 kg infant reaches the TDI with, e.g., 142 g food containing 60 mg/kg ESBO—the jar content is usually around 200 g). The problem has first appeared in scientific literature in 1998 and is still not solved.

For general foods with free oil getting into contact with the lids and gaskets containing ESBO as the main or exclusive plasticizer, the legal limit of 60 mg/kg ESBO in food is nearly always exceeded. The mean ESBO concentration in 86 products was 166 mg/kg, the maximum 580 mg/kg. This does not come as a surprise: since the 1990 publication of Castle et al. [15] at the latest it is in the literature that fatty food causes large amounts of the ESBO to migrate from plasticized PVC into oily products. As the amount of ESBO in the part of the gasket contacting food is in the order of 100 mg, the risk seems obvious that the limit would be exceeded.

The legal limit for ESBO in general food is not only an OML, but also an SML derived from the TDI of 1 mg/kg bw. Since the consumption of the oily products in jars is limited, the risk of exceeding the TDI is low. For a proper exposure assessment it must, however, be taken into account that lids are not the only source of ESBO consumers are exposed to.

There were some oily products containing substantially less or even no ESBO. In these cases, the analysis of the lid revealed that the plasticizer was not ESBO—mostly it turned out to be phthalates, often with a minor addition of ESBO as PVC stabilizer. There is no reason to assume that less phthalates migrate into foods, and the TDI is some ten times lower than for ESBO [16].

To ensure the safety of ESBO, an additional problem must be solved, addressed the first time more than 25 years ago [17, 18]: the identification and toxicological evaluation of the products formed by the intended reaction of ESBO with the HCl cleaved from the PVC during manufacture of the lids. Recently these reaction products were shown to primarily consist of the chlorohydrins of epoxy oleic acid and of cyclic chloro fatty acids from diepoxy linoleic acid [19, 20]. Piccinini et al. [21] found concentrations up to 226 μg/kg in infant food. The toxicological evaluation of these compounds is about to be started; for the time being the limits referring to ESBO do not take these derivatives into consideration [7].

Copyright information

© Springer-Verlag 2005