Elevated antimony concentrations in commercial juices

Abstract

Antimony concentrations up to a factor of 2.7 above the EU limit for drinking water were found in commercial juices and may either be leached from the packaging material or introduced during manufacturing, pointing out the need for further research on the area.

---

Environmental impact

Recent research has documented a dramatic increase of antimony in the environment, raising concern regarding the long-term effects on ecosystems and humans from such increased levels. Evidence indicates that the increased background levels are due to an increased industrial use of antimony and subsequent release from products. In this work we report elevated antimony concentrations in commercial juices. In eight juices (ready to drink) the antimony concentration exceeded the European limit for antimony in drinking water. No previous reports of beverages exceeding this limit exist. Regular monitoring of antimony in consumables and in the environment may be a future requirement.

The human exposure to antimony is increasing, and since Sb has no known biological function there is concern about its long-term effects.¹ Antimony trioxide (Sb₂O₃) is a suspected human carcinogen² and is listed as a priority pollutant by the US EPA and EC. As Sb trioxide is extensively used as a polycondensation catalyst in the production of polyethylene terephthalate (PET), most commercial PET material contains Sb, typically at a concentration of 100–300 mg kg⁻¹.³ Due to the common use of PET as a packaging material in the food industry, Sb frequently comes into contact with food and drink. Consequently, increased Sb concentrations have been observed in products for human consumption. Recent studies have reported leaching of Sb from PET bottles into mineral water^4–7 and citrus fruit juices.⁸ The majority of the examined beverages contained less than 1 μg L⁻¹ though concentrations up to 2.57 μg L⁻¹ have been reported. Thus, the measured Sb levels were below the established safe limits for drinking water (5 μg L⁻¹ (Commission of the European Communities),⁹ 6 μg L⁻¹ (US EPA)¹⁰ or 20 μg L⁻¹ (WHO)¹¹).

Due to the toxicological concern for Sb in drinking water, several studies have analysed the influence of changing physicochemical factors on the dissolution of Sb from PET bottles into water. Westerhoff et al.⁶ reported on the influence of storage temperature (22–85 °C), pH (6.3–8.3) and sunlight exposure (maximum of 7 d). Storage temperature was found to have the most influence, but also the bottle quality and contact area to liquid volume ratio were important factors. Keresztes et al.⁷ obtained similar results but also found that the Sb leaching rate was higher in the case of sparkling mineral waters compared to still waters and attributed this to pH difference (4.94–5.27 and 6.3–8.12 respectively). Shotyk and Krachler¹² studied in detail the rate of change of Sb concentrations in bottled waters kept at room temperature as a function of storage time. They re-analysed 132 waters contained in PET bottles after 6 months of storage. The increase of Sb observed was variable and the authors concluded that the tendencies of the individual bottles to leak varied. Takahashi et al.³ studied the Sb speciation in PET bottles in order to establish a correlation between the Sb speciation in the PET material and the rate of leaching. The results showed large differences between bottles; i.e. in some bottles Sb(III) was present solely whereas the Sb(V) fraction reached 50% in others. However, no correlation between the Sb speciation in the bottles and the rate of leaching could be established, and it was concluded that the degradation of PET itself is more important for the Sb leaching rather than Sb speciation in the PET material.

So far the majority of studies concerning Sb leaching from PET material have focused on leaching into waters under variable physical conditions (time, temperature and sunlight). The only chemical condition that has been thoroughly investigated is pH. However, some evidence exists that the influence of other chemical parameters may be influencing the release of Sb. Westerhoff et al.⁶ showed that higher salt content in waters tend to result in higher Sb concentrations in the bottled waters. Also, some of the highest Sb concentrations measured in commercially available drinks are in fruit juices.⁸ In the relevant study it was speculated that the observed differences in the extraction ability of the individual drinks was due to their different chemical composition and content of, for instance, efficient extractants such as citrate (extracting agents are commonly applied to obtain a quantitative extraction of Sb from solid material).

Stability studies of Sb species have shown that Sb(III) oxidation in natural or synthetic solutions can be prevented in the presence of organic acids,¹³ presumably due to a stabilization via complex formation with the relevant acids. In the study on leaching of Sb from PET bottles into juices of citrus fruit,⁸ it was shown by high performance liquid chromatography-inductively coupled plasma-mass spectrometry (HPLC)-ICP-MS that both a Sb(V)-citrate complex and inorganic Sb(III) occurred, the latter presumably due to stabilization of the trivalent oxidation state by complexation.

In this study, total Sb concentrations have been determined in a selection of different juices, mainly of red fruit juices, contained in either bottles of PET material, glass or in Tetra Pak® cartons. The juices that were analysed were either ready to drink or cordials (to be diluted with water prior to consumption). The study was motivated by (1) previous studies reporting the leaching of Sb from PET materials into the beverages stored in the bottles, and (2) the fact that fruit juices contain high amounts of several organic acids such as citric acid, malic acid and ascorbic acid known to be efficient extractants of Sb and stabilizers of Sb(III). Sb concentrations were determined in 42 beverages of 16 different brands. The beverages analysed included juices of blackcurrant, mixed fruit, strawberry, raspberry, sour cherry, mint and synthetic caramel, obtained from local grocers in Greece, Denmark or Scotland. In total 28 different products were analysed. Because one brand of blackcurrant juice showed particular high Sb concentrations in the initial screening, and the brand is widely available to consumers (marketed in several European countries as in Asia, USA, Australia and New Zealand), 16 beverages (9 different products) were obtained from this manufacturer (referred to here as ‘brand A’). One brand A sample had passed the expiry date; it was included in the study to gain better evidence for correlations between Sb content and expiry date of the samples.

Total Sb concentrations were measured by ICP-MS using In as an internal standard. Quantification (¹²¹Sb/¹¹⁵In) was by the method of standard addition in 2% HNO₃ (w/w). The limit of detection (LOD) was calculated from the linear regression from the value of blanks added 3 times the standard deviation. The LODs for ¹²¹Sb varied between 0.01–0.08 ng g⁻¹. The relative standard deviation (RSD) of the juices varied between 0.23 and 11.3% (except one sample with a concentration of 0.2 ng mL⁻¹ which has an RSD of 44.4%. For quality control, standards (0.2 ng g⁻¹ or 1.0 ng g⁻¹) and standard reference material (SRM 1643d, Trace Elements in Water, from National Institute of Standards & Technology, Gaithersburg, MD USA) were analysed regularly during each experiment and compared to the calculated value. The standards had an RSD < 3.35% (29 measurements in total during the experiments). The concentration of Sb in the SRM 1643d was quantified to 54.2 ± 0.9 ng g⁻¹ (12 measurements) (certified; 53.2 ± 1.1 ng g⁻¹).

Several juices contained in either glass, Tetra Pak® or PET bottles had elevated concentrations of Sb (Fig. 1) compared to background levels normally present in drinking water.⁵ The concentrations shown are those present in the juices when ready for drinking i.e. after cordials have been diluted as prescribed. Eight of the juices contained Sb concentrations above the legal maximum concentration for Sb in drinking water in EU (5 μg L⁻¹). The highest Sb concentration in a ready to drink juice was in a juice of sour cherry (glass bottle) produced in Greece (13.6 μg L⁻¹). It has been reported that glass containers can release Sb to their contents,¹⁴ thus Sb may originate from either the bottle or be due to local contamination. The issue deserves more attention and will not be dealt with in this communication.

Fig. 1 Sb concentrations (μg L⁻¹ of sample) in juices ready for drinking (diluted after suggested dilution factor) contained in different bottles. Error bars represent standard deviations on the concentrations. * indicates beverages of same brand (brand A in the following—the remaining 26 beverages were of 15 different brands). The dotted line shows the maximum limit of Sb in drinking water in the EU (5 μg L⁻¹).

Blackcurrant juices of a specific brand (marked with * in Fig. 1) showed very high Sb concentrations. From this brand, both juices contained in PET bottles and in Tetra Pak® had elevated Sb concentrations. The highest Sb concentrations measured in non-diluted cordials were 12.3, 30.2 and 44.7 μg L⁻¹ (samples 30, 35 and 41 in Fig. 1—all from brand A). The latter two samples thereby exceed the maximum limit for Sb in drinking water in EU when diluted as prescribed (i.e. six-fold); however, it should be noted that the cordial with the highest concentration was past its expiration date. This might indicate the importance of consumers not using cordials that have expired, as this could lead to an increased Sb intake. Likewise, will consumers diluting their cordial less than prescribed, liking it strong in taste, get an enlarged exposure to Sb?

It is worth noting that the guideline limits referred to earlier are defined for drinking water, in which Sb takes the form of the Sb(V) oxo-anion ([Sb(OH)₆]⁻), which is the less toxic form of Sb.^11,15 It is obvious that the form of Sb in the drink is a key determinant of its toxicity. In the study concerned with Sb in citrus juices,⁸ speciation analysis revealed that either the more toxic inorganic Sb(III) (44 ± 17%) or an Sb(V)-citrate complex of unknown toxicity (41 ± 20%) were the main species present in the juices. Sb speciation, using the conventional HPLC-ICP-MS technique as applied on juices,⁸ was attempted but failed, possibly due to the high sugar content in the relevant juices (57–66% w/v) interfering with the chromatography.

When analysing all the samples as one data set, no obvious correlations could be established between Sb concentration and expiration date or any chemical properties (% of juice, density, carbohydrate content or pH). This was not unexpected due to variable reactivities of different bottle types^3,12 and the significant differences in the types of juices analysed.

Next, the results for only the brand A samples (n = 16) were examined. All brand A samples were blackcurrant, and although they were obtained in Denmark, Greece or Scotland, all were produced in UK. Among them were PET bottles as well as Tetra Pak® cartons, cordials as well as ready to drink-juices, and light product as well as regular. Thus, the carbohydrate content varied between 5.2 and 660 g L⁻¹ (according to the nutritive information on the containers). The highest Sb concentrations were measured in the juices with the highest carbohydrate content, and especially for the juices with high carbohydrate content there was a trend towards increasing Sb concentration with decreasing days left until expiration (Fig. 2). The dependence of two factors blurs an obvious correlation, and although the influence of expiration date and carbohydrate content are both statistically significant (P = 0.047 and 0.001 respectively, Pearson product moment correlation analysis), poor correlation coefficients were observed (R² < 0.54). However, multiple linear regression, which takes into account both the time to expiration and carbohydrate concentration simultaneously, revealed a much better correlation (R² = 0.857) and as observed with individual correlation analysis, both factors were found to have statistically significant influence (P ≤ 0.0001). The regression (y = 11.78 − (0.0737 × days to expiration) + (0.0375 × carbohydrate)) suggests that carbohydrate concentration as well as days from expiration influences the Sb concentration.

Fig. 2 The 16 samples of brand A: Sb concentration as a function of carbohydrate content and days to expiration. The multiple linear regression fit is shown.

The expiry date will most likely be connected to the production date of a sample and thus to sample's age. Thus, the correlation between Sb content and expiry date suggests that the Sb content is higher in the elder of the brand A samples. This might indicate that Sb is leached from the packing material over time. The significance of carbohydrate concentration might indicate that carbohydrate aids extracting Sb. Carbohydrates can form complexes with Sb(V) via vicinal hydroxyl groups¹⁶ and could potentially function as extracting agents.

On the other hand, it is peculiar that the concentrations in brand A juices are so much higher than for other brands. A possible explanation is that the quality of the PET material used for bottling is poor. However, even the juices in Tetra Pak® displayed elevated concentrations, and we have not been able to find any documentation indicating that PET should be in contact with these drinks, which suggests that Sb was present in the juice prior to packaging. Thus, the Sb might originate from some contaminated ingredient or from the production equipment.

The juices with the highest carbohydrate contents are cordials and will therefore have higher concentrations of most ingredients compared to the ready to drink-juices. Thus, the statistical significance of carbohydrate concentration could be caused by the presence a contaminated ingredient. To account for this, Pearson product moment correlation analysis was performed on the Sb vs. carbohydrate concentration in the brand A juices when diluted as suggested by the producer, and likewise when Sb and carbohydrate content were normalised to the percentage of blackcurrant concentrate in the products. In both cases, the correlation still proved statistically significant (P = 0.003 and 0.015, respectively). As carbohydrate and Sb concentration correlates even when the concentration differences between the juices are levelled, it is not obvious that the correlation between Sb and carbohydrate concentration should be due to some contaminated ingredient; however, the possibility cannot be completely excluded (if, for instance, the sugar used in the production was contaminated we should obviously still expect a statistically significant correlation).

Of course, the Sb may partly be leached from packing materials and partly be present prior to packaging. Such a mixed origin will make it harder to draw conclusions about the source.

In conclusion, we have measured Sb in juices with up to 17-fold higher concentrations compared to previous reports on beverages in PET-bottles. Trends in the data indicate that the Sb has leached from the packing material; however, it cannot be excluded that the Sb was present prior to packing. Thus, further studies are warranted.

Acknowledgements

We acknowledge Kaj Jørgen Hansen for providing sample material and H. R. H acknowledges the assistance provided by the Royal Society of Chemistry in the form of a Jones Travelling Fellowship.

References

1. W. A. Maher, Environ. Chem., 2009, 6, 93–94.
2. Iarc, Monographs on the evaluation of carcinogenic risk to humans: Some organic solvents, resin monomers and related compounds, pigments and occupational exposures in paint manufacture and painting, International Agency for Research on Cancer, Lyon, France, 1989.
3. Y. Takahashi, K. Sakuma, T. Itai, G. Zheng and S. Mitsunobu, Environ. Sci. Technol., 2008, 42, 9045–9050.
4. Bundesamt für Gesundheit, Bulletin 44/05, Bundesamt für Gesundheit, Bern, Switzerland, 2005, pp. 796–797.
5. W. Shotyk, M. Krachler and B. Chen, J. Environ. Monit., 2006, 8, 288–292.
6. P. Westerhoff, P. Prapaipong, E. Shock and A. Hillaireau, Water Res., 2008, 42, 551–556.
7. S. Keresztes, E. Tatár, V. G. Mihucz, I. Virág, C. Majdik and G. Záray, Sci. Total Environ., 2009, 407, 4731–4735.
8. H. R. Hansen and S. A. Pergantis, J. Anal. At. Spectrom., 2006, 21, 731–733.
9. EC 2003, European Commission directive 2003/40/EC of 16 May 2003.
10. US EPA 2009 National primary drinking water regulations, United States Environmental Protection Agency, 2009.
11. World Health Organization, Antimony in drinking-water, Geneva, 2003.
12. W. Shotyk and M. Krachler, Environ. Sci. Technol., 2007, 41, 1560–1563.
13. M. de la Calle-Guñtias, Y. Madrid and C. Cámara, Fresenius’ J. Anal. Chem., 1992, 344, 27–29.
14. M. J. Hynes, S. Forde and B. Jonson, Sci. Total Environ., 2004, 319, 39–52.
15. M. Filella and P. M. May, Geochim. Cosmochim. Acta, 2003, 67, 4013–4031.
16. H. R. Hansen and S. A. Pergantis, Anal. Bioanal. Chem., 2006, 385, 821–833.

---