The formation of benzene in soft drinks has been a longstanding problems for the drinks industry because of its high reliance on sodium benzoate and benzoic acid as a preservative. Here we look at some of the chemistry behind the phenomenon.
Sodium benzoate is a recognised antimicrobial which is very effective in synergy with potassium sorbate. Its optimum microbial activity occurs in the pH range 2.5- 4.0, indicating benzoic acid (pKa = 4.2) is the active form of this agent. It is used in beverages which are naturally acidic (Chichester and Tanner, 1968).
The sodium salt is preferred because of its much higher solubility in aqueous media. It is permitted in beverages at a level of less than 0.1% (ca 7 mM), although higher levels are permitted in some foods. Benzoic acid is found in substantial amounts in cranberry juice.
Mathew & Sangster (1965) showed sodium benzoate was decarboxylated by hydroxyl radical attack. Sagone et al., (1980) used 14C-labelled benzoate decarboxylation to show hydroxyl radical generation by phagocytic leukocytes. Alkoxy1 radicals do not decarboxylate benzoate, suggesting benzoate decarboxylation is a relatively specific indicator of hydroxylradical attack (Winston et al., 1983). Although the three isomeric hydroxybenzoates have also been measured as products of hydroxyl radical attack on benzoate (Loeble et al., 1951; Armstrong et al., 1960), Gardner & Lawrence (1993) did not report benzene production by this reaction.
A report by the German regulatory authorities gives legal guidance on the avoidance of benzene formation & describes the experiments thus:-
The study by Gardner & Lawrence (1993) according to information provided by the authors, the concentrations of the reactants were adjusted to the conditions in foods (e.g. 6.25 mmol/l sodium benzoate and 8 mmol/l ascorbate). In a typical reaction batch for this study the benzene concentration increased during the first 10 minutes and remained more or less steady during the subsequent 30 minutes. During further investigations the benzene concentration was, therefore, determined after a reaction time of 15 minutes at 25 °C. The benzene levels measured were lower than 50 nmol/l and <1 ppb, respectively (please note: 50 nmol/l benzene is equivalent to 3.9 µg/l and 1 ppb equivalent to 1 µg/kg). The detection limit for benzene was approximately 1 nmol/l. In a reaction batch containing 6.25 mmol/l sodium benzoate, the benzene concentration correlated positively with the concentration of the added ascorbate up to an ascorbate concentration of approximately 8 mmol/l. At higher ascorbate concentrations the benzene concentration correlated negatively with the concentration of the added ascorbate. The pH-dependency of the reaction was measured across pH values from 2 – 7. At pH 2 the benzene formation reached a maximum (approximately 37 nmol/l); at higher pH values significantly less benzene was formed. Moreover, an association between the amount of benzene formed and the concentration of copper sulphate (0.05 – 4 mmol/l) and iron sulphate (0.05 – 1 mmol/l) added was examined. The highest level of benzene was reached at 1 mmol/l copper sulphate and 0.05 mmol/l iron sulphate respectively.’
In a study by McNeal et al. (1993) an aqueous solution containing 0.04 % sodium or potassium benzoate (this is equivalent to 2.8 mmol/l sodium benzoate and 2.5 mmol/l potassium benzoate, respectively) and 0.025 % ascorbic acid (this is equivalent to 1.4 mmol/l), was examined.
According to these authors this corresponds to the usual concentrations in beverages. Part of the solution was exposed to UV light, another part was heated to 45 °C and a third part was stored in the dark at ambient temperature. After 20 h UV radiation and heating to 45 °C, respectively, approximately 300 µg benzene/kg were formed in each case. Four µg benzene/kg were formed in the solution stored in the dark at ambient temperature. After 8 days the benzene content in the solution stored at ambient temperature had increased to 266 µg/kg whereas the benzene contents in the heated and UV eradiated solutions remained steady. It is likely that both heat & light together or singly will raise the rate of decomposition of benzoic acid.
In a study by Chang and Ku (1993) results similar to those of the above-mentioned studies were observed. This publication cannot be evaluated in detail because the text is written in Asian fonts but the English abstract and the English captions of the figures show that after 8 days reaction time benzene contents of up to 200 µg/kg were measured and that, for varying ascorbate and/or benzoate concentrations, a reaction optimum was observed for certain concentrations in each case.
To minimize the risk of benzene formation, beverage developers have removed sodium benzoate as a preservative to prevent the formation of benzene. The content of benzene formed is dependent on the concentration of ascorbic acid because the hydroxyl radical formed can either be mopped up by benzoic or ascorbic acid. In so doing, competing agents can reduce the impact that any radical attack has on benzene production.
The pH effect on benzene formation shows that pH 2 & below is severe, & falls away dramatically to pH 7. The hydroxyl radical attack on benzoic acid may yield benzene but attack on the benzoate anion yields other products such as hydroxybenzoates, phenol or biphenyl (Loebl et al., 1951; Armstrong et al., 1960; Sugimore et al., 1960; Sakumoto et al., 1961).
The Measurement of Benzene In Drinks
In the study by McNeal et al. (1993) foods were examined, too. In foods (including some non-alcoholic beverages) containing benzoic acid and ascorbic acid naturally, benzene levels of ≤ 1 µg/kg were detected. In foods to which benzoate and ascorbic acid had been added, benzene levels in a range of < 1 to 38 µg/kg were measured; in non-alcoholic beverages the
levels were ≤2 µg/kg. In foods which allegedly contain benzene according to some older publications, benzene levels of ≤2 µg/kg were measured in the study by McNeal et al. (1993). The authors assume that the higher benzene levels described in the older publications are attributable to laboratory-related contamination.
In a study by Page et al. (1992) carbonated non-alcoholic beverages and fruit drinks were examined. A detection limit for benzene of 0.02 µg/kg was stated. In 20 carbonated nonalcoholic beverages on whose packaging a benzoate additive was declared, the benzene levels ranged between 0.013 and 3.8 µg/kg (mean value 0.79 µg/kg), whilst for six corresponding beverages on which a benzoate additive was not declared the benzene contents levels were between 0.029 and 0.12 µg/kg (mean value 0.062 µg/kg). For 16 fruit drinks which were produced from cranberries (with natural benzoic acid content) and on whose packaging no benzoate was declared, the benzene levels ranged between 0.011 and 1.8 µg/kg (mean value 0.29 µg/kg) and for 13 other fruit drinks which were produced without cranberries they ranged between 0.011 and 0.66 µg/kg (mean value 0.12 µg/kg). The authors report that based on further experiments (which are not, however, described in more detail) laboratory-related benzene contamination was minimal. Although production-related benzene contamination of foods cannot be completely ruled out, the results of this study suggest that benzene can be formed in beverages from benzoic acid.
In a study in which numerous different foods were examined for the occurrence of volatile organic compounds, benzene was found in three Cola beverages (whereby it is not clear how many Cola beverages were examined overall). The benzene levels measured in Cola beverages ranged between 1 and 138 µg/kg (Fleming-Jones and Smith 2003). It is not clear whether benzene was formed in these cases from benzoic acid or reached the beverages as a contaminant during production. In principle, it must also be considered whether any contamination may have occurred during chemical analysis.
References http://www.bfr.bund.de/cm/245/indications_of_the_possible_formation_of_benzene_from_benzoic_acid_in_foods.pdf Indications of the possible formation of benzene from benzoic acid in foods. BfR Expert Opinion No. 013/2006, 1 December 2005
Armstrong, W. A.; Black, B. A.; Grant, D. W. (1960) The Radiolysis of Aqueous Calcium Benzoate and Benzoic Acid Solutions. J. Phys. Chem., 64, pp. 1415-1419.
Chichester, D. F.; Tanner, F. W., Jr. 1968. Antimicrobial Food Additives. In ‘Handbook ofFood Additives’ Furia, T. E., Ed.; The Chemical Rubber Co.: Cleveland, OH USA.
Gardner, L.K. & Lawrence, G.D. (1993) Benzene Production from Decarboxylation of Benzoic Acid In The Presence of Ascorbic Acid and a Transition-Metal Catalyst. J. Agric. Food Chem. 41 (5) pp. 693-695
Loebl, H.; Stein, G.; Weiss, J. (1951) Chemical Reactions of Ionizing Radiations in Solution: VIII. Hydroxylation of Benzoic Acid by Free Radicals Produced by X-rays. J . Chem. Soc., pp. 405-407.
Mathew, R. W.; Sangster, D. F. Measurement of Benzoate Radiolytic Decarboxylation: Relative Rate Constant for Hydroxyl Radical Reaction. J . Phys. Chem. 1965,69,1938 – 1946.
Sagone, A. L., Jr.; Decker, M. A.; Wells, R. M.; Demoko, C. A (1980) New Method for the Detection of Hydroxyl Radical Production by Phagocytic Cells. Biochim. Biophys. Acta, 628, pp. 90-97.
Sakumoto, A,; Tsuchihashi, G.(1961) Radiation-induced Reaction in an Aqueous Benzoic Acid Solution. 11. Determination of Products by Isotope Dilution Method. Bull. Chem. SOC. Jpn., 34, 663-667.
Sugimore, A.; Tsuchihashi, G. (1960) Effect of Metal Ions on the Radiation-induced Decarboxylation of Aqueous Benzoic and Salicylic Acid Solutions. Bull. Chem. SOC. Jpn., 33, pp. 713-714.
Winston, G. W.; Harvey, W.; Berl, L.; Cederbaum, A. I. (1983) The Generation of Hydroxyl and Alkoxy1 Radicals from the Interaction of Ferrous Bipyridyl with Peroxide. Biochem. J . 216, pp. 415-421.
Leave a Reply