www.elsevier.com/locate/foodcont
Department of Food Science, Faculty of Food Engineering, State University of Campinas, UNICAMP,
Caixa Postal 6121, CEP 13081-970 Campinas-SP, Brazil
Received 18 July 2001; received in revised form 29 August 2001; accepted 30 August 2001
Abstract
Different brands of soft drink, fruit juice, margarine, yoghurt and cheese available on the Brazilian market were analysed for
benzoic and sorbic acids by high performance liquid chromatography (HPLC) with a photodiode array detector. The levels of
benzoic and sorbic acids were, respectively, in the range of not detected to 804 mg/l and not detected to 1371 mg/kg, respectively.
Only one sample presented a preservative level above that permitted by the legislation in force in Brazil. According to the results it
can be predicted that the real utilisation of benzoates and sorbates is significantly lower than the maximum authorised levels. Ó 2002
Elsevier Science Ltd. All rights reserved.
0956-7135/02/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 5 6 - 7 1 3 5 ( 0 1 ) 0 0 0 8 4 - 6
118 S.A.V. Tfouni, M.C.F. Toledo / Food Control 13 (2002) 117–123
Zealand, Spain, the UK and the USA) provided JECFA muscatel grape), margarine, strawberry yoghurts, petit
information on benzoates intake for assessment during suisse, curd–cheese and processed cheese were purchased
its 51st meeting. This Committee noted that mean intake from supermarkets in the city of Campinas, SP, Brazil,
estimates based on maximum limits specified in national in the period 1999–2000. Three batches of each brand
standards were below the ADI for benzoates. However, were homogenised and analysed in duplicate. Extraction
when the intake estimates were based on the maximum was necessary only for certain food categories, as de-
limits and the range of uses specified in the draft General scribed later. Before injection into the liquid chro-
Standard for Food Additives (GSFA), the ADI was matograph, all samples were filtered through a 0.45 lm
exceeded. The Committee observed that benzoates are filter.
not likely to be used in all foods for which their use is
permitted, concluding that information on these aspects 2.2.1. Soft drink
could be used to revise both the intake assessment and Two hundred ml samples of carbonated soft drinks
the maximum limits specified in the GSFA. The Com- were degassed in an ultrasonic bath for 5 min and di-
mittee also concluded that further information on levels luted with water (Argoudelis, 1984; Veerabhadrarao,
of benzoates in food at the time of consumption is Narayan, Kapur, & Sastry, 1987; Williams, 1986).
needed (WHO, 2000).
There are various methods for the analysis of ben- 2.2.2. Fruit juice
zoates and sorbates in foodstuffs, such as thin layer Samples were centrifuged (2000 rpm) for 10 min and
chromatography, UV spectroscopy, high performance a 1 ml aliquot of the supernatant was diluted with water
liquid chromatography (HPLC) and gas chromato- (Bennet & Petrus, 1977; Bui & Cooper, 1987).
graphy. Nowadays, HPLC is the most common ana-
lytical procedure for the detection and quantification of 2.2.3. Margarine
these preservatives in foods and beverages (Bui & A 5 g sample was dissolved in 50 ml of diethyl ether
Cooper, 1987; Burini & Damiani, 1991; Castellari, En- and extracted twice with 10 ml of 0.1 M sodium hy-
sini, Arfelli, Spinabelli, & Amati, 1997; Chen & Fu, droxide solution in a separating funnel. The basic
1995; Lee, Rouseff, & Fisher, 1986; Leuenberger, aqueous extracts were acidified with 1 ml of 2.5 M sul-
Gauch, & Baumgartner, 1979; Mannino & Cosio, 1996; phuric acid in a volumetric flask and diluted to volume
Monta~ no, Sanchez, & Rejano, 1995; Serrano, L opez, & with water (Leuenberger et al., 1979).
Revilla, 1991).
The objective of the present study was to measure the 2.2.4. Yoghurt and petit suisse
concentration of benzoic and sorbic acids in certain food The extraction procedure was based on the method
categories in which their use is approved in Brazil. This described by Bui and Cooper (1987). A 5 g sample was
information combined with food consumption data will placed into a test tube with screw thread and 15 ml of
be used to further estimate the average intake of these 2% H3 PO4 (ortho-phosphoric acid) added. The solution
preservatives in Brazil. was then shaken in a tube rotator for 1 min, 30 ml of
acetonitrile were added and stirred again for 1 min. The
mixture was filtered through a Whatman no. 1 filter
2. Materials and methods paper and diluted with water.
chromatogram was recorded for 15 min. The detection (10%) was used, and the flow rate was increased to
of benzoic and sorbic acids was carried out at the 1:0 ml min 1 , in order to avoid a long run time. These
wavelengths of maximum absorption of the compounds, analytical conditions resulted in an effective separation
228 and 260 nm, respectively. The peaks of benzoic and of the preservatives for all products analysed in a run
sorbic acids in the samples were identified by their UV time of 15 min. The average retention time was 8 min for
spectrum (200–400 nm) and by comparing the retention benzoic acid and 10.7 min for sorbic acid. The identifi-
time with that of the standard. cation and purity of the peaks were confirmed by com-
paring their spectra (200–400 nm) with those of aqueous
2.4. Quantification solutions of the standards.
The extraction of the preservatives was based on
The external standard plot method was used. Du- procedures described in the literature. Recovery for
plicate injections of 20 ll benzoic acid and sorbic acid benzoic acid and sorbic acid varied from 94.0% to
standard solutions were used to construct linear re- 102.2% and from 86.1% to 101.0%, respectively (Table
gression lines (peak ratio versus acid concentrations). 1). In yoghurt and cheese recovery studies were only
The concentration range of the standard curves was carried out for sorbic acid, because benzoic acid was
0:2–10:0 lg ml 1 . The detection limit (expressed as not detected in these products. The detection limits for
concentration) corresponded to the analyte concentra- benzoic and sorbic acid were, respectively, 0.06 and
tion giving a signal equal to the blank signal plus three 0:05 lg ml 1 in soft drinks; 0.2 and 0:2 lg ml 1 in
standard deviations of the blank (Miller & Miller, fruit juice; 0.06 and 0:05 lg ml 1 in margarine; 0.1
1993). and 0:2 lg ml 1 in yoghurt; 0.1 and 0:07 lg ml 1 in
cheese.
2.5. Recovery study Mean concentrations of benzoic and sorbic acids are
given in Tables 2 and 3. The calculated CVs refer to
In order to verify the accuracy and precision of the duplicates analysis of the homogenised samples (3 bat-
analytical procedure, recovery studies were carried out ches). Figs. 1–6 display histograms of benzoic and sorbic
by spiking selected samples of each matrix with stan- acids levels in all analysed samples. Of all the products,
dards. Benzoic and sorbic acid standard aqueous solu- only one sample of processed cheese had sorbic acid in a
tions (for soft drinks, fruit juices, margarine and quantity above the maximum level allowed by the Bra-
yoghurt) or ethanol solutions (for cheese) were added at zilian legislation (Brasil, 1999). Sorbic acid was found in
three different concentrations. Samples of soft drinks only one brand of a typical Brazilian soft drink
were spiked previously to dilution with water and those (‘‘guarana’’) and two brands of fruit juice (passion fruit
of fruit juices before the centrifugation step. For mar- and West Indian cherry). Two out of three brands of
garine, this procedure was conducted before dissolving traditional cola did not contain any of the preservatives.
with diethyl ether, and for yoghurt and cheese, before In general, traditional colas have a high sucrose con-
the addition of 2% H3 PO4 and ethanol. The ranges of centration and very low pH values, which together in-
concentration added were: 0.2–0.8 mg benzoic acid/ml hibit microbial growth without the need of added
soft drink, 0.05–0.2 mg sorbic acid/ml soft drink, 0.2–0.8 preservative. The use of either benzoic or sorbic acid was
mg benzoic and sorbic acids/g margarine, 0.5–1.0 mg noted in different brands of margarine, whereas only one
benzoic and sorbic acids/ml fruit juice, 0.2–0.5 mg sorbic sample presented both preservatives. The analysis of
acid/g yoghurt, 0.1–0.4 mg sorbic acid/g cheese. The yoghurts and cheese confirmed the absence of benzoic
spiked samples as well as the unspiked controls were acid.
analysed in duplicate. According to the results of this study, although both
Recoveries were calculated from the differences in preservatives are added to different foodstuffs in a wide
total amount of either benzoic or sorbic acid between range of concentrations, none of the samples presented
the spiked and unspiked samples.
Table 1
3. Results and discussion Recovery of benzoic and sorbic acids from different products
Product Mean recoverya (%) SD
The analytical method used for the separation of Benzoic acid Sorbic acid
benzoic and sorbic acids in all samples was based on
Cola soft drink 102.2 2.7 99.3 5.2
Mannino and Cosio (1996). The C18 column used by the Pineapple juice 98.7 1.9 101.0 2.5
authors was 10 cm in length and the mobile phase was Margarine 94.0 1.9 94.6 2.8
supplied at a flow rate of 0:5 ml min 1 . Since the C18 Yoghurt – 86.1 2.8
column used in this present work was 30 cm in length, Cheese – 89.1 4.0
a
more acetonitrile (17%) than the quantity suggested Average of three different concentrations.
120 S.A.V. Tfouni, M.C.F. Toledo / Food Control 13 (2002) 117–123
Table 2
Benzoic acid levels in different food categories
Food category n Maximum permitted level Benzoic acid level (mg/kg or litre) % CV (range)
(mg/kg or litre)a
Mean Range
Soft drink 15 500 225 nd–338 0–6.8
Margarine 6 1000 276 nd–721 0–4.4
Fruit Juices 18 1000 495 nd–804 0–1.8
n: number of analysed samples; nd: not detected. Mean values were calculated assuming nd as zero.
a
Source: Brasil (1999).
Table 3
Sorbic acid levels in different food categories
Food category n Maximum permitted Sorbic acid level (mg/kg or litre) % CV (range)
level (mg/kg or litre)a
Mean Range
Soft drink 15 100 1.0 nd–47.1 0–2
Margarine 6 1000 355 nd–803 0–2.0
Fruit juices 18 1000 51 nd–450 0–0.68
Yoghurt 8 600 148 126–213 0.22–2.3
Petit suisse 3 1000 276 155–367 0.12–4.8
Cheese 6 1000 629 376–1371 0.51–5.9
n: number of analysed samples; nd: not detected. Mean values were calculated assuming nd as zero.
a
Source: Brasil (1999).
levels below the range of antimicrobial activity. The The preservatives were detected in all food products
minimum concentrations of benzoic and sorbic acids whose labels mentioned these additives, and were not
required to inhibit microbial growth may vary due to detected in the products that were not labelled as con-
species, strains, pH of the substrate and other factors. taining them. A preference for using benzoic acid in
Most yeasts and molds are inhibited by 0.001–0.1% beverages was detected, despite the fact that both pre-
sorbic acid, and by 0.002–0.07% benzoic acid (L€ uck, servatives are permitted for use in these products. All
1977). samples (except one) contained benzoic and sorbic acids
S.A.V. Tfouni, M.C.F. Toledo / Food Control 13 (2002) 117–123 121
at levels below the maximum permitted by the Brazilian authorised levels. This information is particularly rel-
food legislation. evant in the case of benzoates, for which intakes ex-
ceeding the ADI have been reported for high
consumers, based on maximum limits specified in na-
4. Conclusion tional standards and on model diets (WHO, 2000).
Therefore, in order to reduce an overestimation, it is
Although the evidence provided is limited, it seems recommended that whenever possible the intake esti-
logic to predict that the real utilisation of benzoates mates of additives be refined by a more precise ap-
and sorbates is significantly lower than the maximum proach using analytical data.
122 S.A.V. Tfouni, M.C.F. Toledo / Food Control 13 (2002) 117–123
References