The reagents Tris(hydroxymethyl)aminomethane (TRIS), 2-hydroxyisobutyric acid (HIBA), and sorbic, benzoic and salicylic acids were purchased from Sigma-Aldrich, Merck (St. Louis, MO, USA). The stock solutions of TRIS and HIBA (components of the background electrolyte) were prepared in ultrapure water (Milli-Q, Millipore, Bedford, MA, USA) at the concentration of 100 and 125 mmol/L, respectively. The stock solutions of sorbic, benzoic and salicylic acids were prepared in methanol (Merck) at the concentration of 1000 mg/L and kept under refrigeration ((5±2) °C) until the analysis, when they were diluted to obtain the working concentrations.

The analyses were performed in a capillary electrophoresis (CE) system (model 7100; Agilent Technologies, Palo Alto, CA, USA) equipped with a diode array detector. The data acquisition and treatment software were supplied by the manufacturer (HP ChemStation®).

The electrophoretic separations were performed according to Costa et al. (16) in an uncoated fused-silica capillary of 32 cm (8.5 cm effective length×50 μm inner diameter×375 μm outer diameter). The short-end injection mode was applied in this study because it allowed a faster analysis with excellent separation of the compounds. The applied voltage was 30 kV, with positive polarity at the injection side, the temperature was maintained at 25 °C, and UV detection was conducted at 200 nm for benzoate and salicylate, and at 254 nm for sorbate. The background electrolyte was composed of 25 mmol/L TRIS and 12.5 mmol/L HIBA, at pH=8.1. Salicylate was used as the internal standard.

A total of 30 samples of mustard (N=10), ketchup (N=10), and tomato sauce (N=10) from different brands were purchased in a local supermarket of Florianopolis, Santa Catarina state, Brazil.

The sample preparation procedure was modified from that proposed by Javanmardi et al. (12), who used an aqueous dilution followed by dispersive liquid-liquid microextraction (DLLM). In the original protocol, 5 g of sauce sample were diluted in 50 mL of water and centrifuged at 1784×g for 5 min. An aliquot of 5 mL of the supernatant was transferred to a test tube in which 750 mg NaCl and 1.2 mL acetone (disperser solvent) containing 250 µL chloroform (extraction solvent) were added. The mixture was homogenized for 2 min and centrifuged at 1784×g for 10 min. The sediment phase was quantitatively placed into another test tube and evaporated with nitrogen gas. The residue was redissolved in 100 µL mobile phase for high-performance liquid chromatography (HPLC) analysis. In the present study, a simple sample preparation based only on aqueous dilution was employed as follows: samples ((1.00±0.05) g) were weighed into polypropylene tubes, and 10 mL of ultrapure water were added. The tubes were sealed, and the resulting solution was stirred with a vortex (model 772; Fisatom, São Paulo, Brazil) for 1 min and centrifuged for 5 min at 9861×g (Mini Spin plus, Eppendorf, Hamburg, Germany). The supernatant was filtered through syringe filters (0.45 µm pore size, Millipore) and diluted in the proportion of 5:1 (V/V) with salicylate (internal standard) at the concentration of 120 mg/L before being injected at the CE system. Subsequent dilutions were performed when necessary. For all experiments (quantification and validation) the corrected peak area (A_analyte/A_salicylate) and the corrected migration time (t_m(analyte)/t_{m(salicylate)}) were considered.

All samples were prepared in three independent replicates and analyzed before the expiration date.

The analytical method was originally proposed and validated by Costa et al. (16) to assess sorbate and benzoate in soft drinks, juices and teas. Regarding the application of this method to different matrices (mustard and tomato-based samples) and a change performed in the linear range, it was necessary, in this study, to reassess the validation parameters such as linearity, matrix effect, selectivity, precision and accuracy. The validation protocol followed the Eurachem guidelines (22). In addition, the greenness of the analytical procedures employed in this study was quantitatively determined using the Eco-Scale proposed by Gałuszka et al. (23).

In this study, the capillary zone electrophoresis (CZE) method proposed by Costa et al. (16) for non-alcoholic beverages was successfully applied to determine sorbate and benzoate in mustard, ketchup and tomato sauce, as shown in Fig. 1.

This method allowed the separation of both ions and the salicylate in just 28 s. The total analysis time of the method, which includes flushing, vial permutation, injection and separation, was experimentally determined and took about 1.3 min, achieving an analytical throughput of approx. 40 samples per hour. Moreover, the high throughput of the CE method is noteworthy compared to chromatographic methods such as NMKL (Nordic Committee on Food Analysis) method 124 (24), in which the target analyte separation takes over 15 min.

As shown in Table 1 (16-19, 25, 26), different separation conditions and background electrolyte (BGE) composition can be used for sorbate and benzoate determination by capillary electrophoresis. However, the method proposed by Costa et al. (16) and employed in the present study allowed the separation of both analytes in less than 30 s. In contrast, separation times between 5 and 7 min were usually necessary when other separation conditions were applied. In addition, the limits of detection and quantification found by Costa et al. (16) were similar to those reported by the other studies indicating that the short separation time did not impact the sensibility of the method.

The successful application of the CZE method in this study also depended on the use of suitable sample preparation. The extraction of the analytes was based on modifications of the procedure proposed by Javanmardi et al. (12), who determined sorbate and benzoate in ketchup and other foods using the pre-preparation step followed by DLLM coupled with HPLC. Besides the non-utilization of DLLM, the modifications performed in the present study made the extraction procedure simpler and more appropriate for green chemistry by reducing the sample mass and the residue generation. In addition, the proposed method proved to be adequate to extract the analytes from the analyzed samples, eliminating the need to use one more step of cleaning and concentration of the analytes, such as in DLLM. Despite the use of minimum sample preparation steps, the peaks in electropherograms are well separated, as shown in Fig. 1b and Fig. 1c.

To ensure the fitness for purpose and the veracity of the data, it was necessary to reassess the validation parameters of the CZE method for the new food matrices investigated in this study.

The validated method was applied for the analysis of sorbate and benzoate in 30 commercial samples of mustard, ketchup and tomato sauce from different manufacturers. The results are shown in Table 4.

Among the analyzed samples, six of them declared ’no preservatives’ on the label, and this claim was confirmed experimentally. Also, of the thirty evaluated samples, seven showed some kind of non-compliance. Significant problems involving mislabeling and non-compliance with the Brazilian legislation were found in the ketchup samples, where four out of ten assessed samples had some irregularity.

The Brazilian legislation only allows the addition of sorbic acid and its salts at the maximum mass fraction of 1000 mg/kg in ketchup (10). However, four analyzed samples contained sodium benzoate as a preservative. Ketchup samples A and G reported the use of sorbate on their label, although benzoate was detected experimentally. On the contrary, ketchup samples D and I declared benzoate on the ingredient list, despite the regulatory recommendation. Besides these mislabelings, ketchup samples G and I had benzoate mass fractions ((1054±70) and (1479±36) mg/kg, respectively) higher than the maximum limit of 1000 mg/kg established by the European Commission (9), which represents an infringement if these products are marketed in Europe. All ketchup samples that contained sorbate followed the regulatory limit of both legislations.

Regarding tomato sauce, all samples were in accordance with the limits established by the European and Brazilian legislation that allows the use of sorbate and benzoate at mass fractions of 1000 mg/kg (9, 10). Experimentally obtained results confirmed that tomato sauce samples B and C also contained mislabeling related to the presence of sorbate and benzoate in the ingredient list.

For mustard, European and Brazilian legislation allows the addition of sorbate and benzoate (9, 10), but in nine out of ten evaluated samples only benzoate was detected. The mass fractions of sorbate and benzoate in all samples were in accordance with the legislation (1000 mg/kg) and with their label.

Despite the compliance of almost all investigated samples with the European and Brazilian legislations, the sorbate and benzoate mass fractions found were, in general, higher than those reported in other studies. In Austrian mustards, the mean mass fraction of sorbic acid was 118.9 mg/kg and of benzoic acid 195.5 mg/kg, while in ketchup and tomato puree, the mean mass fraction of sorbic and benzoic acids were 486.0 and 788.7 mg/kg, respectively (32). Low mass fractions of sorbic and benzoic acids were also reported in Chinese ketchup at 175.9 and 307.6 mg/kg, respectively (13). In Iranian ketchup, sorbate and benzoate are not allowed, but Javanmardi et al. (12) found benzoate mass fractions of up to 38.7 mg/kg, however, none of them were detected in tomato paste (2). In contrast, 23.8% samples of Turkish ketchup showed a sum of sorbic and benzoic acids >1000 mg/kg (33).

Our findings reinforce the importance of the constant monitoring of sorbate and benzoate mass fractions in these products, especially ketchup, associated with the labeling inspection since children and young people are frequent consumers of these products, and their excessive intake can represent health risks, especially for these consumers (7). In addition, the fact that the sorbate and benzoate mass fractions found in this study were, in general, higher than those reported in other works lights up a warning signal about the possible excessive intake of these preservatives by Brazilian consumers since they are present in many different food products. Therefore, the constant monitoring and the use of allowed sorbate and benzoate mass fractions in mustard, ketchup and tomato sauce are indispensable to ensure the lower risk to the health of their consumers.