Voltammetric determination of food colorants using a polyallylamine modified tubular electrode in a multicommutated flow system

Talanta 72 (2007) 282–288

Voltammetric determination of food colorants using a polyallylamine modified tubular electrode in a multicommutated flow system M. Lu´ısa S. Silva a , M. Beatriz Q. Garcia a,∗ , Jos´e L.F.C. Lima a , E. Barrado b a

Requimte, Departamento de Qu´ımica-F´ısica, Faculdade de Farm´acia da Universidade do Porto, Rua An´ıbal Cunha, 164, 4099-030 Porto, Portugal Departamento de Qu´ımica Anal´ıtica, Facultad de Ciencias de la Universidad de Valladolid, Prado de la Magdalena s/n, 47005 Valladolid, Spain

b

Received 31 July 2006; received in revised form 17 October 2006; accepted 19 October 2006 Available online 28 November 2006

Abstract This work describes the construction of a polyallylamine modified tubular glassy carbon electrode and its application in the electroreduction of food azo colorants (tartrazine, sunset yellow and allura red) by square wave voltammetry. The electrode modification prevented the surface fouling and, simultaneously, enhanced the analytical signal intensity. The developed unit was coupled to a multicommutated flow system which, given the complexity of samples, was designed to allow the implementation of the standard additions method in an automatic way, using only one standard solution. The described method presented a linear range up to about 2.0 × 10−4 mol l−1 for the referred colorants, with a detection limit of 1.8 × 10−6 mol l−1 for tartrazine, 3.5 × 10−6 mol l−1 for sunset yellow and 1.4 × 10−6 mol l−1 for allura red. The method was applied in the analysis of these colorants in several food samples, and no statistically significant difference between the results obtained by the proposed and the comparative method (HPLC) was found, at a 95% confidence level. Repeatability in the analysis of samples (expressed in R.S.D.) was about 3% (n = 10). © 2006 Elsevier B.V. All rights reserved. Keywords: Modified tubular electrode; Voltammetry; Multicommutation; Polyallylamine; Tartrazine; Sunset yellow; Allura red; Food colorants

1. Introduction The use of colorants as food additives has been exploited by food industry with the aim of enhancing the aesthetic appeal of foodstuffs to the consumer. Azo colorants, such as tartrazine (E102), sunset yellow (E110) and allura red (E129), constitute one of the major synthetic colorant groups, used commercially in food, drinks, medicines and cosmetics. Its vast application is due to an inexpensive production and to a large colour spectrum that can be obtained, when compared with natural colorants [1]. Some colorants can trigger adverse effects, namely tartrazine, which can cause the appearance of allergies and asthma [2] and childhood hyperactivity [3]. The colorants allowed to be used in food products and the authorized maximum levels are regulated by the Portuguese [4] and European [5] legislation. Quantification of colorants in food products prompted the need for the development of analytical methodologies, namely

∗

Corresponding author. Tel.: +351 222078966; fax: +351 222004427. E-mail address: [email protected] (M.B.Q. Garcia).

0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2006.10.032

spectrophotometric [6–12] and chromatographic methods [13]. Voltammetric methods were also reported [1,2,14–19], presenting high sensitivity, simplicity and low cost. However, all of them use mercury electrodes, whose toxicity justifies the search for alternative electrode materials. Besides, the majority of the referred methodologies use the standard additions method for sample analysis, which is carried out in a non-automatic way, increasing the slowness of measurements and demanding specialized operators. In this work, the construction, evaluation and application of a polyallylamine modified tubular glassy carbon electrode is proposed, as an alternative to the use of mercury electrode, for the determination of food colorants, which is based on the electroreduction of the azo dyes. For the first time, the modification of a tubular glassy carbon electrode with a polyelectrolyte coating is described, taking advantage of the ion-exchange and permselectivity characteristics of the polyallylamine film in relation to negatively charged species. The developed unit was coupled to a multicommutated flow system [20], which was designed to enable the implementation, in an automatic way, of the standard additions method, considering the complexity of samples.

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

2. Experimental 2.1. Reagents and solutions Reagents of p.a. quality were used, without having been subjected to any additional purification. In the preparation of solutions, water purified by the Millipore Milli Q system (conductivity <0.1 ␮S cm−1 ) was used. As supporting electrolyte and, simultaneously, carrier solution in the flow system, a hydrochloric acid solution 0.5 mol l−1 , prepared by dilution of concentrated HCl (Merck), was used. Stock solutions of tartrazine (Sigma), sunset yellow (Aldrich) and allura red (Aldrich) 1.0 × 10−3 mol l−1 were prepared by weighing and dissolution of the solid colorants in HCl 0.5 mol l−1 . Working solutions were prepared by dilution of the respective stock solutions in HCl 0.5 mol l−1 . For the working electrode surface coating, a polyallylamine hydrochloride solution 15.0 g l−1 , prepared by weighing and dissolution of the correspondent quantity of the polyelectrolyte (Aldrich), was used. Solid samples were dissolved in HCl 0.5 mol l−1 warmed to about 70 ◦ C, centrifuged for 10 min at 3000 rpm and filtered. Liquid samples were diluted in HCl 0.5 mol l−1 and filtered. 2.2. Equipment In the developed multicommutated flow system (Fig. 1A) solutions and samples were aspirated by an automatic burette (Crison model Micro BU 2031) equipped with a 10 ml syringe. To control the selection and direction of solutions and samples inside the manifold four 3-way solenoid valves (161 T031, NResearch) were used. A homemade power driver, based on an

Fig. 1. (A) Multicommutated flow system for food colorants determination: V1 , V2 , V3 and V4 , three-way solenoid valves; CS, carrier solution (HCl 0.5 mol l−1 ); SS, standard solution 1 × 10−4 mol l−1 ; S, sample; R, reactor; D, tubular voltammetric detector; AB, automatic burette equipped with a 10 ml syringe; W, waste. (B) Schematic representation of the tubular detector: Eref , reference electrode; Ew , polyallylamine modified working electrode; Eaux , auxiliary electrode.

283

integrated ULN 2003 circuit, was used to operate solenoid valves [21]. Control of the analytical system was made through an interface card (PC-LABCard model PCL-711B, Advantech) and a microcomputer. The software was developed in QuickBasic Version 4.5 (Microsoft) and allowed to control the functioning of the burette and the solenoid valves. Connection between the components of the flow system was made with Teflon tubes (Omnifit), of 0.8 mm inner diameter. Voltammetric measurements were carried out in an Autolab electrochemical system (Eco Chemie model PGSTAT 10) and data acquisition was accomplished through GPES software (Version 4.6). Scanning electron microscopy (SEM) micrographs were obtained using an electron microscope JEOL, model JSM-35C. To perform sample analysis by the comparative method, a chromatograph Varian, model 9012, with UV–Vis detector, model 9050, was used. The HPLC system was equipped with a column C18 (150 mm × 4.6 mm i.d. and particle size 5 ␮m) Waters Spherisorb ODS2. 2.3. Voltammetric detector with a polyallylamine modified tubular electrode Usually, in multicommutated flow systems, solutions are aspirated instead of being propelled, which simplifies the flow manifolds since, in this case, only one propulsion device is needed for the driving of all solutions. As a consequence, flow systems present an inner pressure lower than the atmospheric pressure, demanding that all manifold components, including the detector, are tightly fixed, in order to avoid air entrance. The construction of the voltammetric detector was based in a tubular detector with modified electrodes recently described [22], which demonstrated to have the required robustness to be used in flow systems in which solutions are aspirated. The detector, of tubular configuration (Fig. 1B), was constituted by a central Perspex support, which encased the working and auxiliary electrodes, both of glassy carbon, with 2.0 mm thickness and a central orifice of 0.8 mm diameter (this value was diminished after the surface coating with polyallylamine solution), being firmly fixed to the central support by two rubber disks, also perforated in the centre. A Metrohm (Ag/AgCl–KCl 3.0 mol l−1 , model 6.0727.000) electrode was used as the reference electrode, fixed by a threaded screw to the Perspex support. Electric contact with working and auxiliary electrodes was established through two metallic contacts threaded into the Perspex support. The tubular detector had an inner volume of 15.0 ␮l (volume between working and auxiliary electrodes). For cleaning and modification (coating of the active surface) of the working electrode, this was withdrawn from the tubular detector and was firstly polished, using a cotton thread soaked in alumina aqueous slurry of 0.075 ␮m and washed with deionised water. The surface coating, by droplet evaporation method, consisted in the deposition of 20 ␮l of polyallylamine hydrochloride solution 15.0 g l−1 directly into the central orifice of the electrode. It was kept at 70 ◦ C, and was overturned several times until the complete evaporation of the solvent (about 15 min), as a way to assure uniformity of the deposit on the cylindrical wall of the electrode.

284

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

The evaluation of the modified electrode behaviour showed that it was necessary to repeat the cleaning and coating process after 60 determinations carried out with the same polyallylamine film. 3. Results and discussion 3.1. Characterization and behaviour of the polyallylamine modified tubular electrode The electrochemical behaviour of synthetic azo dyes has been widely studied [23–26] and their reduction mechanism postulates a step-wise breakage of the molecule, in acidic media, involving four electrons and four protons. The behaviour of the polyallylamine modified tubular electrode was studied relatively to the reduction of tartrazine, sunset yellow and allura red, and the optimization of the parameters concerning the modification procedure was performed with the purpose of obtaining the maximum sensitivity and reproducibility of the measurements. All experiments were carried out with solutions of each colorant with a concentration of 1.0 × 10−4 mol l−1 , aspirating a volume of 150 ␮l, which was transported towards the detector by the carrier solution, with a flow rate of 0.96 ml min−1 . Voltammetric measurements were made in stopped flow, when the sample plug reached the detector and the analytical signal (cathodic peak current intensity—Ip ) was the highest. The effect of the supporting electrolyte composition, used in the preparation of solutions and as carrier solution in the flow system, was studied, being tested hydrochloric acid solution 0.5 mol l−1 , phosphoric acid and sodium di-hydrogen phosphate buffer solution 0.5 mol l−1 (pH 1.8), acetic acid and sodium acetate buffer solution 0.5 mol l−1 (pH 4.5) and sodium carbonate and sodium hydrogen carbonate buffer solution 0.5 mol l−1 (pH 9.9). Solutions of each colorant were prepared in the referred supporting electrolytes and analyzed. The results indicated that the Ip value increased with the decrease of the pH, so the hydrochloric acid solution was chosen for the following experiments. To evaluate the effect of the electrode modification on the Ip of the colorants, the electrode surface was coated with a polyallylamine solution with concentrations between 0.1 and 1.0 g l−1 and the obtained Ip values were compared. It was shown that the Ip value was higher for the coated electrode, when compared to the bare electrode, and the increase in Ip was in direct proportion to the polyallylamine concentration in the film. The enhancement in the Ip value was due to a selective preconcentration of the colorants by the polyallylamine film. In acidic conditions, this cationic polyelectrolyte becomes highly positively charged [27] binding easily to the anionic dyes via an ion-exchange process. The concentration of the polyallylamine solution used on the electrode coating, which influenced the film thickness, was optimized by coating the electrode surface with polyelectrolyte solution concentrations between 1.0 and 35.0 g l−1 , depositing on the electrode cavity a volume of 20 ␮l. For all colorants, Ip increased with increasing polyallylamine concentrations, up to

Fig. 2. SEM micrograph of the polyallylamine film on the tubular working electrode (polyallylamine solution 15 g l−1 , 20 ␮l), magnification 50×.

15.0 g l−1 , and diminished for higher concentrations. The initial increase of the analytical signal was due to an increase in the ionexchange sites between the polyelectrolyte and the colorants. However, for more concentrated polyallylamine solutions, the obtained films were thicker, which hindered the diffusion of the colorants towards the glassy carbon surface, decreasing Ip . The morphology of the film was examined in detail in SEM micrographs (Fig. 2), and it enabled to observe the uniformity and adherence of the film to the glassy carbon support. It was also possible to verify that the film thickness was about 6 ␮m. To evaluate the reproducibility of the modification procedure, which is reflected in the reproducibility of Ip , three electrodes were modified in 2 different days and used to analyse colorant solutions. It was shown that, for each modified electrode, Ip values obtained in different days presented a R.S.D. of about 3%, evidencing the reproducibility of the modification procedure. The stability of the modified electrode was studied by carrying out consecutive measurements of each colorant solution and monitoring the repeatability of Ip values, whose R.S.D. was used to check the film conditions. Ip values presented good repeatability up to 60 determinations, with a R.S.D. of about 2%, and with a clean baseline. Beyond that number of measurements, the analytical signal became less repetitive; therefore the electrode was withdrawn from the detector and submitted to a polishing to remove the film, followed by a new coating. The square wave voltammetry parameters were optimized, keeping constant the step potential and amplitude (2.5 and 50 mV, respectively) and shifting the frequency value between 10 and 80 Hz. Ip increased for all colorants with the increase of frequency, throughout the tested interval. It was chosen the value of 50 Hz, for which the best relation between Ip and peak width was obtained. Afterwards, step potential value was changed between 2 and 6 mV, and Ip increased with the increase

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

285

in step potential, stabilizing for values higher than 4 mV, being this value selected for the following experiments. Finally, amplitude values were varied between 20 and 100 mV. It was shown that Ip increased almost linearly with the increase in amplitude up to 80 mV, becoming constant beyond that value, which was chosen for the next trials. 3.2. Optimization of the multicommutated flow system parameters Considering the highly complex matrix of the samples intended to be analysed, the multicommutated flow system was designed to enable the implementation of standard additions method, in an automatic way and with continuous flow. In the automatic system based on the multicommutation concept (Fig. 1A), valves V1 and V2 allowed insertion of carrier solution (CS), standard solution (SS) or sample (S) depending on each valve status. Valves V3 and V4 were used to define two parallel analytical pathways, one of them including the reactor (R) and the detector (D) and the other one, meant to be used for sample exchange. The existence of the parallel pathway allowed to minimize the contact time between samples and working electrode, reducing its surface fouling and, additionally, it enabled a faster sample exchange, increasing the sampling rate. The automatic burette (AB) with the syringe was placed at the end of the analytical manifold, after the detector, and was used to aspirate all the solutions. The strategy used to insert sample and standard additions was based on binary sampling, which consisted in the alternate insertion of sample (V2 in position 1), carrier solution (V1 in position 1 and V2 in position 2) and/or standard solution (V1 in position 2 and V2 in position 2) plugs (Fig. 3). The sample volume aspirated to the system was controlled by flow rate and by sampling time, defined as the time interval during which valve V2 remained in position 1. By using binary sampling to insert the sample, the sampling time was defined as the sum of all time intervals during which valve V2 remained in position 1. Insertion by binary sampling allowed a better homogenization between sample, carrier and standard solution plugs. The developed automatic system required only one colorant standard solution to carry out standard additions, which was an advantage considering the time needed to prepare the solutions. The standard additions were performed by increasing the aspiration time of standard for each addition, keeping constant the sample aspiration time. The increase in the standard aspiration time did not change the sample dilution factor because, simultaneously, the carrier solution aspiration time decreased in the same proportion. With the developed system it was possible to perform a variable number of standard additions for each sample, simply requiring an adjustment in the aspiration times of standard and carrier solutions. Three flow system parameters were chosen for optimization, namely sample volume (V), reactor length (L) and flow rate (F), and all trials were carried out with a sunset yellow solution 1.0 × 10−4 mol l−1 , keeping constant all the parameters previously optimized. It was used only one colorant for the system

Fig. 3. Schematic representation of binary sampling for several standard additions.

optimization since the influence of physical parameters in the analytical signal was similar to the three colorants. The parameters were optimized by using a complete factorial design (3 factors and 2 levels). The output variable to optimize was the maximum Ip value, bearing in mind the concentration level predicted in food samples (lower than 1.0 × 10−4 mol l−1 ). Trials were performed in duplicate and in a random way to minimize the effect of uncontrollable factors. The factors levels were selected according to previous studies, which enabled to observe their effect in the analytical signal. Regarding the sample volume, the values of 50 and 150 ␮l were chosen for low (−) and high (+) levels, respectively. Volumes lower than 50 ␮l excessively diminished Ip and values higher than 150 ␮l would make the analytical cycle longer, affecting negatively the sampling rate. As for reactor length, the values of 30 and 75 cm were selected for (−) and (+) levels. The use of longer reactors would unnecessarily decrease Ip and sampling rate, without bringing benefits in the analytical signal repeatability, since binary sampling guaranteed a good homogenization of the plugs. Concerning flow rate, the values of 0.96 and 1.92 ml min−1 were chosen for (−) and (+) levels. It was shown that higher flow rates caused a decrease in Ip for successive determinations, which could be a consequence of film erosion, and flow rates lower than 0.96 ml min−1 diminished sampling rate. The factors levels, as well as the factorial design matrix and the obtained results, are shown in a resumed way in Table 1.

286

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

Table 1 Levels of the factors to be optimized, design matrix and ANOVA of the experimental results Factors

Levels

Injection volume, V (␮l) Reactor length, L (cm) Flow rate, F (ml min−1 )

Low (−) 50 30.0 0.96

Trial no.

High (+) 150 75.0 1.92

Design matrix

Response

V

L

F

V×L

V×F

L×F

V×L×F

Ip (×10−6 A) (n = 2)

1 2 3 4 5 6 7 8

− + − − − + + +

− − + − + − + +

− − − + + + − +

+ − − + − − + +

+ − + − − + − +

+ + − − + − − +

− + + + − − − +

4.868 14.930 3.446 4.266 3.078 14.225 13.425 11.790

Effect Fexp Fcrit

9.678 8122.2 7.571

−1.638 232.55

−0.828 59.396

−0.332 9.580

−0.342 10.165

−0.174 2.629

−0.291 7.337

Analysis of the results (using Yates algorithm) led to the conclusion that V was the factor that most significantly influenced Ip , given that the experimental F value (Fexp ) was much higher than the critical F value (Fcrit ). Ip was maximum for V in the (+) level, independently of the other factors levels. Despite the influence shown by L and F on Ip , it was not possible to deduce clearly, from the Table, which were the levels that generated the higher Ip . Thus, V was kept on (+) level (150 ␮l) and a supermodified simplex [28] was carried out to optimize L and F. As the flow rates allowed by the automatic burette are discrete values, whenever the trial flow rate was not one of the allowed flow rates, it was used the nearest flow rate possible, never with a difference higher than 0.11 ml min−1 between them. According to the obtained results with the supermodified simplex, the levels combination that enabled the highest analytical signal to be obtained was L = 32 cm and F = 1.2 ml min−1 . The analytical cycle was constituted by three major steps, namely sample exchange and manifold cleaning, insertion of sample, carrier and/or standard solutions, by binary sampling, and transport to the detector (Table 2). After optimization of all parameters, solutions of tartrazine, sunset yellow and allura red with concentrations between 1.0 × 10−5 and 3.0 × 10−4 mol l−1 were analyzed, and a linear correlation between colorant concentration and Ip up to 1.5 × 10−4 mol l−1 , for sunset yellow and allura red, and up to 2.0 × 10−4 mol l−1 for tartrazine, was obtained, occurring Table 2 Analytical cycle designed for colorants determination in foodstuffs

electrode saturation for higher concentrations. The detection limits, calculated from the regression equation (with yB = a and SB = Sy/x ) [29] were 1.8 × 10−6 mol l−1 for tartrazine, 3.5 × 10−6 mol l−1 for sunset yellow and 1.4 × 10−6 mol l−1 for allura red. The repeatability of measurements, expressed as R.S.D. of Ip , was evaluated through successive determinations of each colorant solution with a concentration of 1.0 × 10−4 mol l−1 , and the obtained R.S.D. was about 3% (n = 10). 3.3. Interference studies Considering the application of the developed method in the analysis of foodstuffs, the effect of several compounds normally present in these kinds of samples (gelatin, magnesium chloride, sodium benzoate, sodium citrate, ascorbic acid, citric acid, glucose and sucrose) on Ip was evaluated. Solutions containing tartrazine, sunset yellow or allura red, with a concentration of 1.0 × 10−4 mol l−1 , and the foreign compound in a higher concentration (maximum 100:1) were analyzed. The interfering concentration of each compound was considered as being that which caused a variation in Ip greater than or equal to ±5% in relation to the analytical signal obtained in its absence. In accordance with the obtained results, it was possible to conclude that none of the studied compounds interfered in the colorants determination, even when they were present in a 100-fold higher concentration. Gelatin, in contrast, caused a decrease in Ip when was present in a concentration higher than 1.0 g l−1 , which could be due to a blocking effect to the access of colorants to the electrode surface [30]. 3.4. Sample analysis To evaluate the applicability of the proposed method, several food samples containing tartrazine, sunset yellow or allura

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

287

The high precision with which binary sampling was performed, conferred by the use of solenoid valves, allowed to carry out a single standard addition for each sample, providing the same result in colorant concentration that was obtained with more standard additions. Repeatability of results obtained by the proposed method was evaluated by carrying out 10 successive determinations of all samples, and the R.S.D. value was never higher than 3%. 4. Conclusions Fig. 4. Voltammograms obtained in the analysis of an energetic drink, with the polyallylamine modified electrode. The several voltammograms correspond to the signal obtained for the blank (a), the sample (b) and 4 successive additions of a sunset yellow standard solution (c–f). Electrode modified with 20 ␮l of a 15 g l−1 polyallylamine solution, sample volume 150 ␮l, reactor length 32 cm, flow rate 1.2 ml min−1 , standard additions of sunset yellow 1 × 10−4 mol l−1 .

red were analyzed. For standard additions a 1.0 × 10−4 mol l−1 solution of the respective colorant was used. In the determinations performed by the comparative method (HPLC) [8,11], samples were dissolved or diluted in water, centrifuged and filtered. Fig. 4 shows the voltammograms obtained in the determination of sunset yellow in an energetic drink for dehydration. The several voltammograms correspond to the signal obtained for the blank, the sample and four successive additions of a sunset yellow standard solution. Table 3 shows the results obtained in the analysis of samples by the proposed and the comparative methods. Analysis of results enabled to conclude that all samples presented a colorant content within the authorized limits. The agreement between the results provided by the proposed and the comparative methods was evaluated through the Student t-test for paired samples, in which the t value (−1.70) was lower than the critical t value (2.31, two tail), for a 95% confidence level (n = 9). The regression line between the results obtained by the proposed and the comparative methods showed a correlation coefficient of 0.9996, with a = 1.35 ± 1.68 and b = 0.95 ± 0.02, for a 95% confidence level.

The developed method showed to be an advantageous alternative to the described methods for food colorants determination, providing similar results to those obtained with the chromatographic method, being comparatively faster, cheaper and less laborious. However, mixtures of colorants cannot be quantified by this approach due to peaks overlapping, since the three colorants have similar peak potentials (within the range from 0.15 to 0.30 V). The polyallylamine modified glassy carbon tubular electrode, here described for the first time, showed to be a valuable option as a substitute to the use of the dropping mercury electrode, for food colorants quantification. The polyelectrolyte film enabled, simultaneously, the protection of the electrode surface from contamination by sample matrix constituents and an enhancement in the analytical signal intensity, improving the measurements sensitivity. Although glassy carbon electrodes are not, commonly, employed in reduction reactions, the modified electrode presented good stability in the negative potential range in which the colorants reduction occurred. The tubular configuration of the electrode and the film stability allowed to combine it with a multicommutated flow system, facilitating the analysis of complex samples such as food samples, overcoming the frequent problems of electrode fouling and the necessary sample pretreatment before measurement. The use of the multicommutated flow system enabled automation of standard additions, performed in an easier and faster way, compared to the same procedure when it is carried out in batch conditions. The number of standard additions that could be performed was

Table 3 Results obtained in the determination of the colorants tartrazine, sunset yellow and allura red in foodstuffs, by the proposed and the comparative methods Food sample

Colorant

Reduction peak potential, Ep (V) −0.28

Gelatin powder (pineapple) Gelatin powder (peach) Juice powder (pineapple) Gelatin powder (tutti-frutti) Gelatin powder (banana)

Tartrazine

Energetic drink for dehydration I Energetic drink for dehydration II

Sunset yellow

−0.20

Gelatin powder (strawberry) Alcoholic drink

Allura red

−0.18

a b c

Average ± standard deviation of 3 determinations. Values in mg/kg. Values in mg/l.

Colorant concentrationa Proposed method ± ± ± ± ±

Comparative method (HPLC) ± ± ± ± ±

8b 7b 0.9b 1b 2b

+1.8 −3.7 +2.9 −4.9 −3.3

8.9 ± 0.6c 8.5 ± 0.1c

9.3 ± 0.1c 8.5 ± 0.1c

−4.3 0

68 ± 3b 11.9 ± 0.2c

70 ± 4b 11.7 ± 0.4c

−2.9 +1.7

56 79 39 136 87

3b

Relative deviation (%)

3b 1b 6b 1b

55 82 37.9 143 90

288

M.L.S. Silva et al. / Talanta 72 (2007) 282–288

changeable, requiring only an adjustment in the aspiration times of standard solution. Nevertheless, it was possible to carry out a single standard addition per sample, providing the same accuracy that was obtained when two or more standard additions were performed. Acknowledgements The authors thank to CRUP the financial support provided for the Portuguese–Spanish researcher exchange (Acc¸a˜ o Integrada E-41/03, HP2002-0039), to CEMUP—Centro de Materiais da Univ. Porto for providing the equipment in which the SEM analysis was performed and the Technology Department of Faculty of Pharmacy Univ. Porto for providing the HPLC equipment. M.L.S. Silva thanks FCT and FSE (III Quadro Comunit´ario de Apoio) for a Ph.D. grant (BD/10385/02). References [1] P.L. L´opez-de Alba, L. L´opez-Mart´ınez, L.M. De-Le´on-Rodr´ıguez, Electroanalysis 14 (2002) 197. [2] J.J.B. Nevado, J.R. Flores, M.J.V. Llerena, Fresenius J. Anal. Chem. 357 (1997) 989. [3] K.S. Rowe, K.J. Rowe, J. Pediatr. 125 (1994) 691. [4] Portaria no. 759/96 de 26-12-1996, Di´ario da Rep´ublica I S´erie B, 298, pp. 4647–4670. [5] European Parliament and Council Directive 94/36/EC of 30-06-1994, Official Journal of the European Communities No. L 237, pp. 13–29. [6] P.L. L´opez-de Alba, K. Wr´obel-Kaczmarczyk, K. Wr´obel, L. L´opezMart´ınez, J.A. Hern´andez, Anal. Chim. Acta 330 (1996) 19.

[7] P.L. L´opez-de Alba, L. L´opez-Mart´ınez, L.I. Michelini-Rodriguez, K. Wr´obel, K. Wr´obel, J.A. Hern´andez, Analyst 122 (1997) 1575. [8] J.J.B. Nevado, J.R. Flores, C.G. Cabanillas, M.J.V. Llerena, A.C. Salcedo, Talanta 46 (1998) 933. [9] J.J. Berzas, J.R. Flores, M.J.V. Llerena, N.R. Fari˜nas, Anal. Chim. Acta 391 (1999) 353. [10] E. Dinc¸, E. Baydan, M. Kanbur, F. Onur, Talanta 58 (2002) 579. [11] E.C. Vidotti, J.C. Cancino, C.C. Oliveira, M.C.E. Rollemberg, Anal. Sci. 21 (2005) 149. [12] E.C. Vidotti, M.C.E. Rollemberg, Quim. Nova 29 (2006) 230. [13] M.A. Prado, H.T. Godoy, Quim. Nova 27 (2004) 22. [14] A.G. Fogg, K.S. Yoo, Analyst 104 (1979) 723. [15] F.B. Dominguez, F.G. Diego, J.H. Mendez, Talanta 37 (1990) 655. [16] Y. Ni, J. Bai, L. Jin, Anal. Chim. Acta 329 (1996) 65. [17] Y. Ni, J. Bai, Talanta 44 (1997) 105. [18] J.J.B. Nevado, J.R. Flores, M.J.V. Llerena, Talanta 44 (1997) 467. [19] M.A. Kapor, H. Yamanaka, P.A. Carneiro, M.V.B. Zanoni, Eclet. Qu´ım. 26 (2001) 53. [20] B.F. Reis, M.F. Gin´e, E.A.G. Zagatto, J.L.F.C. Lima, R.A. Lapa, Anal. Chim. Acta 293 (1994) 129. [21] R.A.S. Lapa, J.L.F.C. Lima, B.F. Reis, J.L.M. Santos, E.A.G. Zagatto, Anal. Chim. Acta 351 (1997) 223. [22] M.L.S. Silva, M.B.Q. Garcia, J.L.F.C. Lima, E. Barrado, Anal. Chim. Acta 573–574 (2006) 383. [23] J.P. Hart, W.F. Smyth, Analyst 105 (1980) 929. [24] A.G. Fogg, D. Bhanot, Analyst 105 (1980) 868. [25] A.G. Fogg, D. Bhanot, Analyst 106 (1981) 883. [26] R.N. Goyal, M.S. Verma, N.K. Singhal, Croat. Chem. Acta 71 (1998) 715. [27] S.E. Burke, C.J. Barrett, Biomacromolecules 6 (2005) 1419. [28] M.W. Routh, P.A. Swartz, M.B. Denton, Anal. Chem. 49 (1977) 1422. [29] J.N. Miller, J.C. Miller, Statistics and Chemometrics for Analytical Chemistry, fourth ed., Pearson Education Limited, Harlow, 2000, pp. 120–123. [30] A.A. Barros, J.O. Cabral, Analyst 113 (1988) 853.