ANALYTICAL SCIENCES MARCH 2008, VOL.

24 2008 The Japan Society for Analytical Chemistry

365

Application of Factorial Design and Doehlert Matrix for Determination of Trace Lead in Environmental Samples by On-line Column Preconcentration FAAS Using Silica Gel Chemically Modified with Niobium(V) Oxide
Kalya Cravo Di Pietro ROUX, Heloisa Frana MALTEZ, Jeferson Schneider CARLETTO, Edmar MARTENDAL, and Eduardo CARASEK
Departamento de Qumica, Universidade Federal de Santa Catarina, Florianpolis, 88040-900, Brazil

In this study a new method for Pb determination in water using solid phase extraction coupled to a flow injection system and flame atomic absorption spectrometry was developed. The sorbent used for Pb preconcentration and extraction was silica gel chemically modified with niobium(V) oxide. Flow and chemical variables of the system were optimized through a multivariate procedure. The factors selected were buffer type, eluent concentration, and sample and eluent flow rates. It was verified that the aforementioned factors as well as their interactions were statistically significant at the 95% confidence level. The effect of foreign ions was evaluated using a fractionary factorial experimental design. The detection limit was 0.35 g L1 and the precision was 1.6%. Results for recovery tests using different environmental samples were between 90 and 104%. Certified reference materials were analyzed in order to check the accuracy of the proposed method. (Received January 16, 2007; Accepted May 29, 2007; Published March 10, 2008)

Introduction
Lead (Pb) is one of the most common environmental contaminants due to its wide use in petroleum, mining, paint and pigments, ceramics and weapons industries. This metal is commonly found in many places where humans may be exposed to it. It is considered an element that has toxic effects on human and animals and has no physiological function in living organisms.1,2 The harmful effects of Pb have been known for a long time. Indeed, this metal affects practically all vital organs and systems of the human body.3 Since heavy metals are in trace levels in many parts of the environment, suitable analytical methods are required for In Brazil, the National quantitative determinations.4 Environment Council (Conselho Nacional do Meio Ambiente, CONAMA) regulates the maximum levels of contaminants permitted in environmental samples. The maximum Pb concentration allowed in treated water and in seawater where fishing is carried out is 10.0 g L1.5 Atomic absorption spectrometry (AAS) is a technique that offers simple, fast, precise and low cost analysis.6 However, flame atomic absorption spectrometry (FAAS) has limitations in terms of detectability for low concentrations, such as those required by CONAMA. Furthermore, the complexity of environmental samples is a factor which makes the quantitation of analytes difficult. For these reasons, extraction and preconcentration techniques have been employed. Solid phase extraction,6 liquidliquid extraction,7,8 coprecipitation,9,10 and cloud point extraction11 are commonly applied sample
To whom correspondence should be addressed. E-mail: carasek@qmc.ufsc.br

preparation techniques. Solid phase extraction (SPE) is widely employed due to the low generation of residues, high enrichment factor, easy accommodation of the solid phase in a minicolumn coupled to a flow preconcentration system and the fact that it, generally, does not require the use of toxic solvents.12 Pyrzynska and Cheregi13 proposed a methodology using SPE for lead determination in natural waters after preconcentration on the sorbent Cellex P and reported a detection limit of 1.8 g1. Goswami and Singh14 used 1,8-dihydroxyanthraquinone immobilized on silica gel for the preconcentration and determination of lead by FAAS in tap and river water samples and achieved a detection limit of 0.45 g L1. Other sorbent materials have also been used for solid phase extraction, such as 2-aminothiazole modified silica gel,15 Chromosorb-102,16 Amberlite XAD-2 and Amberlite XAD-7.17 In analytical chemistry, methods proposed for the determination of contaminants in environmental samples are usually optimized using a univariate approach where each variable is optimized individually. This requires a great number of experiments and considerable amounts of reagents and time, and does not take into account the interaction between variables.18 To overcome these disadvantages, researches have used procedures involving multivariate techniques.1820 Among the multivariate procedures available, factorial experimental design has been used for the evaluation of the interaction effects of the selected factors on the analytical response. The factors that have been shown to be statistically significant can then be optimized through a response surface methodology. The Doehlert matrix is a non-factorial response surface methodology, in which two or more factors can be simultaneously optimized. The response is fitted to a secondorder polynomial, and the optimum values for the factors can be

366
estimated through this mathematical relationship.21 The factorial design and Doehlert matrix have been used to optimize preconcentration procedures for the determination of nickel in food samples22 and lead in river waters23 using, respectively, FAAS and inductively coupled plasma optical emission spectrometry (ICP OES) as detection techniques. In this paper, a two-level full factorial experimental design and Doehlert matrix were the multivariate tools used for optimization of the factors affecting the preconcentration system for Pb determination in aqueous samples using FAAS. This methodology was based on sorption of Pb(II) on the sorbent silica gel chemically modified with niobium(V) oxide.

ANALYTICAL SCIENCES MARCH 2008, VOL. 24

Experimental
Instrumentation A Varian SpectrAA 50 (Victria, Australia) flame atomic absorption spectrometer equipped with a deuterium lamp as a corrector for non-atomic absorption was used. The lead hollow cathode lamp was manufactured by Hitachi (Mitorika, Ibaraki, Japan). Operational parameters of the spectrometer were: wavelength, 217 nm; lamp current, 5 mA; slit width, 1 nm. The flame consisted of an oxidizing airacetylene flame. The acetylene was atomic absorption-grade (White Martins, So Paulo, Brazil). A Mettler Toledo 320 pH meter was used to set the sample and working solutions at the desired pH. Two peristaltic pumps manufactured by Ismatec, MS-Reglo (Glattbrugg, Switzerland) with eight channels fitted with Tygon tubes were used to pump the solutions through the flow system. Reagents and solutions The working solutions were prepared with deionized water taken from a Milli-Q purification system (Millipore, Bedford, MA, USA). All reagents were analytical grade. Nitric acid 65% (v/v) (Merck, Darmstadt, Germany) was bi-distilled below its boiling point in a quartz distiller (Krner Analysentechnik, Rosenheim, Germany). The laboratory glassware was cleaned as follows: immersion in a 2% (v/v) Extran (Merck) solution for 24 h; washing thoroughly with distilled water; immersion for 48 h in a 20% (v/v) nitric acid solution (Vetec, Rio de Janeiro, Brazil); ultrasound bath for 1 h and finally washing with deionized water. Sodium citrate buffer solution was prepared by weighing 2.94 g of sodium citrate and diluting with deionized water. The pH was adjusted to 7 with 0.1 mol L1 nitric acid and the final volume was made up to 100 mL. Tris(hydroxymethyl)aminomethane (HOCH2)3CNH2 0.1 mol L1 was prepared by weighing 1.2 g and diluting with water. The pH 7 was reached with 0.1 mol L1 nitric acid and the final volume made up to 100 mL. Lead standard solutions were prepared from a 1000 mg L1 stock solution (SPEX, Edison, NJ, USA). Silica gel chemically modified with niobium(V) oxide (Nb2O5-SiO2) was synthesized as described in the literature.24,25 To evaluate the effect of foreign ions on Pb sorption, we tested the following ions: Na+, Ca2+, K+, Mg2+, Zn2+, Cu2+ and Fe3+. Na+ was obtained as NaCl from Reagen (Rio de Janeiro, Brazil); Ca2+ as CaCO3 and Mg2+ as MgSO4 from Vetec (Rio de Janeiro, Brazil); K+ as KCl from Isofar (Rio de Janeiro, Brazil); and standard solutions of Zn2+, Cu2+ and Fe3+ from SPEX (Edison).

Fig. 1 Diagram of the on-line preconcentration system used in this study. (A) Adsorption process and (B) desorption process. V, Valve; L, open; D, closed; MC, minicolumn containing adsorbent; R, sample or eluent back stream; hatched circle, valve on; white circle, valve off.

On-line preconcentration system The on-line flow system used for the development of the proposed method is illustrated in Fig. 1. The flow system consists of two peristaltic pumps equipped with Tygon tubes, four three-way solenoid valves and a minicolumn filled with 100 mg of the sorbent Nb2O5-SiO2. The flow system was coupled to the FAAS. During preconcentration (A), valve V1 is open and the other valves remain closed; the sample or working solutions are pumped through the minicolumn and the effluent is discarded. After this first step, water is pumped for 5 s in the same way as in the preconcentration procedure to eliminate any sample matrix that might be present in the dead volume of the minicolumn. In the elution step (B), V1 is closed and valves V2, V3 and V4 are open. Thus, the eluent percolates through the minicolumn in the opposite direction to that of the preconcentration step. The eluate is carried directly to the nebulization system of the FAAS.18,25 Optimization strategy for the proposed system The optimization of the parameters affecting Pb sorption by the sorbent Nb2O5-SiO2 was performed using a two-level full factorial experimental design involving four factors and a final optimization using a Doehlert matrix design. All experiments were carried out in duplicate, using 15.0 mL of a 100.0 g L1 Pb solution. The four variables were: buffer type, eluent concentration, sample flow rate and eluent flow rate. The analytical response was taken as peak area. The experimental data were processed using the Statistica 6.0 computer program. Environmental samples and certified reference materials Quantification of Pb was carried out using an external calibration procedure. The analytical curve was obtained in the range of 5.0 to 120.0 g L1, submitting each solution to the optimized preconcentration procedure. A calibration curve with concentrations from 1.0 to 7.0 mg L1 prepared in 0.5 mol L1 HNO3 (the same concentration as the

ANALYTICAL SCIENCES MARCH 2008, VOL. 24

eluent) was constructed without preconcentration for estimation of the enrichment factor. Water samples of acid waste material from a coal mine (Cricima-SC, Brazil) were collected in suitable polypropylene flasks and kept at 4C until analysis. Seawater and well water samples were collected in Florianpolis-SC, Brazil. The following certified reference materials were used for validation of the proposed methodology: NIST 1643e (freshwater) from the National Institute of Standards & Technology (Gaithersburg, USA) and BCR 397 (human hair) from the Community Bureau of Reference, Brussels, Belgium. For the environmental samples, recovery tests were performed. For the preparation of the certified reference material (hair), samples of approximately 250 mg were weighed directly in PTFE flasks. Subsequently, 3.0 mL of HNO3 and 1.0 mL of H2O2 were added to each sample in the PTFE flasks. An MLS1200 microwave (Milestone, Sorisole, Italy) was used for microwave-assisted acid-digestion with the following program: 2 min at 250 W, 2 min at 0 W, 6 min at 250 W, 5 min at 400 W and 5 min at 650 W, followed by 5 min of ventilation. The digested samples were transferred to 50 mL polypropylene tubes (Sarstedt, Nqmbrecht, Germany) and kept under refrigeration until use.

367
significant difference was observed. Thus, pH 7 was selected, since the pH for natural water samples is usually around 7. The sample volume used for the optimization was 15.0 mL with a 100.0 g L1 Pb solution. Factorial design The factors considered for the preconcentration procedure were: buffer type, eluent concentration (HNO3), sample flow rate and eluent flow rate. A total of 16 experiments were carried out, in duplicate, to determine the significance of each variable, as well as the influences of their interactions on the preconcentration system. Table 1 shows the minimum and maximum levels obtained for the selected factors. The maximum and minimum conditions for each variable were selected based on other studies using the same sorbent material.18,25 The combination used for each experiment and the analytical response obtained (as integrated absorbance) are summarized in Table 2. A Pareto chart (Fig. 2) was plotted to check the influence of

Results and Discussion

Preliminary tests were performed to investigate Pb(II) adsorption on the sorbent Nb2O5-SiO2 under different pH conditions. For a pH lower than 5, the amount of Pb(II) adsorbed was extremely low, while between 5 to 9 no

Table 1

Factors and levels used in the factorial design Factor Minimum () Sodium citrate 3.3 0.5 3.3 Maximum (+) TRISa 6.3 2.5 5.3

Buffer type Sample flow rate/mL min1 Eluent concentration/mol L1 Eluent flow rate/mL min1

a. Tris(hydroxymethyl)aminomethane.

Fig. 2 Pareto chart of standardized effects for the variables of the preconcentration system using integrated absorbance as analytical response.

Table 2 Experiment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Matrix of full factorial design and the analytical response for each experiment for preconcentration of Pb(II) Sample flow rate/ mL min1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Eluent flow rate/ mL min1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Eluent concentration/ mol L1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Integrated absorbance Buffer type Replicate 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.373 0.419 0.378 0.323 0.475 0.456 0.398 0.381 0.196 0.188 0.149 0.126 0.146 0.126 0.084 0.068 Replicate 2 0.414 0.415 0.378 0.356 0.477 0.47 0.384 0.381 0.18 0.195 0.156 0.11 0.118 0.135 0.107 0.044

368
Table 3 Doehlert matrix for the final optimization of sample and eluent flow rates Experiment 1 2 3 4 5 6 7 Eluent flow rate/ mL min1 4.3 3.3 3.3 4.3 5.3 5.3 4.3 Sample flow rate/ mL min1 3.3 4.1 5.6 6.3 5.6 4.1 4.8 Integrated absorbancea/s 0.239 0.258 0.246 0.246 0.257 0.273 0.267

ANALYTICAL SCIENCES MARCH 2008, VOL. 24

Table 4 Levels and factors used for evaluation of interference of selected ions through fractionary factorial experimental design Factor Ca2+ Mg2+ Na+ K+ Zn2+ Cu2+ Fe3+ Minimum ()/ mg L1 0 0 0 0 0 0 0 Center point/ mg L1 2.5 1.5 2.5 1.5 0.5 0.5 0.5 Maximum (+)/ mg L1 5.0 3.0 5.0 3.0 1.0 1.0 1.0

a. Average of two replicates.

matrix is shown in Fig. 3. The analytical response was taken as integrated absorbance. The response surface can be described by a quadratic equation (Eq. (1)). Abs = 0.100 + 0.115flowsample + 0.038floweluent 0.011flow2sample 0.001flowsamplefloweluent 0.003flow2eluent

(1)

Abs/flowsample = 0.115 0.022 flowsample 0.001floweluent (2) Abs/floweluent = 0.038 0.001flowsample 0.006floweluent
(3)

Fig. 3 Response surface for optimization of sample and eluent flow rates: sample volume, 15.0 mL; sample concentration, 100.0 g L1; sample pH, 7.0; sorbent mass, 100.0 mg; eluent concentration, 0.5 mol L1 HNO3; citrate buffer, 0.1 mol L1.

The critical point of the quadratic function is obtained by making the partial derivative equations (Eqs. (2) and (3)) equal to zero. The application of the Lagrange criterion19,20 indicates whether this critical point is a maximum, minimum or a saddle point. The critical points obtained were 5.5 and 5.0 mL min 1 for eluent and sample flow rates, respectively. The application of Lagrange criterion indicated that the critical point is the maximum point of the response surface. Thus, as a result of all the optimizations, the following working conditions were selected: citrate buffer, 0.5 mol L1 HNO3 as the eluent, and sample and eluent flow rates of 5.0 and 5.5 mL min1, respectively. Effect of foreign ions on the Pb preconcentration The effects of Ca2+, Mg2+, Na+, K+, Zn2+, Cu2+ and Fe3+ ions were assessed to check whether these ions interfere with the adsorption of 15.0 mL of a solution containing 100.0 g L1 Pb, under the optimized conditions. Silica gel chemically modified with niobium(V) oxide is characterized as being a cationic exchanger. Therefore, competition between concomitant cations and the analyte will be dependent on their relative concentrations, the relative affinity for the sorbent surface, and the number of available active sites at the sorbent surface. Usually, the effect of concomitants is determined using a univariate approach, adding to the working solution a known concentration of only one cation at a time. In this way, the effect of each cation is evaluated individually. This type of study is tedious and time-consuming and is not representative of a real situation, in which a sample may contain several ions at the same time. For this reason, in this study, interference was investigated using a 273 fractionary factorial experimental design including a central point, resulting in 17 experiments. Table 4 summarizes the levels for each ion considered as a factor for this study and Table 5 shows the matrix of the experimental design and the analytical response, obtained using integrated absorbance. Figure 4 shows the main and 2-way interaction effects of the factors on the analytical response. As can be observed, none of

the factors and their interactions in the system using analysis of variance (ANOVA) and p-values (p > 0.05). As can be seen from the Pareto chart, the factors buffer type, sample flow rate, eluent flow rate and the interaction among them are statistically significant at the 95% confidence level. The very high negative value obtained for the effect of buffer type (51.90) indicates that the analytical response increases considerably when citrate buffer is used. Thus, citrate buffer was used for the subsequent experiments. The negative values for sample (11.44) and eluent (2.63) flow rates indicate an improvement in the analytical response when the level changes from maximum to minimum. Eluent concentration was not statistically significant and, therefore, its concentration was fixed at 0.5 mol L1. Final optimization using Doehlert design It was observed from the results of the factorial design that the factors sample flow rate and eluent flow rate required a final optimization. Thus, a Doehlert matrix involving these variables was constructed. Seven experiments were carried out in duplicate, and the levels and responses are summarized in Table 3. The surface obtained through application of the Doehlert

ANALYTICAL SCIENCES MARCH 2008, VOL. 24

Table 5 Fractionary factorial design matrix used to check the influence of selected ions on analytical response Experiment Ca2+ Mg2+ Na+ K+ Zn2+ Cu2+ Fe3+ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 Integrated absorbance 0.5113 0.4947 0.4989 0.5063 0.5067 0.5085 0.4914 0.4994 0.5050 0.5185 0.5094 0.5079 0.5201 0.5200 0.5264 0.5255 0.5241 Table 6 Relevant analytical data 5.0 120.0 0.9982 1.6 0.00411 0.35 1.20 46.4 12

369

Linear rangea/g L1 Correlation coefficient (R) Relative standard deviation (RSD, %) (100 g L1, n = 7) Sensitivity ( )/L g1 Limit of detection (LOD)/g L1 Limit of quantification (LOQ)/g L1 Enrichment factor (EF) Analytical frequency/samples h1 a. Studied linear range.

Table 7 Pb(II) concentration obtained for water samples during the recovery tests Sample Mine water Added/g L1 0 30.0 40.0 0 5.0 10.0 0 5.0 10.0 Founda/g L1 ND 28.7 1.2 41.5 2.0 ND 4.5 1.0 9.6 1.1 ND 4.6 0.7 9.9 0.9 Recovery, % 96.0 104.0 90.0 96.4 92.0 99.0

Sample volume, 15.0 mL; sample concentration, 100.0 g L1 Pb.

Seawater

Well water

ND = not detected. a. Average sd, n = 3.

Table 8 Concentrations obtained for Pb(II) in certified reference materials analyzed using the preconcentration procedure here developed Sample NIST 1643e BCR397 (human hair) Certified/g L1 19.63 0.21 33.0 1.2 Found/g L1 18.71 0.91 30.2 2.0

n = 5. Student t-test for 95% confidence level.

Fig. 4 Pareto chart of standardized effects of foreign ions on lead adsorption in the proposed preconcentration system.

these main and interaction effects is significant with regard to the analytical response at a 95% confidence level. In other words, none of the selected ions interfere with lead adsorption at these concentration levels. Effect of sample volume Using the optimized conditions, a study to determine the sample volume to be preconcentrated was performed. The analyte mass was maintained constant and the sample volume was varied and, therefore, the analyte concentration changed. On increasing the sample volume, it was observed that the analytical signal was almost constant over the volume range studied (20 to 120 mL). Thus, the detection of very low concentrations of the analyte can be improved by increasing the sample volume. However, this procedure can result in some loss of sample throughput. As a compromise between analytical frequency and sensitivity, 20 mL of sample volume was selected for subsequent studies.

Analytical features The analytical data of interest for the method are summarized in Table 6. The method precision was determined by submitting 20 mL of a 100 g L1 sample solution to the optimized procedure. Very good precision (RSD < 2%) was obtained. An analytical curve was obtained to assess sensitivity, linear range, enrichment factor, detection and quantitation limits. The detection limit was calculated as three times the standard deviation of a blank solution divided by the slope. The enrichment factor was obtained through the ratio between the slopes of analytical curves with and without preconcentration. Analysis of samples and method validation The proposed flow system was used for Pb determination in water samples from a coal mine located in Cricima, Brazil, as well as in seawater and well water, both from Florianpolis, Brazil. Recovery tests were performed and two certified reference materials were analyzed in order to check the accuracy of the method. The results are given in Tables 7 and 8, respectively. Good recoveries between 90 104% were obtained for the

370
environmental samples, demonstrating that the method can be successfully applied to these samples. For the certified reference materials, there was good agreement between found and certified values.

ANALYTICAL SCIENCES MARCH 2008, VOL. 24

res35705.pdf. 6. D. Budziak, E. L. da Silva, S. D. de Campos, and E. Carasek, Microchim. Acta, 2003, 141, 169. 7. H. Chen, J. Jin, and Y. Wang, Anal. Chim. Acta, 1997, 353, 181. 8. Y. K. Agrawal and K. R. Sharma, Talanta, 2005, 67, 112. 9. M. Soylak and N. D. Erdogan, J. Hazard. Mater. B, 2006, 137, 1035. 10. M. Soylak, S. Saracoglu, U. Divrikli, and L. Elci, Talanta, 2005, 66, 1098. 11. J. Cheng and K. C. Teo, Anal. Chim. Acta, 2001, 450, 215. 12. V. Camel, Spectrochim. Acta, Part B, 2003, 58, 1177. 13. K. Pyrzynska and M. Cheregi, Water Res., 2000, 34, 4215. 14. A. Goswami and A. K. Singh, Talanta, 2002, 58, 669. 15. I. L. de Alcntara, P. S. Roldan, M. A. L. Margionte, G. R. Castro, C. C. F. Padilha, A. O. Florentino, and P. M. Padilha, J. Braz. Chem. Soc., 2004, 15, 366. 16. S. Saracoglu and L. Elci, Anal. Chim. Acta, 2002, 452, 77. 17. P. K. Tewari and A. K. Singh, Talanta, 2002, 56, 735. 18. R. L. Dutra, H. F. Maltez, and E. Carasek, Talanta, 2006, 69, 488. 19. S. L. C. Ferreira, H. C. dos Santos, M. S. Fernandes, and M. S. de Carvalho, J. Anal. At. Spectrom., 2002, 17, 115. 20. M. Zougagh, P. C. Rudner, A. G. de Torres, and M. C. Pavn, J. Anal. At. Spectrom., 2000, 15, 1589. 21. S. L. C. Ferreira, W. N. L. dos Santos, C. M. Quintella, B. B. Neto, and J. M. Bosque-Sendra, Talanta, 2004, 63, 1061. 22. V. A. Lemos, P. X. Baliza, J. S. Santos, L. S. Nunes, A. A. de Jesus, and M. E Rocha, Talanta, 2005, 66, 174. 23. M. S. Bispo, E. S. da Boa Morte, M. G. A. Korn, L. S. G. Teixeira, M. Korn, and A. C. S. Costa, Spectrochim. Acta, Part B, 2005, 60, 653. 24. S. Denofre, Y. Gushikem, S. S. Castro, and Y. Kawano, J. Chem. Soc. Faraday Trans., 1993, 89, 1957. 25. E. L. Silva, E. M. Ganzarolli, and E. Carasek, Talanta, 2004, 62, 727.

Conclusions
In this paper, we describe a new method for Pb determination at trace levels using solid phase extraction for the sample preparation and silica gel chemically modified with niobium(V) oxide as the sorbent. This sorbent presented excellent stability, having a longer lifetime than that necessary to carry out this study. The use of a multivariate procedure for optimization of the parameters affecting the system, as well as the interference study, was found to be very efficient, requiring a reduced number of experiments. The instrumentation and method described provided a fast and convenient way to preconcentrate and separate Pb(II) from aqueous samples in a simple analytical cycle.

Acknowledgements
The authors thank Conselho Nacional de Desenvolvimento Cientifico e Tecnolgico (CNPq) for financial support.

References
1. S. E. Manahan, Environmental Chemistry, 6th ed., 1994, CRC Press, Boca Raton. 2. D. A. Gidlow, Occup. Med., 2004, 54, 76. 3. L. H. Sann, T. G. Cosso, I. Romieu, and M. H. Avila, Salud Pblica Mex, 1998, 40, 359. 4. A. Ceccarini, I. Cecchini, and R. Fuoco, Microchem. J., 2005, 79, 21. 5. CONAMA: http://www.mma.gov.br/port/conama/res/res05/