Experimental design for the study and optimization of the effect of different surfactants on the spectrophotometric determination of sulfide based on phenothiazine dye production

Abstract

In the present work the effect of three different surfactants Triton X-100, cetyltrimethylammonium bromide (CTAB) and sodium dodecylesulfate (SDS) on the spectrophotometric determination of sulfide based on phenothiazine dye production from phenyl-p-diamine (PPD) is studied. The main step for phenothiazine dye production is the oxidation of aromatic p-substituted amine to yield iminoquinone free-radical intermediary. Fe³⁺ in acidic media is used for oxidation purposes. Electron donor groups in the p-position of aromatic ring of amines are decisive for free-radical stabilization, leading to an increased dye production rate and the analytical sensitivity.

Preliminary studies showed that, the analytical signal was increased in the presence of SDS, CTAB and Triton X-100; but among the surfactants studied, SDS and Triton X-100 had a significant effect on the spectrophotometric determination of sulfide based on phenothiazine dye production. The effects of various experimental parameters in this reaction was investigated using central composite design. The experimental design was done at five levels of operating parameters. As a model surfactant, the concentration of SDS, amine, Fe³⁺ and H₂SO₄ were optimized simultaneously. The effects of Triton X-100 and CTAB as non-ionic and cationic surfactant, respectively, were investigated using a one variable at a time optimization method. The linear range for the determination of sulfide in the presence of SDS and Triton X-100 was 0.1–2 μg ml⁻¹ and that in presence of CTAB was 0.2–2.0 μg ml⁻¹. The proposed method was applied successfully for the determination of sulfide in different spiked real samples.

---

1. Introduction

The monitoring of sulfide is required in a variety of environmental and applications, including odor assessments, health and safety investigations, routine industrial/off-line monitoring programs.¹ Because of toxicity and capacity to remove dissolved oxygen, there are limits on the total level of sulfide permitted in waste discharges in most countries.^2,3 Sulfide in a H₂S form is toxic to fish and other aquatic organisms.⁴ Hence, its concentration needs to be controlled especially in water and wastewater. Sulfide can be determined in different media, using various techniques. These methods include titrimetric,⁵ flow-injection analysis,^6,7 amperometric,⁸ polarographic,⁹ ion chromatography,^10,11 HPLC,¹² GC¹³ and spectophotometric determination.^14,15 Although chromatographic methods have low detection limits, they generally require complex and time-consuming sample preparation procedures. In addition, these methods are expensive, and are not easily adapted to field analysis. Kinetic methods have been widely used in catalytic and non-catalytic determination of micro amounts of sulfide.^15–17 However, most of these methods suffer from the interfering effects of other sulfur anions or low linear dynamic range. Thus, simple, rapid, and selective methods are still required.

Spectrophotometric determination of sulfide based on phenothiazine dye production is based on the reaction of sulfide with acidic solutions of N,Ndimethylphenyl-p-diamine (DMPD) and Fe³⁺.^18–24 Ress et al. have developed a colorimetric method for the determination of trace amounts of sulfide in condensed steam.²⁵ For precise work the colour produced by N,N-diethyl-p-phenylenediamine in the presence of iron(III) ions is measured spectrophotometrically covering the range 0·5 to 100 μg of sulfide–sulfur; the standard deviations at the 100 and 1·0-μg levels are about 3 and 0·08 μg, respectively. They have shown that N,N-diethyl-p-phenylenediamine was superior to the dimethyl homologue.²⁵ Solid-phase extraction and spectrophotometric determination of hydrogen sulfide in air and water utilizing ethylene blue formation has been proposed by Vandana Singh et al.²⁶ Recently a systematic comparison of spectrophotometric flow injection methods for sulfide determination based on phenothiazine dye production from diverse aromatic p-substituted amines was performed. The behavior of N,N-dimethylphenyl-p-diamine (DMPD), N,N-diethylphenyl-p-diamine (DEPD), phenyl-p-diamine (PPD), p-aminophenol (PAP) and other three aromatic amines was investigated and the chemical parameters of the proposed flow methods were optimized²⁷

Our goal was enhancement of the sensitivity for the determination of sulfide by phenothiazine dye production method (Scheme 1) using different surfactants.

Scheme 1 p-substituted amine PPD (R = NH2) (1) and the reaction pattern for phenothiazine dye (2) production.²⁷

The effects of three different surfactants (non-ionic, cationic and anionic) on the enhancement of this reaction were studied in this work. Central composite design (CCD) and response surface methodology (RSM) were developed as experimental strategies for modeling and optimization of the influence of some variables on the performance of spectrophotometric determination of sulfide from aqueous solution.

The optimization of the influencing parameters using a “one variable-at-a-time” method requires many experiments. Furthermore, there are may be interactions between the investigated parameters, which cannot be easily studied by the one variable at a time method.

On the other hand, the total number of required experiments can be reduced using experimental design techniques. It is essential that any experimental design methodology be very economical for extracting the maximum amount of complex information, so saving significant experimental time as well as material used for analyses and personal costs.²⁸ Such designs have been applied to the optimization of different process.^29–34

Central composite designs are of three types. Circumscribed designs consist of cube points at the corners of a unit cube that is the product of the intervals [−1, 1], star points along the axes at, or outside the cube, and center points at the origin. Inscribed designs are as described above, but scaled so that the star points take the values −1 and +1, and the cube points lie in the interior of the cube. Faced designs have the star points on the faces of the cube. Faced designs have three levels per factor, in contrast with the other types, which have five levels per factor.³⁵ In this work, concentration of sulfuric acid, amine, SDS (as a case surfactant), and Fe³⁺ solution were optimized simultaneously using Circumscribed CCD. The influence of each of the factors and their interactions could be clearly identified for the responses screened. The use of experimental design, as mentioned, enabled a subsequent benefit in terms of labor time and number of experiments to optimize the conditions. The main purpose of this study was to perform the central composite design as a novel procedure in order to investigate the effect of surfactant on the spectrophotometric determination of sulfide based on phenothiazine dye production using phenyl-p-diamine (PPD).

2. Experimental

2.1. Reagents and solutions

All chemicals used were of analytical grade and the solutions were prepared with freshly purified water. Acidic solutions of 1.00 g L⁻¹ PPD (Merck), was prepared in (0.1–1 mol L⁻¹) H₂SO₄. Acidic Fe³⁺ solutions (0.1 mol L⁻¹) were prepared in H₂SO₄ using FeCl₃·6H2O (Sigma-Aldrich). Na₂S salt was used to prepare the stock solution of sulfide (100 mg L⁻¹). This solution was prepared daily in 1 mol L⁻¹NaOH to prevent sulfide pour out. Working sulfide reference solutions were also prepared in 1 mol L⁻¹ NaOH by proper dilutions of the stock solution.

2.2. Instruments

The spectrophotometric measurements were made with a PG mode T80 UV-Vis double-beam spectrophotometer (Japan) utilizing a 1-cm quartz cell.

2.3. Statistical software

Essential Regression and Experimental Design for Chemists and Engineers (EREGRESS), as a MS Excel Add-In software,^35,36 was used to design the experiments and to model and analyze the results.

2.4. Procedures

An aliquot of the sulfide solution (final concentration of sulfide in the range 0.1–2.0 μg mL⁻¹), 1 mL of 0.2 mol L⁻¹ PPD reagent, 1 ml of 1 mol L⁻¹ sulfuric acid, and 2 mL of 0.2 mol L⁻¹ Fe³⁺ solution was transferred into a 10 mL volumetric flask. 1 ml of surfactant (10% v/v Triton X-100, 10% w/v SDS or 5% w/v CTAB) was added and made up to the mark with double distilled water. The solution was allowed to stand for 10 min and transferred into a 3 mL quartz cell to measure its absorbance at 590 nm against a reagent blank as the reference. The blank solution was prepared as sample solution except that it had no sulfide.

2.5. Central composite design

The experimental conditions were optimized using central composite design (CCD). Four independent factors, namely the PPD (F1), Fe⁺³ (F2), H₂SO₄ (F3) and SDS concentration (F4) were studied at five levels with four repeats at the central point, using a circumscribed central composite design. For each of the studied variables, high (coded value: +2) and low (coded value: −2) set points were selected as shown in Table 1. In addition, Tables 2 shows the coded and real values of designed experiments based on CCD methodology, which was designed using EREGRESS software.

Table 1 The variables and values used for central composite design (CCD)

Variable	Name	−2(low)	−1	0	1	+2(high)
F1	PPD/%	0	0.005	0.01	0.015	0.02
F2	Fe^3+/M	0.02	0.04	0.1	0.16	0.22
F3	Acid/M	0	0.1	0.2	0.3	0.4
F4	SDS/%	0	0.5	1	1.5	2

---

Table 2 List of experiments in the CCD for model optimization (coded values)

Exp #	F1	F2	F3	F4	Absorbance
a indicates 4 repeat of center points.
1	0	2	0	0	0.335
2	1	1	−1	1	0.227
3	−1	−1	1	1	0.356
4	1	−1	1	−1	0
5	0	0	0	−2	0.869
6	−1	1	1	1	0
7	0	0	0	2	0.337
8^a	0	0	0	0	0.95
9^a	0	0	0	0	0.994
10	0	0	−2	0	0.729
11	1	−1	−1	−1	0.7
12^a	0	0	0	0	0.92
13	−1	−1	−1	1	0.237
14	2	0	0	0	0.166
15	1	−1	1	1	0
16	1	−1	−1	1	0.452
17^a	0	0	0	0	0.95
18	−1	1	−1	1	0.009
19	−1	−1	1	−1	0
20	−1	1	1	−1	0
21	−1	1	−1	−1	0.303
22	1	1	−1	−1	0.199
23	0	−2	0	0	0
24	−1	−1	−1	−1	0.498
25	1	1	1	−1	0.24
26	1	1	1	1	0.297
27	0	0	2	0	0
28	−2	0	0	0	0.25

---

Polynomial equations and response surfaces for a particular response are produced using EREGRESS. For an experimental design with four factors, the model including linear, quadratic and cross-terms can be expressed as:

Response = b0 + b1F1 + b2F2 + b3F3 + b4F4 + b5F1F1 + b6F2F2 +b7F3F3 + b8F4F4 + b9F1F2 + b10F1F3 + b11F1F4 +b12F2F3 + b13F2F4 + b14F3F4 (1)

Within eqn (1), F1–F4 are the variables, and b0–b14 are the coefficient values obtained by multiple linear regression (MLR) using EREGRESS software. The response surface plots are obtained by a statistical process that is described in the design and the modeled CCD data. Response surface methodologies graphically illustrate relationships between the parameters and the response(s) and are the way to obtain an exact optimum.^{35, 36}

The analysis of variance (ANOVA) and least squares techniques were used to evaluate the statistical significance of the constructed models. The ANOVA consists of determining which factor(s) significantly affects the response, using a Fisher's statistical test (F-test). The significance and the magnitude of the estimated coefficients of each variable and all their possible interactions on the response variable(s) are determined. Such coefficients for each variable represents the improvement in the response, that is, to expect as the variable setting is changed from low to high. Effects with a confidence level less than 95% (effects with a p-value higher than 0.05) were discarded and pooled into the error term and a new analysis of variance was performed for the reduced model. Note that the p-value represents a decreasing index of the reliability of a result.³⁵ Replicates ( n = four) of the central points were performed to estimate the experimental error.

In order to show the fitness of the model, regression coefficient (R) maybe be used. However, the adjusted regression coefficient (R_adj) and the prediction regression coefficient (R_pred) are a better criteria than the absolute regression coefficient (R). Since the regression coefficient (R) always decreases when a regression variable is eliminated from the model, in statistical modeling the R_adj, which takes the number of regression variables into account, is usually selected.^35,36 In addition, R_pred, which indicates the predictive power of the model, is chosen for the same reason. This parameter was approximated using prediction error sum of squares or PRESS that is calculated from residuals. So, the regression coefficient (R), adjusted R (R_adj), and R for prediction (R_pred) together are very convenient to get a quick impression of the overall fit and the prediction power of a constructed model.

3. Results and discussion

The main step for phenothiazine dye production is the oxidation of aromatic p-substituted amine to yield iminoquinone free-radical intermediary. Fe³⁺ in acidic media is used for oxidation purposes. Electron donor groups in p-position of aromatic ring of amines are decisive for free-radical stabilization, leading to an increase dye production rate and the analytical sensitivity.²⁷

The phenothiazine dye obtained from PPD showed a maximum absorption at 590 nm. Preliminary studies showed that, the analytical signal increased in the presence of SDS, CTAB and Triton X-100; but among the studied surfactants, SDS and Triton X-100 had the significant effect on the spectrophotometric determination of sulfide based on phenothiazine dye production. The effects of various experimental parameters in this reaction was investigated using central composite design. The experiments were designed at five levels of operating parameters. The concentration of SDS (as a model surfactant), amine, Fe³⁺ and H₂SO₄ were optimized simultaneously. The effect of, a non-ionic surfactant, Triton X-100 and a cationic one, CTAB, was optimized by the one variable at a time method.

The absorbance of samples (λ_max, 590 nm) were collected as response in order to optimize four independent factors, namely the PPD(F1), Fe⁺³(F2) H₂SO₄ (F3) and SDS concentration (F4). Tables 1 and 2 present the levels of coded and actual experimental variables examined. Table 2 also presents the corresponding response of each run.

The aims of this CCD strategy were: (i) to study the effect of surfactant (in this case the effect of SDS) on the determination of sulfide based on phenothiazine dye production (ii) to identify the variables that have a higher impact on determination of sulfide; (iii) to give an insight into the robustness of the method operating close to the optimum conditions; and (iv) to show, eventually, interactions between the variables.

In order to find the important factors and build a model to optimize the procedure, we started with a full quadratic model including all terms of eqn (1). To obtain a simple and yet a realistic model, the insignificant terms (p > 0.05) were eliminated from the model through a ‘backward elimination’ process. The constructed model using all 15 terms of eqn (1) showed a relatively good fit. For this model, the regression coefficient (R) for calibration close to 1 was achieved. Although the adjusted regression coefficient (R_adj) as well as the prediction coefficient (R_pred) were very low (<0.2). Some of these 15 regression variables are insignificant or at least have low significance and therefore should be eliminated from the model. As mentioned R, R_adj and R_pred together are very convenient to get a quick impression of the overall fit and the predictive power of the model. In a good model, these regression coefficients should not be too different from each other. However, for small data sets, it is very likely that every data point is influential. In these cases, a high value for R_pred cannot be expected.³⁶ By eliminating insignificant terms in eqn (1) from the constructed model, calibration R decreased to 0.88 but adjusted R (R_adj), and R for prediction (R_pred) increased nearly to 0.82 and 0.68, respectively. The reduced model using significant linear, quadratic and interaction parameters is shown in Table 3. The adjusted R values were well within the acceptable limits^37–39 and there were not large differences between R values which revealed that the experimental data shows a good fit to the quadratic equations and therefore an acceptable model has been achieved.

Table 3 Some characteristics of the constructed models

a prediction error sum of squares.
Regression equation	Response = b0 + b1*PPD/_% + b2*Fe/M + b5*PPD/_%*PPD/_% + b6*Fe/M*Fe/M + b12*Fe/M*Acid/M + b7*Acid/M*Acid/M + b14*Acid/M*SDS% + b8*SDS%*SDS%	Coefficient	Value
		b0	−1.304
		b1	113.49
		b2	114.99
		b5	−1897.0
		b6	−7992.8
		b12	951.32
		b7	−257.65
R of calibration	0.88	b14	12.67
		b8	−0.332
Adjusted R (Radj)	0.82
R of prediction (Rpred.)	0.68
Standard Error	0.173
No. Points	28
PRESS^a	1.40

---

3.4. Response surfaces and selection of the optimum conditions

In order to gain insight about the effect of each variable, the three-dimensional (3D) plots for the predicted responses were formed based on the model function. Some of the response surface plots are represented in Fig. 1, which show the 3D plots of absorbance of samples (590 nm) versus pairs of variables while other variables were kept in center levels. As shown in Fig. 1, there is a non-linear relationship between response and the variables F1–F4, because the surface plots of the response are curvature. It is shown in Table 3 that several linear, squared and interaction parameters are statistically significant.

Fig. 1 Response surface of full quadratic model between absorbance (590 nm) and four variables; PPD concentration (F1), Fe⁺³ concentration (F2), H₂SO₄ concentration (F3) and SDS amount (F4).

The selection of optimum conditions of the method is possible from the response surface plots (Fig. 1). These results demonstrate that all the response surfaces have a flat optimum. These plots show the interaction of pairs of mentioned variable when the remaining factors have been fixed using the constructed model by EREGRESS software. The results show a pronounced dependency of absorbance by all of the investigated experimental variables. Also these plots show that there is significant interaction between Fe³⁺–sulfuric acid concentration and SDS–sulfuric acid concentration (see also Table 3). The contributions of some other terms are clearly presented in Fig. 1. Finally, the optimum conditions can be selected from the obtained model for further examinations, shown in Table 4. The proposed method showed excellent repeatability (RSD < 2% for 6 replicate determinations of 0.5 μg ml⁻¹ sulfide) using the optimum values obtained by CCD.

Table 4 Optimum conditions obtained by response surface modeling

Variable name		Optimum values	Selected values
F1	PPD//mol L^−1	0.020–0.037	0.03
F2	Fe^+3/mol L^−1	0.009–0.013	0.01
F3	H2SO4/mol L^−1	0.03–0.04	0.04
F4	SDS/%	0.90–1.2	1.00

---

3.5. The effect of Triton X-100 and CTAB

The effect of Triton X-100 and CTAB on the spectrophotometric determination of sulfide based on phenothiazine dye production was also studied (Fig. 2). As Fig. 2 shows, the absorbance of the solution increased by increasing Triton X-100 concentration up to 1% (v/v), remained nearly constant in the range 0.8–1.2% (v/v) and decreased slowly at higher concentrations. Therefore, 1% (v/v) of Triton X-100 concentration was used as the optimal concentration for further study. Also as can be seen from Fig. 2, in the presence of CTAB the absorbance of the phenothiazine dye product was increased up to 0.6% w/v and remained nearly constant at higher concentrations. Therefore, 0.8% w/v of CTAB was used to construct the calibration curve.

Fig. 2 The effect of CTAB and Triton X-100 on the spectrophotometric determination of sulfide; Conditions: sulfide, 2 μg ml⁻¹; PPD, 0.03 mol L⁻¹; Fe3+, 0.01 mol L⁻¹; sulfuric acid, 0.04 mol L⁻¹

3.6. Analytical characteristics

Table 5 summarizes the analytical characteristics of the optimized method, including regression equation, linear range, and limit of detection. These results show that, the analytical signal was increased in the presence of SDS, CTAB and Triton X-100 but of the surfactants studied SDS and Triton X-100 have a significant effect on the spectrophotometric determination of sulfide based on phenothiazine dye production.

Table 5 Analytical characteristics of the proposed method

Experiment in the presence of	Regression equation	R 2	Linear range/μg ml^−1	LOD/μg ml^−1^b
a Squared regression coefficient. b The limit of detection, defined as CL = 3 Sb/m, where CL, Sb and m are the limit of detection, standard deviation of the blank, and the slope of the calibration graph, respectively.
SDS	A = 0.3435C + 0.01	0.994	0.1–2.0	0.04
Triton X-100	A = 0.374C + 0.015	0.995	0.1–2.0	0.035
CTAB	A = 0.287C + 0.05	0.9918	0.2–2.0	0.07
Without surfactant	A = 0.181C + 0.0385	0.994	0.4–3.0	0.10

---

3.7. Application of the proposed method

The proposed method (using SDS surfactant) was applied to the determination of sulfide ion in spiked water, tobacco and human plasma samples. The results are presented in Table 6.

Table 6 Determination of sulfide in spiked real samples

Sample	Added (μg ml^−1)	Found (μg ml^−1)	Recovery (%)
Tap water	0.30	0.29	96.7
	0.60	0.62	103.3
Tobacco (1)	0.30	0.28	93.3
	0.60	0.59	98.3
Tobacco (2)	0.30	0.30	100.0
	0.60	0.63	105.0
Human plasma	0.30	0.32	106.7
	0.60	0.61	101.7

---

In view of the fact that tobacco contains more than 4000 chemical components,⁴⁰ it is very difficult to separate and accurately analyze some of the components such as sulfide, sulfate etc using chromatographic methods. Another aspect of analyzing tobacco samples is pretreatment of the samples. Recently, direct acid and alkaline hydrolysis were introduced into tobacco analysis for the analysis of free and conjugated phytosterols in tobacco using gas chromatography–flame ionization detection.⁴¹ Pretreatment of tobacco samples by acidic hydrolysis must be done carefully and needs a closed system to trap the H₂S gas produced. In this work, 1 g of dry sample was powdered and added into 10 ml alkaline solution (1 mol L⁻¹ KOH) and was left for 1 h in order to extract the analyte into the water. The sample was then filtered and 3 ml of this solution was treated under optimum conditions for the purpose of determination.

Human plasma sample was obtained from a nearby hospital. Plasma samples were deproteinized using acetonitrile.⁴² An aliquot of 5 ml sample was transferred to a 10 ml tube and 5 ml of acetonitrile was added. The sample was mixed vigorously for 20 s and then centrifuged for 2 min at 3000 rpm in order to precipitate protein. These samples were found to be free from sulfide and so spiked samples were prepared by adding known amounts of sulfide. The results are presented in Table 6. Recoveries are in the range 93–105% which indicates the applicability of the proposed method for the determination of sulfide in real samples.

4. Conclusions

In the present work the effect of three different surfactants Triton X-100, cetyltrimethylammonium bromide (CTAB) and sodium dodecylesulfate (SDS) on the spectrophotometric determination of sulfide based on the phenothiazine dye production from phenyl-p-diamine (PPD) is studied.

This study showed that the use of experimental design enables subsequent benefit, in terms of labor time and number of experiments needed to optimize the conditions for modelling the influence of variables on the performance of spectrophotometric determination of sulfide from aqueous solution. By using central composite design and subsequently response surface methodology: (i) the effect of the surfactants on determination of sulfide based on phenothiazine dye production could be studied; (ii) variables that had a high impact on the determination of sulfide were identified; (iii) the optimum conditions of the proposed method were obtained and (iv) the possible interactions between variables were shown. The results also showed that the analytical signal increased in the presence of SDS, CTAB and Triton X-100; but of the studied surfactants, SDS and Triton X-100 had the most significant effect on the spectrophotometric determination of sulfide based on phenothiazine dye production. The proposed method was applied successfully for determination of sulfide in different spiked real samples. The analysis of different spiked water, human plasma and tobacco samples with satisfactory results indicated the applicability of the proposed method for the determination of sulfide in real samples.

References

1. P. Patnaik, A Comprehensive Guide to the Hazardous Properties of Chemical Substances; 2nd edn; Wiley: New York, 1999.
2. R. E. Kirk, D. F. Othmer, Encyclopedia of Chemical Technology; 3rd edn; Wiley: New York, 1981; Vol. 17.
3. L. Ebdon, S. J. Hill, M. Jameel, W. T. Corns and P. B. Stockwell, Analyst, 1997, 122, 689.
4. A. D. Eaton, L. S. Clesceri, A. E. Greenberg, Standard Methods for the Examination of Water and Wastewater; 19th edn; American Public Health Association: Washington, 1995.
5. C. H. Liu and S. Shen, Anal. Chem., 1964, 36, 1652.
6. J. L. Lambert and D. T. Manzo, Anal. Chim. Acta, 1969, 48, 185.
7. M. Roman Ceba, F. Vinagre Java and J. A. Munoz Leyva, Analyst, 1982, 107, 781.
8. J. Kurzawa, Anal. Chim. Acta, 1985, 173, 343.
9. D. R. Canterford, Anal. Chem., 1975, 47, 88.
10. I. G. Casella, M. R. Guascito and E. Desimoni, Anal. Chim. Acta, 2000, 409, 27.
11. P. Steinmann and W. J. Shotyk, J. Chromatogr., A, 1995, 706, 287.
12. D. Tang and P. D. J. Santschi, J. Chromatogr., A, 2000, 883, 305.
13. J. Radford-Knoery and G. A. Cutter, Anal. Chem., 1993, 65, 976.
14. C. F. Wood and I. L. Marr, Analyst, 1988, 113, 1635.
15. K. Han and W. F. Koch, Anal. Chem., 1987, 59, 1016.
16. W. Lei and P. K. Dasgupta, Anal. Chim. Acta, 1989, 226, 165.
17. A. A. Ensafi, Anal. Lett., 1992, 25, 1525.
18. D. Chen, M. D. Luque de Castro and M. Valcarcel, Analyst, 1991, 116, 1095.
19. V. Kuban, P. K. Dasgupta and J. N. Marx, Anal. Chem., 1992, 64, 26.
20. R. J. Cassela, L. G. Teixeira, S. Garrigues, A. C. S. Costa, R. E. Santelli and M. La Guardia, Analyst, 2000, 125, 1835.
21. M. S. P. Silva, I. S. da Silva, G. Abate and J. C. Masini, Talanta, 2001, 53, 843.
22. M. Kass and A. Ivaska, Anal. Chim. Acta, 2001, 449, 189.
23. M. S. P. Silva, C. X. Galhardo and J. C. Masini, Talanta, 2003, 60, 45.
24. A. Afkhami and R. Norooz-Asl, Sep. Sci. Technol., 2009, 44, 983.
25. T. D. Rees, A. B. Gyllenspetz and A. C. Docherty, Analyst, 1971, 96, 201.
26. V. Singh, S. Gosain, S. Mishra, A. Jain and K. K. Verma, Analyst, 2000, 125, 1185.
27. J. C. C. Santos, E. B. G. N. Santos and M. Korn, Microchem. J., 2008, 90, 1.
28. M. Kincl, S. Turk and F. Vrecer, Int. J. Pharm., 2005, 291, 39.
29. X. Zhang, R. Wang, X. Yang and J. Yu, Chemom. Intell. Lab. Syst., 2007, 89, 45.
30. A. Avila, E. I. Sanchez and M. I. Gutierrez, Chemom. Intell. Lab. Syst, 2005, 77, 247.
31. P. Vanloot, J.-L. Boudenne, L. Vassalo, M. Sergent and B. Coulomb, Talanta, 2007, 73, 237.
32. M. Capella-Perio, D. Bose, M. F. Rubert and J. Esteve-Romero, J. Chromatogr., B, 2006, B 839, 95.
33. C. Goncalves, J. J. Carvalho, M. A. Azenha and M. F. Alpendurada, J. Chromatogr., A, 2006, A 1110, 6.
34. Apha-Standard Methods for Examination of Water and Wastewater, 19th edn, American Public Health Association, Washington, 1995.
35. D. D. Stephan, J. Werner, R. P. Yeater, Essential Regression and Experimental Design for Chemists and Engineers, MS excel add in software package (1998–2001).
36. I. Bulacov, J. Jirkovsky, M. Muller and R. B. Heimann, Surf. Coat. Technol., 2006, 201, 255.
37. S. R. Senthilkumar, B. Ashokkumar, K. Chandra Raj and P. Gunasekaran, Bioresour. Technol., 2005, 96, 1380.
38. T. Sivakumara, R. Manavalana, C. Muralidharanb and K. Valliappan, J. Pharm. Biomed. Anal., 2007, 43, 1842.
39. T. Lundstedt, E. Seifert, L. Abramo, B. Thelin, A. Nystrom, J. Pettersen and R. Bergman, Chemom. Intell. Lab. Syst., 1998, 42, 3.
40. R. R. Baker, in: D. L. Davis, M. T. Nielsen (ed.), Tobacco Production, Chemistry and Technology, Blackwell Science, Oxford, UK, 1999, p. 398.
41. W. H. Liu, B. Ding, X. M. Ruan, H. T. Xuc, J. Yang and S. M. Liu, J. Chromatogr., A, 2007, 1163, 304.
42. H. J. Mascher and C. Kikuta, J. Chromatogr., A, 1998, A 812, 381.

---