Application of response surface methodology for determination of methyl red in water samples by spectrophotometry method

Accepted Manuscript Application of response surface methodology for determination of methyl red in water samples by spectrophotometry method Saeid Khodadoust, Mehrorang Ghaedi PII: DOI: Reference:

S1386-1425(14)00692-1 http://dx.doi.org/10.1016/j.saa.2014.04.119 SAA 12083

To appear in:

Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy

Received Date: Revised Date: Accepted Date:

4 March 2014 13 April 2014 22 April 2014

Please cite this article as: S. Khodadoust, M. Ghaedi, Application of response surface methodology for determination of methyl red in water samples by spectrophotometry method, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy (2014), doi: http://dx.doi.org/10.1016/j.saa.2014.04.119

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 2 3 4 5 6

Application of response surface methodology for determination of methyl red

7

in water samples by spectrophotometry method

8 9 10 11 12 13

Saeid Khodadoust, Mehrorang Ghaedi*

Department of Chemistry, Yasouj University, Yasouj 75914-35, Iran * Corresponding author E-mail: [email protected], Tel-fax: (098)-741-2223048.

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 1

1 2

Abstract:

3

In this study a rapid and effective method (dispersive liquid–liquid microextraction (DLLME))

4

was developed for extraction of methyl red (MR) prior to its determination by UV–Vis

5

spectrophotometry. Influence variables on DLLME such as volume of chloroform (as extractant

6

solvent) and methanol (as dispersive solvent), pH and ionic strength, extraction time were investigated.

7

Then significant variables were optimized by using a Box-Behnken design (BBD) and desirability

8

function (DF). The optimized conditions (100 µL of chloroform, 1.3 mL of ethanol, pH 4 and 4 %

9

(w/v) NaCl) resulted in a linear calibration graph in the range of 0.005–100 mgmL−1 of MR in initial

10

solution with R2 = 0.995 (n = 5). The limits of detection (LOD) and limit of quantification (LOQ) were

11

0.001 and 0.005 mg mL−1, respectively. Finally, the DLLME method was applied for determination of

12

MR in different water samples with relative standard deviation (RSD) less than 5% (n = 5).

13 14

Keywords:

15

Methyl red; Dispersive liquid–liquid microextraction; Box-Behnken design; Desirability function.

16 17 18 19 20 21 22 23 24

2

1

1. Introduction:

2

Dye and dye stuffs are extensively used in various areas such as textile, plastic, food, cosmetic,

3

carpet and paper industries [1, 2]. Wastewaters of these industries contain dye with metals, salts, and

4

other chemicals which may be toxic to aquatic environment [3 – 5]. The release of colored wastewater

5

from industry may produce an eco-toxic hazard and introduce potential danger of bioaccumulation,

6

which may eventually affect humans through the food chain [1]. Azo dyes are the largest group of dyes

7

used in industry [2]. The term azo dye is applied to synthetic organic colorants that are characterized by

8

a nitrogen-to-nitrogen double bond: -N=N- [6]. Durability of azo dyes causes pollution to the

9

environment. Besides, some azo dyes are toxic and mutagenic [7]. It is well known that methyl red

10

(MR) dye has been used in paper printing and textile dyeing [8,9] and it causes irritation of the eye,

11

skin and digestive tract if inhaled/swallowed [10].

12

Sample preparation is an important step in analytical methods for determination of analytes in various

13

matrices. So a combination of instruments with novel sample preparation methods has enabled analysis

14

of trace amounts of analytes with higher accuracy. In the last decades design and development of

15

miniaturized alternative methods to the older sample preparation techniques has been one of the most

16

important challenges for analysts [11]. Liquid-Liquid Extraction (LLE) and Solid Phase Extraction

17

(SPE) as the most commonly used techniques for sample preparation suffers from the disadvantages

18

including time consuming, high cost, and need large volumes of samples and toxic organic solvents.

19

Microextraction methods such as Solid Phase Microextraction (SPME) [12] and Liquid Phase

20

Microextraction (LPME) [13, 14] have attracted much attention in recent years as alternatives for

21

classic liquid–liquid and solvent extraction procedures. These techniques are simple and fast,

22

miniaturize sample pretreatment processes and minimize the use of organic solvents. However, these

23

techniques suffer from some problems such as sample carry-over, relatively high cost, fiber fragility

24

and relatively low precisions [15]. Dispersive Liquid–Liquid Microextraction (DLLME) was developed 3

1

to overcome these limitations [16, 17]. It is based on a ternary component solvent system like

2

homogeneous LLE and Cloud Point Extraction (CPE) [18]. In this method, the appropriate mixture of

3

extraction and dispersive solvents are injected rapidly into aqueous sample by syringe, and a cloudy

4

solution is formed. The analyte in the sample is extracted into fine droplets of extraction solvent. After

5

extraction, phase separation is performed by centrifugation and the enriched analyte in the sediment

6

phase is determined by various methods. From commercial, economical and environmental point of

7

view, DLLME offers several important advantages over conventional solvent extraction methods:

8

faster operation, easier manipulation, no need of large amounts of organic extraction solvents, low time

9

and cost, high extraction recovery and enrichment factor [19-21]. There are several experimental

10

variables affecting the DLLME procedure that must be optimized [22, 23].

11

In recent years, chemometric tools such as Response Surface Methodology (RSM)

12

based on statistical Design of Experiments (DOEs) have been frequently applied to the

13

optimization of analytical methods. Such statistical analyses are more efficient, since they account

14

for interaction effects between the studied variables and determine more accurately the combination of

15

levels that produces the optimum of the process [19, 22]. If there significant interaction affects

16

between variables, the optimal conditions indicated by the univariate studies will be different

17

from the correct results of the multivariate optimization. So the univariate procedure may fail

18

since the effect of one variable can be dependent on the level of the others involved in the

19

optimization process. That is why multivariate optimization schemes involve designs for which

20

the levels of all the variables are changed simultaneously. These methods have advantages 4

1

such as a reduction in the number of experiments that need be executed resulting in lower

2

reagent consumption and considerably less laboratory work. A structured experiment design that

3

could simultaneously take into account several variables seems a more convenient approach searching

4

for the optimal operational conditions in a reasonable number experimental runs [24, 25].

5

The Box-Behnken design (BBD) is a second-order multivariate technique based on

6

three-level incomplete factorial designs that received a wide application for assessment of

7

critical experimental conditions, that is, maximum or minimum of response function. BBD, a

8

spherical and revolving design, has been applied in optimization of chemical and physical

9

processes because of its reasoning design and excellent outcomes [22 - 24].

10

In the present work, DLLME followed by UV-Vis detection was applied for extraction and

11

determination of the MR in water samples. Influence of important DLLME variables such as the kind

12

and volume of extraction and disperser solvent, pH of the sample solution, extraction time and salt

13

effect were investigated and optimized by BBD and Desirability function (DF). The applicability of

14

presented method for the analysis of water samples has also been investigated.

15 16

2. Experimental

17

2.1. Reagents and Instrumentation

18

All chemicals that used in this work were of analytical reagent grade and were used without further

19

purification. Double distilled deionized water was used throughout which was produced by a Milli-Q

20

system (Millipore, Bedford, MA, USA). Carbon tetrachloride, chloroform, dichloromethane, methanol,

21

acetonitrile (HPLC grade), acetone and tetrahydrofuran (for spectroscopy) were obtained from Merck 5

1

(Darmstadt, Germany). The methyl red (MR) (Fig. 1) and sodium chloride were purchased from

2

Merck. A stock standard solution of MR (200 mg L−1) was prepared in methanol. The working standard

3

solutions were prepared in double distilled/deionized water.

4

Recording the absorption spectra and absorbance measurements were carried out with a UV–Vis

5

spectrophotometer (model V-530, Jasco, Japan) using 1.0 cm quartz cells. A pH meter (model-686,

6

thermometer Metrohm, Switzerland) equipped with a combine Ag/AgCl glass electrode was used to

7

check the pH of the solutions. A Hermle Labortechnik GmbH centrifuge model Z206A (Germany) was

8

used to accelerate the phase separation. Insert Fig. 1

9 10

2.2. Dispersive liquid–liquid microextraction (DLLME) procedure

11

For DLLME, 5.0 mL aliquot of water sample was placed in a 10 mL screw cap glass tube with

12

conic bottom and spiked at the level of 0.5 mgmL−1 of MR. A mixture of 1.3 mL of ethanol (as

13

disperser solvent) and 100 µL chloroform (as extraction solvent) was injected rapidly into a sample

14

solution by using 2.0 mL syringe and mixture was gently shaken. A cloudy solution that consists of

15

very fine droplets of chloroform dispersed into aqueous sample was formed, and the MR was extracted

16

into the fine droplets. After centrifugation at 4000 rpm min−1 for 5 min the chloroform phase was

17

sediment (about 70 µL) at the bottom of the centrifuge tube and entirely transferred into a vial using

18

100µL syringe for evaporation of solvent. The residue was dissolved in 2.00 mL methanol and was

19

conveyed to a UV–Vis spectrophotometer to measure its absorbance at λmax (486 nm).

20 21 22 23 24

2.3. Calculation of enrichment factor and extraction recovery The enrichment factor (EF) was defined as the ratio between the analyte concentration in the settled phase (Cset) and the initial concentration of the analyte (C0) in the aqueous sample [26]. EF= Cset/C0

(1) 6

1

The extraction recovery (ER %) was defined as the percentage of the total analyte which was extracted

2

in the settled phase.

3

ER% = (Cset × Vset/C0 × Vaq)×100 = EF× Vset/ Vaq

(2)

4

where Vsed and Vaq are the volume of the sediment phase and the volume of the aqueous sample,

5

respectively [27]. All the experiments were performed in triplicates and average of the results reported

6 7

2.4. Selection of extraction solvent and disperser solvent

8

The selection of a suitable extraction solvent is very important for the DLLME process. The

9

extracting solvent has to meet two properties: to extract the analytes well and to have density higher

10

than of water [25]. Hence, carbon tetrachloride (density, 1.59 gmL−1), chloroform (density, 1.48

11

gmL−1) and dichloromethane (density, 1.32 gmL−1) were considered for this purpose.

12

As for the choice of dispersive solvent in DLLME, the miscibility in organic phase (extraction solvent)

13

and aqueous phase (sample solution) is a key factor, which can disperse extraction solvent into very

14

fine droplets in aqueous phase. Acetonitrile, methanol, ethanol and tetrahydrofuran were compared as

15

disperser solvent in the extraction of MR. For selecting the best extraction and dispersive solvents, 150

16

µL of CCl4, CHCl3, CH2Cl2 combined with 1.5 mL of acetonitrile, ethanol, methanol, THF and the ER

17

of them were examined. According to the results shown in Table 1, ethanol as the disperser solvent and

18

chloroform as the extraction solvent provided maximum ER of 80.0%. Therefore, we selected mixture

19

of ethanol and chloroform as a suitable set for subsequent experiments. Insert Table 1

20 21

2.5. Effect of pH

22

pH of solution is an important factor during liquid–liquid extraction (LLE) process involving

23

analytes that possess an acidic or basic moiety. The ionic form of a neutral molecule formed upon

24

deprotonation of a weak acid or protonation of a weak base normally does not extract through the 7

1

organic solvent as strongly as does its neutral form. Thus, pH should be adjusted to ensure that neutral

2

molecular forms of the analytes are present prior to performing the microextraction step. The effect of

3

pH of solution on the amount of extracted MR was investigated in the range of 1–8.

4 5

2.6. Effect of salt addition

6

Addition of salt often improves extraction of analytes in conventional liquid–liquid extraction, as a

7

result of the salting-out effect. Generally, addition of salt enhanced extraction of analytes, because the

8

presence of the salt reduced the solubility of the analytes in water and forced more of them on to the

9

organic phase. To evaluate the possibility of salting-out effect, extraction recovery of MR by DLLME

10

was studied in the range of 0 - 6 % (w/v) of NaCl (obtained results are shown in results and discussion

11

section).

12 13

2.7. Effect of extraction time

14

In miniaturized preconcentration methods such as DLLME and LPME, extraction time is one of

15

the most consequential parameters. The time of extraction is defined as an interval time between

16

injection of mixture of disperser solvent (ethanol) and extraction solvent (chloroform) before starting to

17

centrifugation. In this work effect of extraction time was examined from 0 to 10 min. The obtained

18

results showed that the variations of ER against the extraction time were not significant. In DLLME,

19

the surface area between the extraction solvent and the aqueous phase is significantly large, so that the

20

transfer of the analyte from the aqueous phase into the extraction phase is carried out quickly.

21

Therefore, the time of extraction was very succinct because equilibrium state was obtained very fast.

22

For this reason 1 min was chosen for extraction time in subsequent experiments.

23 24

3. Results and discussion 8

1

In DLLME several variables such as volume of extraction and disperser solvents, salt effect,

2

extraction time and pH of sample solution affect the ER of MR. In order to obtain the optimum

3

condition for extraction of MR from the water samples. BBD and DF were used for searching the

4

optimal experimental conditions. RSM is an empirical modeling technique used to evaluate the

5

relationship between a set of controllable experimental factors and observed results. This optimization

6

process involves three major steps: (i) performing statistically designed experiments, (ii) estimating the

7

coefficients in a mathematical model, and (iii) predicting the response and checking the adequacy of

8

the model [20]. A class of three level complete factorial design for the estimation of the parameters in a

9

second-order model was developed by Box–Behnken [21]. For this purpose STATISTICA 7.0

10

statistical package was used to generate the experimental matrix and to evaluate the results.

11 12

3.1. Box-behnken design (BBD)

13

According to the principle of BBD, extraction volume, dispersion volume, pH of sample solution

14

and ionic strength, which were identified to have strong effects on the response in preliminary one-

15

factor-at-a-time experiments, were taken as the variables tested in a 27-run experiment to determine

16

their optimum levels. In order to optimize the critical factors and describe the nature of the response

17

surface in the experiment, RSM was employed. A BBD with four variables was used to determine the

18

response pattern and then to establish a model. As presented in Table 2, four independent variables

19

used in this work were the extraction solvent volume (X1), dispersion solvent volume (X2), pH (X3) and

20

concentration of NaCl (w/v %) (X4), which were prescribed into three levels coded (−1, 0, + 1) were

21

selected for each set of experiments. This design led to study the effects of four variables in a three

22

block of 27 sets of test conditions and three central points. The total number of design points needed

23

(N) is determined by the following equation:

24

N= 3(K-1) + C

(3) 9

1

where K and C is the number of variables and center points (K = 4, C = 3), respectively. The order of

2

the experiments was fully randomized. According to applied design, 30 combinations (see Table 3)

3

were executed and the mathematical relationship between the four independent variables can be

4

approximated by the second order polynomial model [29]:

5


=  + ∑    + ∑  ∑

   + ∑   

(4)

6

where y is the predicted response (extraction recovery); Xi's are the independent variables (ethanol,

7

CHCl3, pH and NaCl) that are known for each experimental run. The parameter β0 is the model

8

constant; βi is the linear coefficient; βii are the quadratic coefficients; and βij are the cross-product

9

coefficients (interaction). Analysis of variance (ANOVA) was used to estimate the statistical

10

parameters. The extent of fitting the experimental results to the polynomial model equation was

11

expressed by the determination coefficient R2. F-test was used to estimate the statistical significance of

12

all terms in the polynomial equation within 95% confidence interval. Table 4 represents the design

13

matrix of the variables together with the experimental results.

14

The adequacy of the model was checked using analysis of variance (ANOVA) which was tested using

15

Fisher's statistical analysis. The regression coefficients in the response surface model, for the linear,

16

quadratic and interaction effects of the variables are shown along with F- and P-value in Table 4. A p-

17

value less than 0.05 in the ANOVA table indicates the statistical significance of an effect at 95%

18

confidence level. The quality of fit of the polynomial model equation was expressed by the coefficient

19

of determination (R2 = 0.995 and adjusted R2 = 0.945).

20

Insert Table 2

21

Insert Table 3

22

In the next step of the design, a response surface model was developed by considering all the

23

significant interactions in the BBD. Data analysis gave a semi-empirical expression of ER % with

24

following equation: 10

1

 % = 47.483 + 17.550 − 10.654  + 13.894  + 7.798  + 8.839  + 8.131  − 9.501  (5)

2

The final step was to find the conditions of the variables (extractor volume, disperser volume, pH and

3

ionic strength) that maximize the response (ER %) of the dependent variables. The interaction effects

4

and optimal levels of the variables were determined by plotting the response surface curves. The

5

response surface curves are represented in Fig. 2a-c. The curvatures of these plots indicate the

6

interaction between the variables.

7

The surface plot (Fig. 2a) showed that for low pH and extraction solvent (CHCl3), the ER%, is low

8

(≈35%), but with increasing pH to 3-4, and increasing volume of CHCl3 to 90- 120 µL, the ER was

9

reached to ≈88 %. In Fig. 2b, by increasing the extraction volume to 100-120 µL, while the percentage

10

of NaCl was 4% the ER increased to about 90% and by more increasing the percentage of NaCl,

11

because of the increased the volume of sediment phase, the ER was decreased.

12

The surface plot (Fig. 2c) showed that for low volume of ethanol and CHCl3 the ER% is low (≈25%).

13

This may be attributed to the fact that when low volumes of ethanol were used, extraction solvent could

14

not be dispersed efficiently and therefore, cloudy solution was not formed completely. For medium

15

amounts of CHCl3 as ethanol increases, the recovery slightly increases reaching a plateau. However, at

16

higher volumes of CHCl3 (120 µL) excess of ethanol displayed a negative effect on the ER%. This may

17

be attributed to the fact that as higher amounts of disperser solvent was used, solubility of the target

18

analytes was increased and therefore, ER% decreases. The highest ER≈90% was found in the region

19

where value of ethanol (1.0 –1.4 mL) and high values of CHCl3 (90 – 130 µL) were used as dispersive

20

and extractive solvents, respectively.

21

Insert Fig. 2

22

Insert Table 4

23

3.2. Optimization of BBD by DF for extraction of MR

11

1

The profile for predicted values and desirability option in the STATISTICA 7.0 software is used

2

for the optimization process (Fig. 3). Procedure and operation functions of desirability were explained

3

in pervious publication [16]. Profiling the desirability of responses involves specifying the DF for each

4

dependent variable (ER%), by assigning predicted values a scale ranging from 0.0 (undesirable) to 1.0

5

(very desirable). The BBD design matrix results from Table 3 represented the maximum (93.5%) and

6

minimum (35.2%) ER of MR. According to these values, DF settings for each dependent variable of

7

ER% are depicted at the right hand side of Fig. 3: desirability of 1.0 was assigned for maximum ER%

8

(93.5%), 0.0 for minimum (35.2%) and 0.5 for middle (64.3%). On the left hand side of Fig. 3

9

(bottom), the individual desirability scores are illustrated, respectively, for the ER %. Since desirability

10

1.0 was selected as the target value, the overall response (ER%) obtained from these plots with the

11

current level of each variable in the model are depicted at the top (left) of Fig. 3. These figures allow

12

seeing at a glance how changes in the level of each variable affect not only response (ER%) but also

13

overall desirability of the responses. On the basis of these calculations and desirability score of 1.0, ER

14

of MR was optimized at 95.3% with calculating the optimized model variables of 100.0 µL CHCl3, 1.3

15

mL ethanol, 4.0 for pH of sample solution and 4.0% (w/v) of NaCl. Finally, for validation, duplicate

16

assenting experiments were conducted using the optimized variables. The results are closely co-related

17

with the data obtained from desirability optimization analysis using BBD, indicating that BBD with DF

18

could be effectively used to optimize the ER% of the target analyte. Insert Fig. 3

19 20

3.3. Analytical performance of DLLME

21

A series of working solution containing MR at six concentration levels of 0.005, 0.025, 0.08, 1.0,

22

5.0 and 10.0 mg mL−1 were prepared for the establishment of the calibration curve. For each

23

concentration, five replicate extractions were performed under optimized conditions. The detection

24

limit (LOD) of the method was calculated from 3σ, using the calibration curve data. The standard 12

1

deviation of the blank (sb) was estimated from that of the residuals (sy/x) for this calculation. At

2

optimum conditions method has linear response over 0.005 –10.0 mg mL−1 with detection limit 0.001

3

mg mL−1 in water samples with correlation determinations (R2) 0.995.

4

The water samples were spiked with the standards of the MR at the concentration of 0.1 and 0.2 mg

5

mL−1. For each concentration, five replicate experiments with the whole analysis process were

6

performed and the results are given in Table 5. The result repeatability expressed as relative standard

7

deviations (RSDs, n = 5) was less than 5.0 %, for MR. These results (Table 5) show that the proposed

8

method has a high sensitivity and repeatability.

9

Insert Table 5

10

The effect of potential interferences, encountered in real samples, on of 1 mg mL−1 MR standard

11

solution in the presence of various amounts of individual interfering ions and some dyes, were

12

examined. The tolerance level was defined as the maximum concentration of the foreign species

13

causing a change in the analytical signal not higher than 90 % when it was compared with the signal of

14

dye lonely. The obtained results are given in Table 6. The results showed that the recoveries of the

15

analytes, in the presence of interfering ions at the ratios that usually occur in real samples are almost

16

quantified by using DLLME method and four dyes could be tolerated up to 55. Insert Table 6

17 18

4. Conclusion

19

This work investigated the determination of MR in water samples. Experiments were made as a

20

function of different variables (volume of extraction and dispersive solvents, pH and ionic strength of

21

sample solution). Response Surface Methodology (RSM) by the Box–Behnken model was used to

22

examine the role of four process variables on MR determination. It was shown that a second-order

23

polynomial regression model could properly interpret the experimental data with coefficient of

24

determination (R2) value of 0.995. The simultaneous optimization of the multi response system by 13

1

desirability function indicated that 95.3% extraction recovery of MR can be possible by using the

2

optimal conditions of 100 µL chloroform, 1.3 mL ethanol, pH 4 and 4 % (W/V) NaCl. The optimized

3

DLLME combined with UV-Vis spectrophotometer allowed quantification of trace levels of MR in the

4

water samples. This technique provides good repeatability and reproducibility, high extraction recovery

5

and short extraction and separation time and cloud be applied for routine analysis and determination of

6

MR in water samples.

7 8

Acknowledgments

9

The author expresses their appreciation to the Yasouj University, Yasouj, Iran for financial support of

10

this work.

11 12

References:

13

[1] J.X. Lin, S.L. Zhan, M.H. Fang, X.Q. Qian, H. Yang, J. Environ. Manag. 87 (2008) 193–200.

14

[2] Z.A. ALOthman, Y.E. Unsal, M. Habila, A. Shabaka, M. Tuzen, M. Soylak, Food Chem. Toxicol.,

15 16 17

50 (2012) 2709-2713. [3] M. Ghaedi, H. Tavallali, M. Montazerozohori, E. Zahedi, M. Amirineko, S. Khodadoust, G. Karimipour, Environ. Monit. Assess 184 (2012) 6583–6591.

18

[4] S. Khodadoust, A. Sheini, N. Armand, Spectrochimica Acta Part A 92 (2012) 91– 95

19

[5] M. Ghaedi, H. Tavallali, M. Montazerozohori, S.D. Mousavi, S. Khodadoust, M. Soylak, Fresenius

20 21 22 23 24

Environ. Bull. 20 (2011) 2783–2793. [6] M. Ghaedi, A. Najibi, H. Hossainian, A. Shokrollahi, M. Soylak, Toxicol. Environ. Chem. 94 (2012) 40-48. [7] C. Zhang, C. Fu, L. Bishop, M. Kupferle, S. FitzGerald, H.H. Jiang, C. Harmer, J. Hazard. Mater. 41 (1995) 267–85. 14

1

[8] C. Sahoo, A.K. Gupta, P. Anjali, Desalination 181 (2005) 91–100.

2

[9] H. Lachheb, E. Puzenat, A. Houas, Mohamed Ksibi, E. Elaloui, C, Guillard, J.M. Herrmann, Appl.

3

Catal. B 39 (2002) 75–90.

4

[10] Y. Badr, M.G. Abdul El-Wahed, M.A. Mahmoud, J. Hazard. Mater. 154 (2008) 245–253.

5

[11] E. Psillakis, N. Kalogerakis, Trends Anal. Chem. 22 (2003) 565-274.

6

[12] C.L. Arthur, J. Pawliszyn, Anal. Chem. 62 (1990) 2145–2148.

7

[13] S. Pedersen-Bjergaard, K.E. Rasmussen, Anal. Chem. 71 (1999) 2650–2656.

8

[14] L. Zhao, H.K. Lee, J. Chromatogr. A 919 (2001) 381–388.

9

[15] G. Shen, H.K. Lee, Anal. Chem. 74 (2002) 648–654.

10

[16] F. Ahmadi, Y. Assadi, M.R. Milani Hosseini, M. Rezaee, J. Chromatogr. A 1101 (2006) 307–312.

11

[17] W. Li, X. Zhang, P. Tong, T. Wu, H. Hu, M. Wang, Y. Du, Spectrochimica Acta Part A 124,

12

(2014) 159-164

13

[18] M.K. Purkait, S. DasGupta, S. De, J Hazard Mater 137 (2006) 827-835.

14

[19] S. Khodadoust, M.R. Hadjmohammadi, Anal. Chim. Acta 699 (2011) 113– 119.

15

[20] R. Khani, J. B. Ghasemi, F. Shemirani, Spectrochimica Acta Part A 122 (2014) 295-303.

16

[21] G.E.P. Box, J.S. Hunter,W.G. Hunter, Second ed., Wiley-Interscience, New York, 2005.

17

[22] S. Khodadoust, M. Ghaedi, J. Sep. Sci. 36 (2013) 1734–1742.

18

[23] S. Khodadoust, M. Ghaedi, M.R. Hadjmohammadi, Talanta 116 (2013) 637–646

19

[24] M. Muthukumar, D. Mohan, M. Rajendran, Cem. Concr. Compos. 25 (2003) 751-758.

20

[25] E. Zeini Jahromi, A. Bidari, Y. Assadi, M.R. Milani, M.R. Jamali, Anal. Chim. Acta, 585 (2007)

21 22 23

305–311. [26] M. Rezaee, Y. Assadi, M.R. Milani Hosseini, E. Aghaee, F. Ahmadi, S. Berijani, J. Chromatogr. A, 1116 (2006) 1–9.

15

1 2 3

[27] X. Zang, J. Wang, O. Wang, M. Wang, J. Ma, G. Xi, Z. Wang, Anal. Bioanal. Chem. 392 (2008) 749–754. [29] Y. Lai, G. Annadurai, F.C. Huang, J.F. Lee, Technol. 99 (2008) 6480–6487.

4 5 6 7

Figure captions

8 9 10 11 12 13

Fig. 1: Molecular structure of MR. Fig. 2: Response surfaces for the 3(4-1) BBD: (a) pH-volume of CHCl3 ; (b) NaCl % (W/V)–volume of CHCl3; (c) volume of ethanol–volume of CHCl3. Fig. 3: Profiles for predicated values and desirability function for ER of MR. Dashed line indicated current values after optimization.

14 15 16 17 18 19 20 21 22 23

16

1 2 3 4 5 6 7 8 9 10

Table 1: Effect of extraction solvent and disperser solvent on extraction recovery obtained from DLLME technique.

11 Disperser solvent

Extraction solvent

THF

Acetonitril

Ethanol

Mehanol

70.4 ±3.6

64.5±2.8

41.0±4.6

44.6±2.8

CHCl3

68.7±4.2

75.2±3.4

80.7±3.8

62.3±3.7

CH2Cl2

-

-

45.4±2.7

-

CCl4

a

12

a: Extraction recovery (%)

13

b: Relative standard deviation (RSD)

b

14 15 16 17 18 19 20 17

1 2 3 4 5 6

7

Table 2:

8

Level of different variables used in the Plackett–Burman screening design for the extraction of MR.

9 10

Levels Factors

Low (-1)

Central (0)

High (+1)

(X1) Volume of CHCl3 (extraction solvent) (µL)

70.0

100.0

130.0

(X2) Volume of ethanol (disperser solvent) (mL)

1.0

1.3

1.6

(X3) pH of sample solution

2.0

4.0

6.0

(X4) Ionic strength (NaCl concentration; w/v) (%)

0.0

3.0

6.0

11

12

13 14 15 16 17 18 19

20

21 18

1

2

Table 3: Design matrix for the 3(4-1) Box-Behnken design and obtained result for each run.

3

Runs 1

Block 1

X1 1

X2 0

X3 -1

X4 -1

ER (%) 38.2

2

1

0

1

0

1

68.2

3

1

-1

-1

0

1

45.3

4

1

0

1

-1

-1

67.3

5

1

1

0

0

1

80.2

6

1

-1

-1

1

0

48.9

7

1

1

0

1

0

58.5

8

1

-1

-1

-1

-1

52.3

9

1

0

1

1

0

78.2

10

2

-1

0

0

0

45.3

11

2

-1

0

1

-1

75.2

12

2

1

1

-1

1

53.2

13

2

0

-1

1

-1

69.3

14

2

0

-1

0

0

89.2

15

2

0

-1

-1

1

86.2

16

2

1

1

0

0

62.5

17

2

-1

0

-1

1

48.2

18

2

1

1

1

-1

82.3

19

3

0

0

1

1

35.2

20

3

0

0

0

-1

92.4

21

3

-1

1

-1

0

45.3

22

3

0

0

-1

0

64.2

23

3

1

-1

1

1

72.0

24

3

1

-1

0

-1

58.3

25

3

1

-1

-1

0

63.5

26

3

-1

1

1

1

72.3

27

3

-1

1

0

-1

93.5

(c)

28

1

0

0

0

0

86.5

29(c)

2

0

0

0

0

85.2

(c)

3

0

0

0

0

84.3

30

4

(c): center point

19

1 2 3 4 5 6 7

Table 4: Analysis of variance (ANOVA) for Box-Behnken design. Source of variation

Sum of square

Degree of freedom

Mean square

F- value

P- value

81.250

2

40.6250

2.57609

0.223239

X1 (L+Q)

1658.021

2

829.0106

52.56884

0.004621

X2 (L+Q)

226.054

2

113.0271

7.16723

0.071997

X3 (L+Q)

733.951

2

366.9754

23.27047

0.014902

X4 (L+Q)

500.969

2

250.4847

15.88362

0.025347

X1X2

67.005

4

16.7512

1.06222

0.500324

X1X3

594.736

4

148.6841

9.42829

0.047795

X1X4

99.121

2

49.5606

3.14272

0.183645

X2X3

143.622

3

47.8742

3.03577

0.193061

X2X4

197.367

2

98.6837

6.25769

0.085023

X3X4

157.503

1

157.5025

9.98748

0.050859

Pure error

47.310

3

15.7700

Total error

8410.180

29

Blocks

8

L: linear

9

Q: quadratic

10 11

20

1 2 3 4 5 6 7

Table 5: Extraction recoveries and RSD in different water samples at spiked level by the DLLME-UV-Vis method. Samples Distillate water

River water

Tap water 8

Added (mg mL-1)

Found (mg mL-1)

0.000 0.100 0.200 0.000 0.100 0.200 0.000 0.100 0.200

ND 0.095 0.191 ND 0.097 0.18 ND 0.097 0.185

ND: not detect

9 10 11 12 13 14 15 16 17 21

ER ± RSD (%)

95.3 ± 4.7 95.5 ± 3.6 97.2 ± 4.2 90.4 ± 3.6 97.3 ± 3.9 92.7 ± 4.5

1 2 3 4 5 6 7

Table 6: Tolerance limits of interfering species in the determination of 1 mgmL−1 of MR. Interference I

-

F

Br

-

Tolerance ratio (w/w)

ER (%)

1000

100

1000

90

750

90

3+

1000

99

3+

750

99

2+

750

94

2+

1000

96

Ca

2+

1000

98

Zn

2+

1000

96

1000

99

Fluorescein

55

75

Gongo Red

65

80

Sunset yellow

150

84

Brilliant green

200

90

Eosin B

120

87

Ce

Co Ni

Pb

Mg

2+

8 9 10 11 12 13 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Fig. 1

15 16 17 18 19 20 21 22 23 24 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Fig. 2

29 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Fig. 3

22 23

25

1 2

3

Highlight:

4

5

•

A simple method for the extraction of methyl red by using chemometrics was developed.

6

•

The significant variables were optimized by using a BBD combined with DF.

7

•

This technique provides good repeatability and high extraction recovery.

8 9

26

1 2

Graphical abstract

3

4

27