BOX-BEHNKEN DESIGN FOR SEQUESTRATION OF DYE: STATISTICAL OPTIMIZATION AND MECHANISM STUDIES Sivahankar Raja 1 , Sathya. A. Bose 2 , Kanimozhi Jayaram 3 , Sivasubramanian Velmurugan 4 * 1 Department of Chemical Engineering, Hindustan Institute of Technology and Science, Chennai, India-603103 2 Department of Biotechnology, Aarupadai Veedu Institute of Technology, Paiyanoor, India- 603104 3 Department of Biotechnology, Kalasalingam Academy of Research and Education, Krishnankoil, India- 626128. 4 Department of Chemical Engineering, National Institute of Technology, Calicut, India-673601. *Corresponding mail id: [email protected]Abstract: The wastewater treatment containing dyes are challenging due to their recalcitrant properties. Sequestration of dyes via adsorption principle seems to be promising. The purpose of this study was to optimize the operational condition and enhance the adsorption of MY dye using response surface modeling. Aquatic macrophyte (Salvinia molesta), iron oxide synthesized via chemical co-precipitation, and iron oxide modified aquatic macrophyte have been successfully exploited for the sequestration of dye from aqueous solution. For the investigation of effects of major operational variables and optimization conditions, response surface modeling based three factors, three levels Box-Behnken experimental design associated with quadratic programming has been utilized for maximizing the sequestration processes. Initial dye concentration (10 to 100 mg/L), adsorbent dosage concentration (0.5 to 1 g), and initial solution pH (2.0 to 12.0) are the three independent variables were studied from 17 experimental run in a batch system. For analysis of variation with 95 % confidence limit, the significance of the model terms contained in the Regression equations have been assessed using the F - and P - test. The adsorbent dosage concentration of 1.0 mg/100 ml, pH of 2, and initial dye concentration of 10 mg/L were found to be the economical operating conditions for maximum removal efficiency of 96.42, 97.25 and 98.51 using aquatic macrophyte, iron oxide and iron oxide modified aquatic macrophyte, respectively. As a result, Box-behnken design was suggested as a virtuous statistical tool for optimal design of dye sequestration that provides appropriate results for the maximum percentage decolorization. Keywords: Box-behnken design, optimization, sequestration, dye. 1. INTRODUCTION Dyes are substances that provide color by a process that alters temporarily or permanently. The chemicals applied for producing dyes are frequently highly toxic, carcinogenic, or even explosive. About 10,000 different dyes and pigments are industrially used, with annual global production of more than 7x105 tons of synthetic dyes [1]. Industries such as textile, cosmetics, paper, pharmaceutical, leather, plastics, and food are the major users of these synthetic dyes. Azo dyes that consist of one or more ( – N=N–) azo bonds are the most significant class of synthetic organic dyes. These azo dyes contribute to more than 60% of all used synthetic dyes and are difficult to degrade because of their complex structure and synthetic nature [2]. Many azo dyes are thought to be recalcitrant and toxic contaminants prevent photosynthetic activity and harmful to aquatic organisms. It is a renowned fact that exposure to these dyes results in aesthetic problems and sometimes mutagenic and carcinogenic effect in human beings. Further, degradation of these azo dyes gives rise to the formation of toxic amines [3]. Hence, sequestration of these dyes from aqueous solution is desirable. Adsorption is an easy and cost-effective wastewater processing technique, among several techniques available for the removal of dyes from effluent, such as flocculation, oxidation and electrolysis [4]. Generally, it is difficult to separate the used adsorbents after saturation adsorption. In these circumstances, magnetic separation of used adsorbents would be one of the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
BOX-BEHNKEN DESIGN FOR SEQUESTRATION OF
DYE: STATISTICAL OPTIMIZATION AND
MECHANISM STUDIES
Sivahankar Raja1, Sathya. A. Bose
2, Kanimozhi Jayaram
3, Sivasubramanian
Velmurugan4*
1Department of Chemical Engineering, Hindustan Institute of Technology and Science, Chennai,
India-603103
2Department of Biotechnology, Aarupadai Veedu Institute of Technology, Paiyanoor, India-
603104
3Department of Biotechnology, Kalasalingam Academy of Research and Education, Krishnankoil,
India- 626128.
4Department of Chemical Engineering, National Institute of Technology, Calicut, India-673601.
The quadratic model‟s F-value of 194.33, 2242.87 and 594.80 shows that the model was
significant for MY dye sequestration using SM, IO and MBC respectively. In addition,
the maximum decolorization percentage for all the three adsorbents was estimated from
the second order quadratic equation. The R2 values (0.9967, 0.9997and 0.9970), adjusted
R2 values (0.9924, 0.9992 and 0.9970) and predicted R2 (0.9755, 0.9964 and 0.9829) are
close to one for SM, IO and MBC respectively (Table 4). This indicates a greater
correlation and acceptable agreement. The adequate precision value of 43.440, 131.528
and 65.570 was estimated for SM, IO and MBC respectively and it is due to the signal to
noise ratio that indicates an adequate signal. Such models will then be used to handle the
RSM model design space. For the second-order model, the differences in the means were
tested by ANOVA and the results are exhibited in Table 5. In general, F-values and P-
values determine the significance of all coefficients. The greater F values and the lower P-
values imply that the corresponding coefficients are more significant. The model is
suggested to be significant when the (Prob>F) is less than 0.1 and not significant if the
values are greater than 0.05 under specified conditions. However, the values of "Prob > F"
less than 0.05 indicates that model terms are significant. From the results of ANOVA, the
model terms such as A, B, C, AC, BC, A^2, B^2, C^2 were found to be significant. In
general, a regression model exhibits lack-of-fit when it fails to adequately describe the
functional relationship between the experimental factors and the response variable.
However, the “lack of fit P-value” was estimated as 0.514, 0.2413 and 0.1715 which is
greater than 0.05 and this suggests that lack-of-fit is not significant for the selected
variables. Further, this may be due to the quadratic model that has a good agreement and
greater correlation amongst selected variable for MY dye adsorption onto the adsorbents.
Similarly, the "lack of Fit F-value" of 0.9, 2.11 5.25 implies that the lack of Fit is not
significant relative to the pure error. There is a 24.13% chance for "lack of Fit F-value"
could be large due to noise. Non-significant lack of fit is good for the model to fit.
Regression analysis was carried out to fit the response functions, i.e. percentage
decolorization of MY dye. An empirical relationship between the response and input
variables were expressed as response surface reduced quadratic polymeric model
equations. The regression equation for the sequestration of MY dye by SM, IO and MBC
were predicted from the model and presented in Eq. (6), (7) and (8) respectively.
(6)
(7)
(8)
It can be noted from a statistically validated model that a randomly constructed series of
experiments has verified the theoretical adsorption potential of the model, apparently
confirming that the model equation fits well with the experimental response at 95 percent
certainty. Table 4 Response surface quadratic model summary statistics
Model R2 Adjusted
R2 Predicted
R2 Adequate precision
SD* CV* PRESS*
Salvinia molesta
0.9967 0.9924 0.9755 43.440 2.98 9.51 458.34
Iron oxide 0.9997 0.9992 0.9964 131.528 0.96 3.46 67.13 Magnetic biocomposite
0.9970 0.9970 0.9829 65.570 1.89 6.51 325.57
*SD: Standard deviation; CV: Coefficient of variation; PRESS: Predicted residual error sum of square
3.2. Effect of process variables and interactions Figures 1, 2 and 3 exhibit the 3D surface plot of sorption process of MY dye on to SM, IO
and MBC, respectively. The adsorbent concentration is one of the significant parameter
that determines the potential of adsorbent in removal of dye from aqueous solution. The
effect of dosage concentration with dye concentration against decolorization efficiency
were studied for all adsorbents and presented in Figures 1a, 2a, and 3a. The decolorization
efficiency has been found to be increased as the adsorbent dosage concentration increased
gradually. More binding site and increased surface area have been offered by increasing
the dosage concentration of adsorbent. Moreover, the figure shows clearly that by
increasing the amount of catalyst, the adsorption rate has increased, and that the amount
adsorbed per unit mass has decreased. Depending on the availability of active sites of
adsorbent in the aqueous solution, the extent of adsorption of MY dye varies. It is well
seen that the number of available adsorption sites increased by increasing the adsorbed
amount but the slump in adsorption capacity is fundamentally due to the sites remaining
unsaturated during the adsorption process. Another main parameter in the adsorption
process is the pH solution. The pH range was varied from 2 to 12 in this study. The effect
of pH with adsorbent dosage concentration against decolorization efficiency was studied
and shown in Figures 1b, 2b, and 3b. Maximum decolorization percentage was obtained
with increase in dosage concentration and decrease in solution pH. In addition to its
adsorbent dosage, it can be perceived that the influence of pH in the process of adsorption
of MY dye played a crucial function. All the prepared adsorbents SM, IO and MBC
showed similar scheme of decolorization for the range of pH. The increase in pH
decreases the decolorization percentage of MY dye for all the prepared adsorbents.
However, the percentage decolorization in the neutral and alkaline pH was significantly
lower than the acidic solution. The influence of pH on solid–liquid equilibrium could be
explained on the basis of ion exchange, i.e, the formation of –NH+3 group occurs at
acidic pH because of the excess availability of protons to protonate the amino group of
adsorbents. Thus increased protonation may contribute to the increased electrostatic
attraction amongst the negatively charged anions of dye molecules and positively charged
active sites of adsorbents which results in the greater percentage decolorization. In
addition, the rise in the solution pH from acidic to neutral or alkaline condition lead to the
engagement of negatively charged OH- ions on the surface of adsorbents and this results
in the decreases in adsorption efficiency due to the formation of repulsive force with
anionic dye at neutral and alkaline condition. Therefore, the dye decolorization decreases
since the external surface of the adsorbent does not produce exchangeable anions at
higher pH. As shown in Figures 1c, 2c, and 3c, decolorization efficiency of MY dye on to
all the prepared adsorbents decreased with increase in initial solution pH. A maximum
decolorization percentage of 96.42, 97.25 and 98.51 were obtained for SM, IO and MBC,
respectively. However, dye decolorization percentage increases with increase in adsorbent
concentration and decrease in solution pH.
Table 5 Analysis of variance (ANOVA)*
Source Sum of Squares
Mean Square
F Value
p-value Prob > F
Remarks
Salvinia molesta
Model 18615.80 2068.42 233.48 < 0.0001 significant A 718.39 718.39 81.09 < 0.0001 B 6.21 6.21 0.70 0.4300 C 12291.55 12291.55 1387.47 < 0.0001 AB 141.85 141.85 16.01 0.0052 AC 42.58 42.58 4.81 0.0645 BC 250.43 250.43 28.27 0.0011 A
2 16.10 16.10 1.82 0.2196
B2 15.45 15.45 1.74 0.2282
C2 5038.63 5038.63 568.76 < 0.0001
Residual 62.01 8.86 Lack of Fit 25.04 8.35 0.90 0.5141 not significant Pure Error 36.98 9.24 Cor Total 18677.81
Iron oxide
Model 18860.39 2095.60 2275.68 < 0.0001 significant A 495.34 495.34 537.90 < 0.0001 B 24.96 24.96 27.10 0.0012 C 11878.03 11878.03 12898.73 < 0.0001 AB 0.081 0.081 0.088 0.7751 AC 59.91 59.91 65.06 < 0.0001 BC 205.92 205.92 223.62 < 0.0001 A
2 12.62 12.62 13.70 0.0076
B2 23.73 23.73 25.76 0.0014
C2 6114.05 6114.05 6639.44 < 0.0001
Residual 6.45 0.92 Lack of Fit 3.95 1.32 2.11 0.2413 not significant Pure Error 2.49 0.62 Cor Total 18866.84
Magnetic biocomposite
Model 19041.32 2115.70 594.78 < 0.0001 significant A 452.85 452.85 127.31 < 0.0001 B 29.76 29.76 8.37 0.0232 C 12178.92 12178.92 3423.83 < 0.0001 AB 1.78 1.78 0.50 0.5019 AC 6.60 6.60 1.86 0.2152 BC 24.11 24.11 6.78 0.0353 A
2 9.31 9.31 2.62 0.1497
B2 12.64 12.64 3.55 0.1014
C2 6284.98 6284.98 1766.88 < 0.0001
Residual 24.90 3.56 Lack of Fit 19.86 6.62 5.25 0.1715 not significant Pure Error 5.04 1.26 Cor Total 19066.22