Optimization of Reactive Blue 21 removal by Nanoscale Zero-Valent Iron using response surface methodology

2.1 Materials and apparatus

Reactive Blue 21 (Molecular formula (C₄₀H₂₆CuN₁₀O₁₆S₆), Molecular Weight 1159.62 g/mol, ${\lambda }_{\text{max}}$ 614 nm), was obtained from ALVAN SABET company, Iran, and was used without any further purification (Fig. 1 shows the chemical structure of RB21.). Sodium Borohydride (NaBH₄), Iron (III) chloride (FeCl₃·6H₂O), Acetone, HCl and NaOH were obtained from Merck. Deionized distilled water (DDW) was used in all experiments. The amount of residual RB21 was analyzed by double beam UV–Vis spectrometer (Cary 100-varian).

Close

Figure 1

Molecular structure of Reactive Blue 21.

Export to PPT

2.2 Synthesis of Nano ZVI and characterization by SEM and XRD

Nano-scale Zero-Valent Iron particles used in this study were chemically synthesized in aqueous solution via the reduction of ferric iron by sodium borohydride (Frost et al., 2010; Liao et al., 2007). The experiments were carried out in the room temperature by applying N₂ atmosphere gas. In this work, NaBH₄ solution (0.3 M) was gradually dropped into FeCl₃·6H₂O solution (0.1 M) and the solution was mixed by a tunable mechanical stirrer at 415 rpm. The resulting reaction can be given as below (Eq. (1)) (Sun et al., 2007):

Black particles of NZVI, as a form of sedimentation, appeared immediately after introducing the first drop of NaBH₄ solution. The generated iron particles were separated by vacuum filtration and washed by DDW. In order to stabilize NZVI against immediate oxidation, the solid was washed at least two times with 98% acetone.

Morphology and the sizes of NZVI particles were identified by scanning electron microscope (SEM) (LEO-1455VP, England) as shown in Fig. 2. This image demonstrates that the iron particles are in the form of nano-spheres and they exist in contact with one another forming chains with diameters of <100 nm.

Close

Figure 2

Scanning electron micrograph (SEM) of iron nanoparticles: (a) 10.000 KX; (b) 20.000 KX.

Export to PPT

XRD analysis of newly-synthesized iron nanoparticles was carried out with Philips, PW3050, the XRD pattern is shown in Fig. 3 indicating that the Zero-Valent Iron has only a characteristic 2θ value of 44.7°, and no signals for iron oxides with a 2θ value of 36° were found. It also indicates that the ZVI is mainly seen in the sample.

Close

Figure 3

XRD pattern of Nano Zero-Valent Iron (NZVI) particles.

Export to PPT

The treatment was carried out by adding 0.1–0.9 g NZVI into 50 ml dye solution.

2.3 Experimental design and data analysis

In the present study, a five-level-four-factor Central Composite Rotatable Design, which is a widely used form of RSM, was employed for optimization of RB21 removal. The procedure of optimization has been represented in the form of a flowchart as shown in Fig. 4. A set of 30 experiments were designed to optimize by Nano ZVI catalyst. Four independent variables called catalyst dosage, pH, removal time and initial concentration were investigated and each variable in the design was studied at five different coded levels (Table 1). The variables were coded according to the following equation (Eremia et al., 2008):

Close

Figure 4

The procedure of using RSM for optimization.

Export to PPT

3.1 Model fitting and statistical analysis

Experimental data to remove RB-21 from wastewater are given in Table 3 along with the predicted values. The data were fitted with various models and their subsequent ANOVA showed that reaction of RB21 removal was most suitably described by quadratic polynomial model. The final model to predict the percentage of RB21 removal by NZVI is shown in Eq. (4):

F-value calculated from the data (19.89) was higher than tabular value. ANOVA and values of the factors are shown in Table 4. Coefficient of determination, R², always lies between 0 and 1 (Priya and Kanmani, 2011). (However, Masoumi et al., (2011) stated that if R² value lies between 0.90 and 1.00, the fitted regression equation is considered as a model having a high correlation.)

In this study, R² of the model was 0.9335. It indicates 93.35% of the experimental data and predicted data can be explained by the model. In addition, the value of adjusted determination coefficient Adj R² = 0.8866 was also satisfactory, confirming the significance of the model.

The value of coefficient of variation (CV = 3.69%) implied that there is good validity in the experimental data and the model. It can be considered reasonably reproducible if its CV is not greater than 10%. A lower value of CV clearly showed a high degree of precision. The model also showed adequate precision by measurement of signal to noise ratio. The ratio greater than 4 is desirable. Thus, a ratio of 13.620 indicated an adequate signal (Masoumi et al., 2011).

Fig. 5 indicates a plot of predicted value vs actual value of percentage removal from response surface design, where the coefficient of determination (R²) was (0.9335) and indicated a good agreement with model. In the meanwhile, the residual plots represent appropriateness of the model and correlation of the data (Fig. 6). The residuals are randomly scattered in a constant width band about the zero line. The root mean squares error (RMSE = 2.42) and the absolute average deviation (AAD = 2.30) were determined. The RMSE and AAD are calculated by the following equations:

Close

Figure 5

Scatter plot of predicted removal % value vs actual removal % value from RSM design.

Export to PPT

Close

Figure 6

Residual plot of model.

Export to PPT

3.2 Interaction between influencing factors

The three-dimensional response surface plots were used to determine the interaction between the four variables. The results are displayed in Fig. 7. The optimum situation of the relative variables is similar to the coordinates of central point in the upmost level in each of these figures. Fig. 7a represents that RB21 removal increases as time passes during reaction and decreases with the increase of dye concentration (while keeping other variables at a constant level). The reason is that in dye’s low concentrations, there is a close relation between surface active sites and the total dye molecules. Consequently, all molecules stick to NZVI surface and then are removed from the solution. However, there is not enough space for all molecules in high concentration of dye.

Close

Figure 7

Response surfaces plot. (a) Interaction plot of removal time (X₃) and dye concentration (X₄). (b) Interaction plot of pH (X₂) and dye concentration (X₄). (c) Interaction plot of catalyst amount (X₁) and dye concentration (X₄). (d) Interaction plot of catalyst amount (X₁) and pH (X₂).

Export to PPT

Fig. 7b shows the response surface plot of RB21 removal as a function of initial concentration of RB21 and pH. The pH was also one of the important factors in RB21 removal rate. More than 80% of dye removal by ZVI was observed around the neutral range of pH 4–9 (Shu et al., 2007). The removal rate increased at pH value below 9. This case can be ascribed to the fact that lower pH_ZPC (<pH_ZPC ∼ 8.3) is favorable for removal of anionic dye on the iron surface (Fan et al., 2009) but extremely basic or acidic conditions gave poor removal results. These results agree with previous reports on the ZVI transformation of dyes (Fan et al., 2009; Shu et al., 2007). As it is obvious from Fig. 7b RB21 removal increased at weak acidic range and decreased at extremely basic or acidic conditions within increase initial concentration of RB21. Fig. 7c demonstrates that with an increased amount of the catalyst and a decreased dye concentration, RB21 removal increases. Fig. 7d shows the effect of the catalyst amount and pH on RB21 removal. This diagram clearly states that around the neutral range of pH (lower pHzpc) with an increased catalyst amount, the efficiency of RB21 removal increases. The surface charge of NZVI under pHzpc is positive and it helps the removal of anionic compounds such as reactive dyes.

3.3 Optimization of reaction and model validation

To confirm the model’s adequacy for predicting maximum RB21 removal, three additional experiments were used. Such optimum conditions were performed and are now presented in Table 5.

The optimum reaction parameters were a catalyst amount of 0.5 g, initial concentration of 30 ppm, pH of 6.5, and removal time of 90s. All the optimum conditions can be used to produce high percentage Reactive Blue 21 removal. The desired value of the residual standard error (1.25) is also obtained from the optimum condition of RSM experimental design. Comparison of predicted and actual values revealed good correspondence between them, implying that the empirical model derived from design expert software can describe the relationship between the factors and outputs.