3.1 Characterization of catalysts
The spectrum of the XRD associated with pure ZIF-8, pure NiFe2O4, and ZIF-8/ NiFe2O4 samples are illustrated in Fig. 3 (a-c) respectively. Figure 3a, the ZIF-8 characteristic peaks clearly appear at 2θ in 7.5°, 10.4°, 12.7°, 14.8°, 16.5°, 18.4°, and 19.7° which correspond to (011), (002), (112), (022), (013), (222), and (114) crystal planes, respectively, matched well with the published patterns [42]. The typical peaks of pure NiFe2O4 sample are detected at 18.84°, 30.41°, 35.68°, 37.55°, 43.26°, 53.89°, 57.37°, 62.91°, and 74.05° which can be indexed to the (111), (290), (311), (222), (400), (422), (511), (440), and (533) crystal planes of NiFe2O4, respectively. These peaks are consistent with the cubic spinel phase of NiFe2O4 [43]. The XRD pattern of the ZIF-8/NiFe2O4 composite confirms the successful growth of ZIF-8 on the surface of the NiFe2O4 particles.
Figure 4 shows the FTIR spectra of (a) ZIF-8, (b) NiFe2O4, and (c) ZIF-8/NiFe2O4. The broad band at about 3100 to 3500 cm− 1 of pure ZIF-8, pure NiFe2O4, and ZIF-8/NiFe2O4 is attributable to the –OH stretching of the adsorbed water molecules. In Fig. 4a, the absorption bands around 3132 and 2927 cm− 1 correspond to C-H from the aromatic ring and the aliphatic chain in H-MIM. The band appears at 1550 − 1350 cm− 1 and 1581 cm− 1 is ascribed to the vibration of the entire ring and the C = N stretching vibration, respectively. The wavenumber region between 1350 and 950 cm− 1 indicated various bands assigned to the in-plane bending of the ring and those seen at 754 and 689 cm− 1 are indexed to the out of plane bending. The peaks of synthesized ZIF-8 concur with the published reports [44–46]. The strong peak at around 594 cm− 1 of pure NiFe2O4 can be ascribed to the stretching vibrations of Fe –O bonds (Fig. 4b) [47]. The surface of ZIF-8/NiFe2O4 which includes characteristic peaks of ZIF-8 and NiFe2O4 confirms the successful synthesis of the nanocomposite materials (Fig. 4c).
Figure 5 shows SEM images of (a) pure ZIF-8, (b) NiFe2O4, and (c) ZIF-8/NiFe2O4 composite. The FESEM image of the ZIF-8/NiFe2O4 shows many ZIF-8 polyhedron crystals grow on the surface of NiFe2O4 nanoparticles, which allows easy separation from solution by a magnet.
The structural information of the ZIF-8/NiFe2O4 sample was further obtained using the TEM analysis (Fig. 5d). In Fig. 5d, NiFe2O4’s region is darker than the ZIF-8’s region as magnetic materials have a higher capacity for electron absorption. As shown in Fig. 5(d) the ZIF-8 nanoparticles were deposited on the surface of NiFe2O4.
According to Fig. 6, compared with pure ZIF-8, ZIF-8 doped NiFe2O4 composite showed much weaker fluorescence signals, proving that the incorporation of NiFe2O4 could effectively improve the inhibition of electron-hole recombination; therefore, it has the higher photocatalytic efficiency.
Figure 7 shows the magnetization curves of the pure NiFe2O4 and ZIF-8/NiFe2O4 nanocomposite measured by VSM in fields of ± 1.5 T at room temperature. It can be observed that the values of the saturation magnetization (Ms) of pure NiFe2O4 and ZIF-8/NiFe2O4 were 38.54 emu/g and 10 emu/g, respectively. Compared with pure NiFe2O4, synthesized ZIF-8/NiFe2O4 has a lower saturation magnetization, which is due to the presence of non-magnetic materials in the composites [48, 49]. Nevertheless, Ms of ZIF-8/ NiFe2O4 is still sufficient than those of reported magnetic photocatalysts [50–51]. Thus, it will enable separation and recovery of the photocatalyst magnetically from an aqueous solution using an external magnetic field.
3.2 Validation of response surface models and statistical analysis
The results of experiments under conditions specified by the CCD for photocatalytic degradation of MB using ZIF-8/NiFe2O4 are presented in Table 2. Based on the results of this table, the following second-order polynomial equation was obtained to express the relationship between the degradation percent of MB (Y) with catalyst dose (A), initial dye concentration (B), irradiation time (C), and pH (D) (Eq. 2):
Table 2
Experimental design conditions and responses of each experimental run
Run
|
Std
|
A
|
B
|
C
|
D
|
Response
|
|
|
|
|
|
|
Actual a
|
Predicted b
|
1
|
12
|
0.05
|
16
|
98
|
7.5
|
63.025
|
65.04
|
2
|
24
|
0.04
|
13
|
165
|
9.0
|
75.675
|
74.89
|
3
|
28
|
0.04
|
13
|
165
|
6.0
|
71.33
|
69.99
|
4
|
2
|
0.05
|
9
|
98
|
4.5
|
63.325
|
64.86
|
5
|
14
|
0.05
|
9
|
233
|
7.5
|
92.1701
|
91.62
|
6
|
9
|
0.02
|
9
|
98
|
7.5
|
60.35
|
60.74
|
7
|
29
|
0.04
|
13
|
165
|
6.0
|
70.46
|
69.99
|
8
|
19
|
0.04
|
5
|
165
|
6.0
|
88.975
|
90.19
|
9
|
20
|
0.04
|
20
|
165
|
6.0
|
49.661
|
49.79
|
10
|
3
|
0.02
|
16
|
98
|
4.5
|
43.34
|
44.26
|
11
|
13
|
0.02
|
9
|
233
|
7.5
|
61.7
|
62.91
|
12
|
5
|
0.02
|
9
|
233
|
4.5
|
56.375
|
54.72
|
13
|
17
|
0.01
|
13
|
165
|
6.0
|
20.7298
|
21.24
|
14
|
15
|
0.02
|
16
|
233
|
7.5
|
31.075
|
29.90
|
15
|
25
|
0.04
|
13
|
165
|
6.0
|
69.35
|
69.99
|
16
|
4
|
0.05
|
16
|
98
|
4.5
|
59.7
|
57.48
|
17
|
1
|
0.02
|
9
|
98
|
4.5
|
50.025
|
48.71
|
18
|
7
|
0.02
|
16
|
233
|
4.5
|
33.7
|
33.81
|
19
|
16
|
0.05
|
16
|
233
|
7.5
|
55.375
|
55.68
|
20
|
8
|
0.05
|
16
|
233
|
4.5
|
51.986
|
51.96
|
21
|
27
|
0.04
|
13
|
165
|
6.0
|
70.35
|
69.99
|
22
|
22
|
0.04
|
13
|
300
|
6.0
|
49.665
|
50.67
|
23
|
10
|
0.05
|
9
|
98
|
7.5
|
85.65
|
84.52
|
24
|
18
|
0.06
|
13
|
165
|
6.0
|
63.025
|
63.17
|
25
|
26
|
0.04
|
13
|
165
|
6.0
|
69.15
|
69.99
|
26
|
21
|
0.04
|
13
|
30
|
6.0
|
54.375
|
54.02
|
27
|
6
|
0.05
|
9
|
233
|
4.5
|
76.675
|
75.80
|
28
|
23
|
0.04
|
13
|
165
|
3.0
|
57.7
|
59.14
|
29
|
11
|
0.02
|
16
|
98
|
7.5
|
44.34
|
44.19
|
a Actual MB degradation determined by experiment. |
b Predicted MB degradation was calculated according to Eq. (1) |
At this stage, the analysis of variance (ANOVA) was applied to evaluate the significance and adequacy of the fitted model and their interactions on the removal efficiency of dye [52]. Statistically, the P-values less than 0.05 and large F-value indicates the significance of model terms [53]. The results of ANOVA showed in Table 3 suggested a quadratic model which was highly significant for MB removal because of its very high F-value (Fmodel= 248.26) and very low P-value (< 0.0001) [54]. As observed, the P-value related to the linear terms A, B, C, D and the quadratic term A2, C2, D2 and the AC, AD, BC, BD, CD interactions were less than 0.05, so they had a significant effect on the model and significant impacts on the MB removal efficiency. Other quadratic and interaction terms did not have a significant effect on the model. The lack of fit p-value in this model was 0.1206, suggesting that the lack of fit was not significant. In the other words, the proposed statistical model fits well. The R2 test was also used for the validation of the mathematical model. In this study, the value of R2 was 0.996. This indicates that 99.6% of the total variations for MB removal are explained by the independent variables and only about 0.4% of total variations were not explained by the model [55].
Adjusted R2 (Adj-R2) Value of 0.992 was very close to the corresponding value of R2 (i.e., the difference < 0.2). These values, which are quite close to one, indicate a satisfactory adjustment of the model with the experimental results. The C.V. is 2.47%, indicating that differences between actual and prognosticated results are unimportant. Moreover, the present model has an adequate precision ratio of 66.027 (it measures the signal to noise ratio), which is > 4 [56].
Assessed by the F-values of variables, the order of independent parameters that influenced the MB removal efficiency could be graded as: catalyst dose (A) > initial dye concentration (B) > initial pH (D) > irradiation Time (C).
3.2.1 Graphical analysis
Figure 8a shows the normal graph of the residuals. The residuals from the least squares fit are important for judging model adequacy. As shown, the data points are on a line and follow a normal distribution. Figure 8b presents the values of actual data in terms of the predicted results of the model. It can be seen that the points are close to the straight line (y = x), which indicated that actual data and data obtained from the models have a high degree of accordance. All these results show a high correlation and adequacy of the proposed model to predict the MB removal with ZIF-8/NiFe2O4 nanoparticles as a catalyst using the photocatalytic reactor.
which confirms that the error values between the actual and predicted values have zero mean and the regression models could be used to predict the MB removal from the initial experimental conditions.
Table 3
Analysis of variance (ANOVA) test for response function Y (MB degradation efficiency)
Source
|
Sum of Squares
|
Degree of freedom
|
Mean Square
|
F-value
|
p-value
|
Remark
|
Model
|
7616.14
|
14
|
544.01
|
248.26
|
< 0.0001
|
significant
|
A
|
2637.43
|
1
|
2637.43
|
1203.58
|
< 0.0001
|
|
B
|
2447.37
|
1
|
2447.37
|
1116.85
|
< 0.0001
|
|
C
|
16.87
|
1
|
16.87
|
7.70
|
0.0149
|
|
D
|
372.17
|
1
|
372.17
|
169.84
|
< 0.0001
|
|
AB
|
8.61
|
1
|
8.61
|
3.93
|
0.0674
|
|
AC
|
24.28
|
1
|
24.28
|
11.08
|
0.0050
|
|
AD
|
58.18
|
1
|
58.18
|
26.55
|
0.0001
|
|
BC
|
270.92
|
1
|
270.92
|
123.64
|
< 0.0001
|
|
BD
|
146.30
|
1
|
146.30
|
66.76
|
< 0.0001
|
|
CD
|
14.80
|
1
|
14.80
|
6.76
|
0.0210
|
|
A²
|
1259.33
|
1
|
1259.33
|
574.69
|
< 0.0001
|
|
B²
|
0.2952
|
1
|
0.2952
|
0.1347
|
0.7191
|
|
C²
|
509.46
|
1
|
509.46
|
232.49
|
< 0.0001
|
|
D²
|
15.16
|
1
|
15.16
|
6.92
|
0.0198
|
|
Residual
|
30.68
|
14
|
2.19
|
|
|
|
Lack of Fit
|
27.51
|
10
|
2.75
|
3.48
|
0.1206
|
not significant
|
Pure Error
|
3.17
|
4
|
0.7915
|
|
|
|
Cor Total
|
7646.82
|
28
|
|
|
|
|
3.3 Effect of variables as response surface and contour plots
In order to gain insight into the effect of the independent variables and their interactions on the response factor (removal %), the three-dimensional (3D) for the predicted responses were presented in Fig. 9 (a-e).
3.3.1 Effect of catalyst dose and irradiation time
Figure 9a shows the interactive effect of catalyst dose and irradiation time on the MB removal from the aqueous phase. Other parameters are in constant values (reaction time = 120 min and catalyst dose = 0.2 g/ 100 mL). As can be seen with the increase in the catalyst dose up to 0.05 g/100mL the removal efficiency increased and then decreased.
In fact, the increase in the dose of the catalyst increases available surface area and active sites on the catalyst surface. But, the increase in catalyst dosage beyond an optimum value leads to enhance turbidity, decrease light penetration and finally diminish the availability of active sites [57, 58]. As for irradiation time, it was found that the removal efficiency significantly increased by prolonging the exposure time of photocatalyst to the light. In fact, with increasing exposure time, the number of absorbed photons on the surface of photocatalyst becomes greater which boost the photocatalytic performance (produce more free radical hydroxyl concentration), however, the photocatalytic process was less efficient after 120 min probably due to the slow reaction of more complex organic compounds with OH− radicals [59].
3.3.2 Effect of catalyst dose and initial dye concentration
The combined effect of catalyst dose and initial dye concentration on dye removal is shown in Fig. 9b. The figure illustrates that the efficiency of removal increased with increasing catalysis dose and decreasing dye concentration. This is because, with increasing the initial dye concentration the ratio of surface active sites of the nanocomposite to the total dye molecules decreases which result in decrease in the degradation efficiency.
3.3.3 Effect of catalyst dose and pH
Figure 9c presents the response surface plot showing the effect of catalyst dose, pH, and their interactions on the MB removal, while the other variable was set at the middle value. In many photocatalytic reactions, the pH of the dye solution is an important factor as it directly influences the adsorption and degradation efficiency [60].
From this figure, the pH had a strong effect on the response and the MB removal efficiency was increased with an increase in pH value and increase of catalyst dose. The variation in the dye degradation with respect to the initial pH can be explained with the help of the zero point charge (pHZPC) of the ZIF-8/NiFe2O4. For synthesized ZIF-8/NiFe2O4 nanocomposite, the zero point charge (pHZPC) is determined to be 6.
At acidic pH (pH < pHPzc), the composite shows a positive surface charge that causes reduce the adsorption of MB on the photocatalyst due to the repulsion force between dye molecules and nanocomposite. But if the solution pH is over pHPZC, a relatively higher number of negatively charged sites on the catalyst surface enhance the sorption of dye [61]. In fact, the photocatalysis reactions accurse after the adsorption process of MB on the catalyst surface. The maximum removal efficiency was achieved at pH values around 6.5-7 and an exposure time of 120 min due to strong electrostatic attraction between the composite and cationic organic pollutants.
3.3.4 Effect of irradiation time and initial dye concentration
The interaction effect of irradiation time and initial dye concentration on removal efficiency (Fig. 9d) shows that the maximum removal efficiency was acquired at higher irradiation times and lower initial dye concentration. As illustrated time has a positive effect on degradation efficiency. The removal efficiency decreased with increasing the initial dye concentration. In fact, at higher dye concentration the adsorbed dye molecules hinder the direct interaction between the light and the surface of the catalyst led to a decrease in the amounts of reactive radicals produced on the surface of the catalyst. Therefore, a longer irradiation time was needed to increase the chance of reaction between radical species and MB molecules [62].
3.3.5 Effect of pH and initial dye concentration
The plot for the combined effect of the solution pH and initial dye concentration (Fig. 9e) shows that increasing initial dye concentration results in decreasing the removal efficiency, whilst it increases with increasing the value of pH. The increasing trend in dye degradation with increasing pH may be related to a greater tendency to absorb the dye molecule with the positive charges in the alkaline environment. In general, with the increase in pH the efficiency of photocatalytic oxidation increases [58], due to the availability of more number of hydroxyl ions to generate hydroxyl radicals for efficient degradation [63]. The obtained results illustrate that the optimum pH is within the limit of 6.5_7.5. This pH is in the range of wastewater effluent discharge standards and no need for the adjustment of pH after the treatment process.
3.4 Optimal conditions for photocatalytic decolorization of methylene blue
The optimization process was carried out to determine the optimum parameters for the photocatalytic decolorization process. According to the response surface and desirability functions, the optimum conditions for the highest percentage of the dye degradation were obtained. In this case, all variables were in the studied range and removal efficiency was defined as maximum. Maximum dye removal (100%) was predicted in optimum conditions by software under treatment conditions of 0.044 g/100mL catalyst dose, 5 mg/L initial dye concentration, 150 min irradiation time, and dye solution pH of 7.4 based on desirability function of 1.00. In order to compare the actual value with the predicted optimum value and control the presented model, a further experimental test was done at optimum conditions. The resulted decolorization was 98%, which was found to agree well with the optimum value predicted by the model. The relatively small error in the experimental and predicted values shows good agreement of the results obtained from values predicted by the models and the experimental values. These results confirmed the system to optimize the operational conditions using RSM for the photocatalytic degradation of MB under the experimental conditions was successful.