3.1 Characterization of nanometer Fe3O4 catalyst
Figure 1 shows the XRD patterns of the as-prepared sample. Compared with the data in the standard card JCPDS(NO.65-3107), it can be seen that there are no other miscellaneous peaks. The pure phase magnetite Fe3O4 with cubic inverse spinel structure is obtained, and the highest diffraction peak of the crystal plane is near 35.56° (2285). The as-synthesized sample has shown sharp diffraction peaks, high signal-to-noise ratio, and good crystallinity. Figure 2 depicts an SEM scan image and EDS analysis of the as-prepared sample. Figure 2 (a), (b), and (e) show the even distribution of Fe and O, with a ratio of about 3:4. This indicates that the as-prepared sample could be Fe3O4.
3.2 Feasibility Study
Parallel experiments were carried out in a Fenton-like system catalyzed by nano-Fe3O4 and Fenton system catalyzed by FeSO4, and the removal efficiencies of the two catalysts were compared under the same conditions. At room temperature of 22 ℃, the dosage of Fe3O4 nanoparticles and FeSO4 was firstly set to 2 g/L, which were added into three flasks containing AMA production wastewater. To the above mixture, the H2O2 solution was then added at various dosages of 5, 10, 25, 50, 75, 100, 125 and 150 mmol/L, respectively. The solution was continuously stirred for 360 mins. The indexes of AMA production wastewater treated by the two systems were determined, as shown in Fig. 3.
As shown in Fig. 3 (a), (b), and (c), the Fenton-like system of nano-Fe3O4 and the Fenton system of FeSO4 have obvious treatment effects on AMA, COD, and chromaticity in AMA production wastewater. The treatment efficiency of nano-Fe3O4 microspheres is obviously better than that of FeSO4 in terms of AMA degradation and COD removal, indicating that nano-Fe3O4 microspheres are investigable in AMA wastewater treatment process.
During the change of H2O2 solution dosage from 5 to 150 mmol/L in Fig. 3, the removal of AMA and COD first decreased and then increased whereas the chromaticity decreased to nearly complete removal. The best AMA removal efficiency was achieved at 75 mmol/L dosage of H2O2 solution. The removal rate of AMA, COD and Chromaticity reaches 99.74%; 95.00% and 99.80%, respectively. The concentration of AMA and COD reached 804.35 g/L and 240 mg/L, respectively. While the chromaticity was 78 pcu. With the continual increase of H2O2 dosage, the removal rate of AMA began to decrease. The experimental results revealed that increasing the dosage of H2O2 can produce more ·OH radicals, which eventually enhance the removal rate of AMA. However, the addition of too much H2O2 results in ·OH consumption, which generates less oxidizing HO2·. HO2 ·can not only react with H2O2 to prevent it from generating ·OH but also it can react with ·OH to prevent it from degrading AMA. Successful catalytic oxidation of AMA in the Fenton system was attained at 75 mmol/L, which could be considered as the optimum dosage of H2O2.
3.3 Single Factor Test
3.3.1 Effect of pH on Removal Efficiency
To study the effect of pH alteration on the Fenton removal of the AMA, COD and chromaticity, 2 g/L of the nano-Fe3O4 catalyst along with 75 mmol/L of H2O2 solution was subjected into the system at the reaction temperature of 22 ℃. The reaction time was set to 360 mins. The pH range of 2–9 was considered in the experiment. Figure 4 (a), (b), and (c) depicts the experimental results.
With the pH rising from 2 to 9, removal of both AMA and COD initially increases and then decreases gradually whereas the chromaticity removal gradually declines. The maximum removal efficiency of AMA, COD and Chromaticity appeared at the pH of 5 with respective removal rates of 99.99%, 94.61% and 99.98%. The concentration of AMA and COD were 15.76 g/L and 258 mg/L, respectively; while the initial chromaticity was 8 pcu. The results revealed that Fe3O4 nanospheres have good catalytic activity under weak acid conditions, but the catalytic activity is inhibited under alkaline conditions. Therefore, it can be stated that the best solution pH value for better degradation and removal of AMA in Fe3O4 + H2O2 Fenton system is 5.
3.3.2 Effect of Dosage of Nanometer Fe3O4 on Removal Efficiency
The effect of Fe3O4 catalyst dosage on the removal of AMA, COD and Chromaticity was investigated using 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5 and 4.0 g/L catalyst loading and the results were depicted in Fig. 5 (a), (b), and (c). Apart from pH set to 5, the H2O2 solution dosage along with the temperature and reaction time remains the same as in the case of pH alteration on the removal rate.
As the Fe3O4 catalyst dosage increases from 0.5 g/L to 4.0 g/L, the chromaticity removal rate also increases. The initial increase of Fe3O4 dosage to 1.0 g/L showed strong change in terms of the chromaticity removal rate (Fig. 5 (c)) while the removal rate is almost constant after 1.0 g/l catalyst loading. The removal efficiency of AMA and COD followed similar pattern. Both AMA and COD first increased and then decreased gradually. The maximum removal efficiency for either AMA, COD or Chromaticity was achieved at 1.5g/L Fe3O4 catalyst loading with removal rate of 99.99%, 98.90% and 99.98%, respectively. The concentration of AMA and COD were 10.33 g/L and 54 mg/L, respectively, while the chromaticity was 6 pcu. In line of the above, it can be stated that the best Fe3O4 catalyst dosage for the degradation of AMA in the Fe3O4 + H2O2-like Fenton system is 1.5 g/L.
3.3.3 Effect of Reaction Time on Removal Efficiency
The effect of reaction time on the removal of AMA, COD and Chromaticity was studied considering various reaction times (60, 120, 180, 240, 300, 360, 420 and 480 mins). The values for temperature, nano- Fe3O4 dosage, H2O2 solution and pH were kept constant. Figure 6 (a), (b), and (c) depicts the experimental results.
As the reaction time increases from 60 min to 480 min, the removal rates for chromaticity, AMA and COD also increase. When the reaction time reaches 360 min, the removal efficiency does not increase significantly or even decreases to a certain extent. The removal rate for AMA, COD and Chromaticity reaches 99.99%, 98.50% and 99.85%, respectively. The respective concentration of AMA and COD were 10.46 g/L and 72 mg/L while the chromaticity was 57 pcu. At reaction time of 360 mins, the optimum AMA degradation rate was attained by Fe3O4 + H2O2 Fenton system.
3.3.4 Influence of Temperature on Removal Efficiency
The effect of reaction temperature on Fenton removal of AMA, COD and Chromaticity was also studied by varying temperatures (0, 5, 10, 15, 20, 25, 30 and 35 ℃). The reaction time was set to 360min and other experimental parameters and conditions were kept constant. The results are shown in Fig. 7 (a), (b), and (c).
As the reaction temperature rises from 0 ℃ to 35 ℃, the chromaticity removal rate also rises first and is nearly completely removed after reaching 10 ℃. The removal rate of AMA and COD first increased and then decreased. The best removal rates were attained when the reaction temperature reached 25 ℃. The respective removal rate of COD and Chromaticity were 99.42% and 99.98%. At this reaction temperature, AMA is no longer detectable. COD concentration was detected as 28 mg/L whereas chromaticity was 8 pcu. The results indicated that the reaction temperature affects the activity of hydroxyl radicals. As the reaction temperature rises, the activity of free radicals can be enhanced. However, too high temperature could decompose hydrogen peroxide. Therefore, it can be deduced that the best reaction temperature for degradation and removal of AMA by Fe3O4 + H2O2 Fenton system is 25 ℃.
3.4 Response surface methodology (RSM)
3.4. 1 Design and Data
According to the results of a single-factor experiments, the dosage of nano-Fe3O4 (A), H2O2 (B), pH (C), and reaction time (D) were determined as the main influencing factors. The value range of each factor was 1.0 ~ 2.0 g/L, 50 ~ 100 mmol/L, 4 ~ 6, 300 ~ 420 min, respectively. The concentration of AMA is taken as the response value and recorded as the dependent variable Y. Based on the principle of Box-Behnken central combination design, an experimental formula with 4 factors and 3 levels was designed. See Table 2for the levels and codes of each factor.
Table 2
Levels of the factors taken for RSM
Coded value
|
Name
|
Units
|
Levels
|
-1
|
0
|
1
|
A
|
Fe3O4 dosage
|
g/L
|
1.0
|
1.5
|
2.0
|
B
|
H2O2 dosage
|
mmol/L
|
50
|
75
|
100
|
C
|
pH
|
-
|
4
|
5
|
6
|
D
|
Reaction Time
|
min
|
300
|
360
|
420
|
Taking the concentration (Y) of AMA as the response value, the experimental data are fitted by polynomial regression analysis, and a typical four-factor quadratic polynomial model can be obtained. The model is as follows:
(2)
Where β0 is the constant term representing the center point correction coefficient; Xi、Xj is the experimental factor; βi is the linear coefficient; βij is the quadratic term coefficient; βij is the interaction term coefficient; ε is the residual error of the constructed model.
3.4. 2 Regression equation and variance analysis
Based on the principle of Box-Behnken central combination design, the experimental scheme was designed, and a total of 29 groups of experiments were carried out. The experimental design scheme and results are shown in Table 3.
Table 3
Responses values for different experiment conditions
Run
|
Factors
|
Responses
|
(A) Fe3O4
|
(B) H2O2
|
(C) pH
|
(D) Reaction time
|
AMA
|
|
(g/L)
|
(mmol/L)
|
-
|
min
|
µg/L
|
1
|
1.5
|
75
|
5
|
360
|
10.46
|
2
|
1
|
75
|
5
|
300
|
968.47
|
3
|
2
|
75
|
6
|
360
|
124.79
|
4
|
1.5
|
75
|
6
|
300
|
527.34
|
5
|
1.5
|
75
|
4
|
420
|
122.67
|
6
|
2
|
50
|
5
|
360
|
357.64
|
7
|
1
|
50
|
5
|
360
|
1016.85
|
8
|
1.5
|
75
|
5
|
360
|
9.74
|
9
|
1.5
|
75
|
5
|
360
|
12.14
|
10
|
1.5
|
50
|
6
|
360
|
735.28
|
11
|
1.5
|
100
|
5
|
300
|
288.82
|
12
|
1.5
|
75
|
4
|
300
|
348.91
|
13
|
1.5
|
50
|
4
|
360
|
365.33
|
14
|
1.5
|
100
|
4
|
360
|
259.08
|
15
|
1
|
75
|
4
|
360
|
518.93
|
16
|
1
|
75
|
6
|
360
|
945.93
|
17
|
2
|
75
|
5
|
300
|
177.94
|
18
|
1.5
|
75
|
6
|
420
|
125.14
|
19
|
1.5
|
75
|
5
|
360
|
10.67
|
20
|
1.5
|
100
|
5
|
420
|
72.14
|
21
|
1.5
|
50
|
5
|
300
|
650.98
|
22
|
2
|
75
|
5
|
420
|
76.69
|
23
|
1.5
|
100
|
6
|
360
|
208.61
|
24
|
1.5
|
50
|
5
|
420
|
334.38
|
25
|
2
|
100
|
5
|
360
|
89.46
|
26
|
1
|
75
|
5
|
420
|
387.39
|
27
|
1.5
|
75
|
5
|
360
|
8.76
|
28
|
1
|
100
|
5
|
360
|
603.22
|
29
|
2
|
75
|
4
|
360
|
246.81
|
Using Design-Expert 10.0 software, the experimental data in Table 3 were fitted by quadratic multinomial regression, and the following equation was obtained.
Y = + 12375.257-4279.014*A-48.485*B-570.788*C-27.522*D + 2.909*AB-274.510*AC + 3.999*AD-4.204*BC + 0.0167*BD-0.733*CD + 1144.219*A2 + 0.351*B2 + 162.892*C2 + 0.030*D2 (3)
From Table 4, it can be seen that the F value of the model is 353.98, which is much greater than 1. This indicates that the model is significant. At the same time, the F values of the four influencing factors, nano-Fe3O4 dosage, H2O2 dosage, pH, and reaction time, are greater than 1, respectively. Similarly, they all have significant effects on the AMA removal rate, and according to the size of the F value, it can be judged that the order of their influence degree is nano-Fe3O4 dosage > H2O2 dosage > reaction time > pH. This implies the nano-Fe3O4 dosage is the key factor affecting AMA degradation. The corresponding P values of nano-Fe3O4 dosage (A), H2O2 dosage (B), pH (C), and reaction time (D) are all less than 0.05, which also show that they have a significant effect on AMA removal. In addition, the P values corresponding to the six interaction situations of AB, AC, AD, BC, BD, and CD are all less than 0.05, indicating the significant effect of the interaction between the dosage of nano-Fe3O4 (A), the dosage of H2O2 (B), pH (C) and reaction time (D). The determination coefficient of the model R2 was found to be 99.72%, which indicates that the probability of the model can explain the change of response value by 99.72%. Compared to the corrected determination coefficient (R2adj = 99.44%), only 0.56% of the response value cannot be predicted by the model. According to the coefficient of variation (CV = 6.85% < 10%), it can be stated that the reliability of the experiment and the accuracy of the model are high. The signal-to-noise ratio is equal to 60.304 > 4, which indicates that there is enough signal and the value is within a reasonable range.
Table 4
ANOVA for Response Surface Quadratic model
Source
|
Sum of
Squares
|
df
|
Mean
Square
|
F
Value
|
p-value
Prob > F
|
Source
|
Model
|
2.552E + 006
|
14
|
1.823E + 005
|
353.98
|
<0.0001
|
significant
|
A- Fe3O4 dosage
|
9.450E + 005
|
1
|
9.450E + 005
|
1835.25
|
<0.0001
|
|
B- H2O2 dosage
|
3.134E + 005
|
1
|
3.134E + 005
|
608.56
|
<0.0001
|
|
C-pH
|
54050.39
|
1
|
54050.39
|
104.97
|
<0.0001
|
|
D-Reaction time
|
2.834E + 005
|
1
|
2.834E + 005
|
550.35
|
<0.0001
|
|
AB
|
5288.93
|
1
|
5288.93
|
10.27
|
0.0064
|
|
AC
|
75355.74
|
1
|
75355.74
|
146.35
|
<0.0001
|
|
AD
|
57559.21
|
1
|
57559.21
|
111.79
|
<0.0001
|
|
BC
|
44188.24
|
1
|
44188.24
|
85.82
|
<0.0001
|
|
BD
|
2496.00
|
1
|
2496.00
|
4.85
|
0.0450
|
|
CD
|
7740.48
|
1
|
7740.48
|
15.03
|
0.0017
|
|
A2
|
5.308E + 005
|
1
|
5.308E + 005
|
1030.81
|
<0.0001
|
|
B2
|
3.126E + 005
|
1
|
3.126E + 005
|
607.02
|
<0.0001
|
|
C2
|
1.721E + 005
|
1
|
1.721E + 005
|
334.26
|
<0.0001
|
|
D2
|
74122.84
|
1
|
74122.84
|
143.95
|
<0.0001
|
|
Residual
|
7208.68
|
14
|
514.91
|
|
|
|
Lack of Fit
|
7202.46
|
10
|
720.25
|
463.28
|
<0.0001
|
significant
|
Pure Error
|
6.22
|
4
|
1.55
|
|
|
|
Cor Total
|
2.559E + 006
|
28
|
|
|
|
|
Note: P < 0.01 means the model is "highly significant", P < 0.05 means the model is "significant", P > 0.05 means the model is "not significant", model determination coefficient R2 = 0.9972, correction determination coefficient R2adj = 0.9944, prediction determination coefficient R2pred = 0.9838, coefficient of variation CV = 6.85%, signal-to-noise ratio = 60.304. |
3.4.3 Residual Analysis
Residual refers to the error between the predicted value and the actual value given by the model established by the software. Among them, the internalized residual is used to express the degree to which the standard deviation deviates from the actual and predicted response value, which is mainly shown graphically as to whether each data point is linearly distributed. Foreign student residuals are used to describe whether each group of data is an outlier relative to the fitted equation. As shown in Fig. 8a, the experimental data points are evenly distributed on both sides of the fitting straight line, and the data has no problem. This further confirms that the predicted value is close to the actual value. As shown in Figs. 8b and 8c, the data points are randomly distributed without any regularity, which proves the randomness of the experimental group. The residual values are all within the range of 3, indicating that there are no abnormal points in the fitting process, and the experimental operation is reliable. Figure 8d shows a linear relationship between the actual value of AMA concentration and the predicted value. The experimental data points are basically distributed on a straight line, which highlights that there is a good linear relationship between the predicted value and the actual value of the model. Therefore, this implies the model has good accuracy and persuasion.
3.4.4 Factor effect analysis
To further understand the removal of AMA by experimental factors and the interaction between nano-Fe3O4 dosage, H2O2 dosage, pH, and reaction time, a quadratic response surface diagram was drawn using software, as shown in Fig. 9.
As portrayed in Fig. 9, when one factor is fixed, the response value always increases first and then decreases with the change of another factor. This explains that each factor has an optimal value within the experimental level range. Any two factors taken as the central values, the contour lines corresponding to the response surface are elliptical rather than circular, which confirms the obvious interaction between the four factors. The optimal values of the model are 1.70 g/L of nano-Fe3O4 dosage, 53.52 mmol/L of H2O2 dosage, pH of 5.14, 388.97 min as the reaction time, and − 105. 83 g/L, the concentration of AMA. The negative concentration of AMA could be linked to the removal efficiency which reached the highest extent under the set condition. This highlights that AMA has been completely degraded and removed under this condition.
3.5 Degradation Mechanism
Figure 10 indicates mass spectrogram of AMA degradation in nano-Fe3O4 + H2O2 Fenton system. Before the reaction starts, only AMA was present in the solution with the mass-to-charge ratio (m/z) of 180. After 10min of Fe3O4 catalytic reaction, the compounds with m/z of 237, 196, and 154 appeared on the mass spectrogram. According to the intermediate product and accessory product 12 in the production process of AMA, the compound with m/z of 237 can be judged to be 2, 4-diaminoanisole while the compound with m/z of 196 is presumed to be the compound generated by ·O2 and ·OH radicals destroying the C bond at the para-position of the amino group in AMA. Subsequently, the acetamido functional group of the compound was oxidized by ·O2 and ·OH radicals to form a compound with m/z of 154. When the reaction proceeded for 30 min, new compounds were produced, which were compounds with m/z of 165, 141, 124, and 112, respectively. According to the AMA production method, the compound with m/z of 165 can be judged to be p-methoxy acetanilide. The acetamido functional group of p-methoxy acetanilide can then be oxidized by ·O2 and ·OH radicals to form p-amino anisole with m/z of 124. At the same time, the compound with m/z of 154 could be further oxidized to form a compound with m/z of 141 and resorcinol with m/z of 112.
With the change of reaction time from 30 min to 240 min, each substance was gradually degraded. Over time, various substances are eventually degraded into small molecular organic substances such as H2O and CO2. Only resorcinol with m/z of 112 was found in the final mass spectrogram, and it can be inferred that the possible degradation routes of AMA are the three routes in Figure 11.