3.1. Characterization of the PANI / SiO2 SPME sorbent
To characterize the functional groups, FT-IR spectra was employed of the prepared sorbent. The FT-IR spectrum of Silica showed main characteristic peaks at 1112.85, 808.12, and 470 cm− 1 that were attributed to the of bending vibration for Si-O-Si, symmetric stretching, and bending vibrations of Si-O-Si, respectively. (Fig. 1a) [25]. It can be seen from Fig. 1b that PANI / SiO2 gave peaks at 1583.45 and 1508.23 cm− 1 which were attributed to the stretching mode of C = N and C = C respectively [24, 26] .
The structure and morphology of the PANI / SiO2 nanocomposite was examined using the SEM instrument. Figure 2b shows the SEM images of SiO2 and SiO2/PANI with different magnifications. The smooth surface and uniform morphology of SiO2 nanofibers is evident from Fig. 2(a-b). Polyaniline particles tend to accumulate during the polymerization process, which reduces the sorption capacity and porosity. The synthesis of silica nanoparticles by electrospinning showed that silica nanofibers provided a template to prevent the aggregation of polyaniline particles. As shown in Fig. (2-c), polyaniline nanoparticles are uniformly coated on the silica nanofibers surface, which leads to higher adsorptivity, more flexibility and higher permeability. The EDX spectrum was also carried out on the sorbent (Fig. 3). The elemental mass ratios of silicon, oxygen, carbon, and nitrogen were 6.4, 30.5, 46.57 and 15.94%, respectively. These results show a good accordance with the PANI particles, synthesized on the surface of silica nanofibers
3.2. Optimization of the extraction conditions using CCD
In order to achieve the highest extraction efficiency and high pre-concentration, obtain accurate results in the shortest time, the main parameters including extraction time, ionic strength, extraction temperature, stirring rate, and desorption conditions were optimized. Because the adsorption characteristics of the PANI/SiO2 fiber were unknown, desorption time and desorption temperature were optimized one-variable-at-a-time, to ensure that complete desorption of the adsorbed analytes achieved. For this purpose, desorption time and desorption temperature were optimized over the ranges of 1–5 min and 230–280°C, respectively. According to the results, the highest extraction efficiency for the proposed nanocomposite fiber were obtained in 1 min as the desorption time and at 280 ◦C as the desorption temperature. Then, the CCD design was employed to determine the optimal extraction conditions through RSM. Multivariate optimization has important advantages over one-at-a-time, such as reducing the number of experiments, studying the interaction between variables, which leads to faster more efficient optimization. Central composite design (CCD) is a member of the RSM, which is a way to identify optimal experimental conditions with a reasonable number of runs. In general, the number of experimental runs is given by 2k + 2 k + Cp (where k and Cp are the number of factors and the number of central points respectively). Statistical software of Minitab 17 (State College, PA, USA) was employed to generate the experimental matrix and to assess the results [27]. In this study, four extraction variables, including extraction time, sample temperature, ionic strength, and stirring rate were considered for the RSM-CCD optimization. For each variable, five levels (0, -α, +α, -1, and + 1) were assigned as the central points, low axial runs, high axial runs, the experimental variables and five levels for the CCD optimization are shown in Table 1. The experimental matrix and the responses (peak area) are listed in Table 2. Analysis of variance (ANOVA) was applied to ensure the accuracy of the proposed RSM model [28]. The obtained results are presented in Table 3. Based on the results, the p-values of the independent factors were < 0.05, while the p-values of the Lack-of-Fit were higher than 0.05, confirming that all variables were statistically significant with 95% confidence level. According to the ANOVA results, a second-order polynomial equation was fitted using the following equation:
Φ-C11 = 74075 + 1844 T + 8065 t − 33255 IS − 280.6 stirring rate − 23.09 T2 − 92.78 t2 + 2238 IS2 + 0.2496 stirring rate2- 36.11 T×t + 278.2 T×IS- 0.086 T×stirring rate- 150.1 t×IS − 0.990 time×stiring rate + 8.02 IS×stirring rate
The value of determination regression coefficients (R2 = 0.9935) also indicating that the polynomial model fits well. Besides, predicted-R2 and adjusted-R2 were 0.9585 and 0.9866, respectively, which showing a desirable value for the statistical model validation.
The 3D response surfaces were created to obtain the optimal values and interactive effects of the independent variables, as showed in Fig. 4 for ϕ-C13. The optimal values for extraction temperature extraction time, ionic strength and stirring rate were determined to be 62 ◦C, 15 min, and 8%, and 600 respectively. The interaction between the variables is shown using the charts of the interruption. Figure 5 shows the main effects and interaction effects of the factors on an extraction of ϕ12 by PANI / SiO2 fiber nanosorbent. Figure 6 shows the normal probability plot for ϕ-C13; it is clear that the distribution of residuals is normal, and the model satisfies the assumptions of the analysis of variance.
Table 1
Experimental factors and levels used in the CCD model.
Variable
(Symbol, Unit)
|
|
Levels
|
|
|
|
Low axial (-α)
|
Low factorial (− 1)
|
Center point (0)
|
High factorial (+ 1)
|
High axial (+α)
|
Sample temperature (T, ◦Ϲ)
|
20
|
35
|
50
|
65
|
80
|
Extraction time (t, min)
|
5
|
18
|
31
|
44
|
57
|
Ionic strength (IS, w/v%)
|
0
|
2
|
4
|
6
|
8
|
Stirring rate (SR, rpm)
|
0
|
200
|
400
|
600
|
800
|
Table 2
The CCD matrix and the obtained data at different levels of the experimental factors.
RunOrder
|
T
|
t
|
IS
|
SR
|
Φ-C11
|
Φ-C12
|
Φ-C13
|
Φ-C14
|
1
|
50
|
31
|
4
|
400
|
79398
|
69865
|
70572
|
45990
|
2
|
50
|
31
|
4
|
800
|
85087
|
77143
|
60665
|
5955
|
3
|
50
|
31
|
8
|
400
|
109937
|
70790
|
45432
|
52234
|
4
|
20
|
31
|
4
|
400
|
80841
|
3036
|
28173
|
8394
|
5
|
50
|
31
|
4
|
0
|
160257
|
31989
|
12061
|
38831
|
6
|
50
|
5
|
4
|
400
|
30665
|
32696
|
34758
|
7211
|
7
|
50
|
31
|
0
|
400
|
127162
|
23890
|
9266
|
20576
|
8
|
80
|
31
|
4
|
400
|
43069
|
69174
|
4442
|
41634
|
9
|
50
|
57
|
4
|
400
|
9373
|
10056
|
6930
|
33742
|
10
|
50
|
31
|
4
|
400
|
84056
|
68756
|
80125
|
44423
|
11
|
35
|
44
|
2
|
600
|
79892
|
16056
|
43354
|
40214
|
12
|
50
|
31
|
4
|
400
|
79292
|
70056
|
77908
|
59662
|
13
|
65
|
18
|
2
|
600
|
60392
|
61844
|
12047
|
24824
|
14
|
65
|
44
|
2
|
200
|
78756
|
48682
|
12695
|
55905
|
15
|
65
|
18
|
6
|
600
|
79057
|
152521
|
80362
|
70036
|
16
|
65
|
18
|
6
|
200
|
97702
|
36517
|
55546
|
74432
|
17
|
35
|
44
|
6
|
600
|
49513
|
20493
|
35915
|
30056
|
18
|
35
|
44
|
2
|
200
|
120768
|
48120
|
48901
|
70307
|
19
|
35
|
18
|
6
|
200
|
80189
|
28957
|
43758
|
40602
|
20
|
50
|
31
|
4
|
400
|
75500
|
70073
|
80065
|
59964
|
21
|
65
|
18
|
2
|
200
|
92056
|
13994
|
7872
|
32669
|
22
|
65
|
44
|
2
|
600
|
34954
|
45165
|
8126
|
40569
|
23
|
35
|
18
|
2
|
600
|
75188
|
50612
|
72794
|
12152
|
24
|
35
|
18
|
6
|
600
|
60420
|
80696
|
73946
|
26481
|
25
|
65
|
44
|
6
|
600
|
35193
|
110036
|
38258
|
37336
|
26
|
50
|
31
|
4
|
400
|
79809
|
75056
|
79656
|
57955
|
27
|
50
|
31
|
4
|
400
|
75418
|
66557
|
85015
|
59111
|
28
|
35
|
18
|
2
|
200
|
110425
|
44925
|
43669
|
13745
|
29
|
35
|
44
|
6
|
200
|
77456
|
5243
|
2050
|
50189
|
30
|
65
|
44
|
6
|
200
|
69056
|
51745
|
30056
|
65384
|
Table 3
ANOVA for the suggested CCD model.
Term
|
Φ-C11
|
Φ-C12
|
Φ-C13
|
Φ-C14
|
Coef
|
T
|
P
|
Coef
|
T
|
P
|
Coef
|
T
|
P
|
Coef
|
T
|
P
|
T
|
-7593
|
-10.49
|
0.000
|
14528
|
12.02
|
0.008
|
-6954
|
-5.36
|
0.000
|
7662
|
12.01
|
0.000
|
t
|
-6351
|
-8.77
|
0.000
|
-7075
|
-5.71
|
0.005
|
-9429
|
-7.26
|
0.000
|
6170
|
9.67
|
0.000
|
IS
|
-5762
|
-7.96
|
0.000
|
10442
|
8.42
|
0.000
|
7615
|
5.87
|
0.000
|
6977
|
10.94
|
0.000
|
SR
|
-16756
|
-23.15
|
0.000
|
14565
|
11.75
|
0.003
|
9061
|
6.98
|
0.000
|
-7805
|
-12.24
|
0.000
|
T×T
|
-5195
|
-7.67
|
0.000
|
-6634
|
-5.72
|
0.000
|
-13501
|
-11.12
|
0.000
|
-4587
|
-7.69
|
0.000
|
t×t
|
-15679
|
-23.16
|
0.000
|
-10316
|
8.89
|
0.000
|
-12367
|
-10.18
|
0.000
|
-5721
|
9.59
|
0.000
|
IS×IS
|
8953
|
13.22
|
0.000
|
-3825
|
-3.30
|
0.005
|
-10741
|
-8.84
|
0.000
|
-1739
|
-2.91
|
0.011
|
SR×SR
|
9984
|
14.74
|
0.000
|
-2019
|
1.74
|
0.104
|
-8487
|
6.99
|
0.000
|
-5242
|
-8.79
|
0.000
|
T×t
|
-7041
|
-7.94
|
0.000
|
6627
|
4.36
|
0.001
|
2328
|
1.46
|
0.165
|
-6285
|
-8.04
|
0.000
|
T×IS
|
8347
|
9.41
|
0.000
|
12841
|
8.46
|
0.000
|
13533
|
8.51
|
0.000
|
55144
|
6.59
|
0.000
|
T×SR
|
-259
|
-0.29
|
0.774
|
11126
|
7.33
|
0.000
|
-3438
|
-2.16
|
0.048
|
645
|
0.83
|
0.423
|
t×IS
|
-3904
|
-4.40
|
0.001
|
-6114
|
4.03
|
0.001
|
-7752
|
-4.87
|
0.000
|
-9512
|
-12.18
|
0.000
|
t*SR
|
-2573
|
-2.90
|
0.012
|
-11457
|
-7.54
|
0.000
|
-3522
|
2.21
|
0.044
|
-4103
|
-5.25
|
0.000
|
IS*SR
|
3210
|
3.62
|
0.003
|
27916
|
9.19
|
0.000
|
4618
|
2.90
|
0.012
|
-739
|
-0.95
|
0.360
|
3.4. Analytical figures of merit
Analytical figures of merit of the purpose method, involving linear dynamic ranges (LDRs) relative standard deviations (RSDs), and detection limits (LODs) for the determination of four LABs in aqueous samples were evaluated. The results are showed in Table 4. The calibration graphs were linear in the ranges of 0.05-12 µg mL− 1 for Φ-C11 and Φ-C13 and 0.02-12 µg mL− 1 for Φ-C12 and Φ-C14 with linear regression coefficients greater than 0.9947. The LODs correspond to the analyte amounts for which the signal-to-noise ratio was equal to 3 were found to be in the range of 0.4–0.9 ng mL− 1. The relative standard deviations (RSDs, n = 6) values for a single fiber (repeatability) were determined 4.8–7.3%. The inter-fiber RSDs (reproducibility) for three randomly selected fibers were in the range of 9.2–12.4%. For further evaluation of the reliability and applicability, the analytical performances of the developed method were compared with some similar the previous studies [23, 29, 30], reported on separation and determination of LABs (Table 5). The results clearly showed that the proposed method has a wider LDRs and lower LODs and a compared to the mentioned methods.
Table 4
Analytical figures of merit for the analysis of LABs using the PANI/SiO2 fiber nanosorbent.
Analytes
|
LDR (µg ml− 1)
|
Equation
|
R2
|
LOD
(ng mL− 1)
|
RSD (%)
|
Intra-fiber(n = 6)
|
Inter-fiber (n = 3)
|
Φ-C11
|
0.05-12
|
y = 8672.1x + 14527
|
R² = 0.9947
|
0.8
|
6.9
|
9.2
|
Φ-C12
|
0.02-12
|
y = 6636.5x + 9597.8
|
R² = 0.9961
|
0. 4
|
4.8
|
10.2
|
Φ-C13
|
0.05-12
|
y = 5044.7x + 9775.8
|
R² = 0.9956
|
0.4
|
5.6
|
11.2
|
Φ-C14
|
0.02-12
|
y = 3327.9x + 8989.5
|
R² = 0.9961
|
0.9
|
7.3
|
12.4
|
Table 4
Analytical figures of merit for the analysis of LABs using the SiO2/PANI fiber nanosorbent.
Method
|
RSD (%)
|
LOD
|
LDR
|
Matrix
|
Ref.
|
LPME-HPLC-UV
|
12–17
|
0.7–4.6 (µg L− 1)
|
10–50 (µg L− 1)
|
Aqueous samples
|
[29]
|
SPE-HPLC-UV
|
2.4–5.6
|
0.013–0.021 mg L− 1
|
0.5–100 (mg L− 1)
|
Aqueous samples
|
[23]
|
In-tube SPME-HPLC-UV
|
1.6–12
|
0.02–0.1 µg L− 1
|
0. 2–25 (µg L− 1)
|
Aqueous samples
|
[30]
|
INCAT SPME-GC-FID
|
5.3–16.9
|
0.5-1 ng L− 1
|
0.01-10( ng L− 1)
|
wastewater
|
[31]
|
HS-SPME-GC-FID
|
4.8–12.4
|
0.4–0.9 ng mL− 1
|
0.05-12 (µg mL− 1)
|
wastewater
|
This work
|
3.5. Real sample and matrix effect
To evaluate the reliability and applicability of the developed method, it was employed for the analysis of LABs in three real wastewaters and the matrix effect (ME). One sample (#1) was of a municipal wastewater (Khoramabad, Iran) and two samples (#2 and #3) were from the Khoram River passing through the city of Khoramabad. All samples were collected and stored according to a standard sampling method[32]. In order to check the matrix effect, first each sample was analyzed under optimal conditions and then standard solution was added to each sample at three different concentration levels, one close to LOQ define it, middle, and at the highest concentration level of the calibration curve. The results are showed in Table 6. Matrix effect was determined using the following formula:
$$ME\%=\frac{(C-B)}{A}\times 100$$
where A is the added (spiked) concentration, B is the initial concentration of real sample, and C is the concentration of the fortified real sample. The results showed that the average ME values were distributed over the range 95.40–100.72%, it demonstrated that the matrix effects were not significant.
Table 6
The study of matrix effect on the extraction efficiency of the developed propose procedure.
Sample
|
Added (µg mL− 1)
|
Determined (µg mL− 1)
|
ME (%)
|
Φ-C11
|
Φ-C12
|
Φ-C13
|
Φ-C14
|
Wastewater
|
0
|
0.1(3.1)a
|
0.12(2.4)
|
0.08(4.3)
|
0.05(3.2)
|
-
|
0.2
|
0.29(1.9)
|
0.32(2.50
|
0.27(3.1)
|
0.25(2.3)
|
97.5
|
6
|
5.78(2.6)
|
5.65(3.5)
|
6.01(3.1)
|
5.79(3.2)
|
95.40
|
10
|
10.01(2.5)
|
9.56(2.1)
|
9.89(1.5)
|
10.01(3.)
|
97.8
|
River water #1
|
0
|
0.04(3.1)
|
0.05(3.6)
|
n.d.b
|
n.d.
|
-
|
0.2
|
0.25(2.5)
|
0.23(2.4)
|
0.18(2.7)
|
0.199(2.9)
|
96.13
|
6
|
6.01(1.8)
|
6.01(2.8)
|
5.88(3.6)
|
5.47(4.1)
|
97.00
|
10
|
10.01(2.3)
|
9.56(3.8)
|
9.89(4.6)
|
10.1(3.2)
|
98.68
|
River water #2
|
0
|
0.03(2.5)
|
0.01(5.2)
|
0.01(2.4)
|
0.018(3.1)
|
-
|
0.2
|
0.21(4.1)
|
0.19(3.1)
|
0.25(3.5)
|
0.24(2.1)
|
98.75
|
6
|
6.08(2.3)
|
6.01(2.8)
|
6.08(3.1)
|
6.07(4.6)
|
100.72
|
10
|
10.03(3.1)
|
9.56(1.8)
|
9.79(4.2)
|
10.12(3.9)
|
98.58
|
aThe numbers in parentheses refer to RSD% obtained by three replicated analyses.
bNot detected.