3.1. Experimental design
The model of LDPE pretreatment using REX was studied by response surface methodology. In the current study, the experimental runs were carried out based on the design plan proposed for the studied parameters (screws speed, operating temperature and concentration of SiO2). After each run, the crystallinity index, carbonyl index and weight loss (%) of LDPE were calculated and presented as responses for each run as shown in Table 2.
Table 2
Experimental matrix and observed responses for LDPE pretreatment in BBD
|
Independent variable
|
Dependent variable
|
Run
|
X1 (RPM)
|
X2 (⁰C)
|
X3 (wt%)
|
Y1
|
Y2
|
Y3
(%)
|
1
|
100
|
350
|
1
|
22.34
|
0.85
|
64.03
|
2
|
100
|
350
|
1
|
22.70
|
0.92
|
64.16
|
3
|
50
|
400
|
1
|
15.97
|
0.06
|
56.79
|
4
|
150
|
300
|
1
|
24.16
|
0.76
|
61.90
|
5
|
100
|
300
|
0
|
23.58
|
1.94
|
63.16
|
6
|
50
|
350
|
2
|
22.32
|
1.38
|
63.84
|
7
|
150
|
350
|
2
|
25.12
|
2.03
|
55.29
|
8
|
100
|
300
|
2
|
25.06
|
0.41
|
60.98
|
9
|
150
|
400
|
1
|
20.35
|
0.57
|
50.33
|
10
|
150
|
350
|
0
|
22.67
|
2.03
|
64.35
|
11
|
100
|
350
|
1
|
23.00
|
0.80
|
64.35
|
12
|
50
|
350
|
0
|
23.40
|
2.16
|
62.62
|
13
|
100
|
400
|
2
|
17.51
|
1.00
|
50.10
|
14
|
100
|
400
|
0
|
16.82
|
0.48
|
56.48
|
15
|
50
|
300
|
1
|
26.00
|
0.77
|
62.53
|
X1: Screws speed, X2: Operating temperature, X3: Concentration of SiO2, Y1: Crystallinity index, Y2: Carbonyl index, Y3: Weight loss (%) of LDPE |
The studied responses were then tested against different regression models to determine the best-fitting mathematical model and the significance of varying the process parameters. The quadratic model was chosen as the best fitting model for the studied responses in comparison to the other models. The relationship between the crystallinity index (Y1) and carbonyl index (Y2) and the studied parameters; screws speed (X1), operating temperature (X2), and Concentration of SiO2 (X3) is demonstrated in Table 3.
For crystallinity index (Y1), the coefficients of the quadratic model equation indicated that the increase in operating temperature led to a significant decrease in the crystallinity index of LDPE residues while the increase in the screws speed and concentration of SiO2 showed a positive effect. The screws speed’s interactions with both operating temperature and concentration of SiO2 also showed a significant positive efficacy on crystallinity index. On the other hand, the interaction between the operating temperature and concentration of SiO2 showed a negative effect on crystallinity index. Such results elaborate that all the studied factors had significant effects on the crystallinity index of treated LDPE residue. Despite the factorial levels’ values, all the studied factors led to the creation of amorphous LDPE causing a decrease in the crystallinity index of treated LDPE samples which had an original crystallinity index of 29.20. Thus, making the LDPE residue favorable for microbial biodegradation.
For carbonyl index (Y2), both the operating temperature and concentration of SiO2 showed a significant negative effect on the carbonyl index values while the increase in screws speed had a positive effect. Alternatively, the screws speed interaction with both the operating temperature and concentration of SiO2 showed significant positive effects on the carbonyl index. Thus, based on the obtained results, it can be indicated that the interactions of the studied factors led to an increase in LDPE oxidative degradation which was observed through the increase in the carbonyl index of the treated LDPE residues.
Percentage of LDPE weight loss (Y3) was used as a response to assess LDPE initial degradation after REX. In Table 3, it can be observed that all the independent variables and their interactions influenced LDPE weight loss significantly with a P-value exceeding 0.05. Additionally, the coefficients of the model equation showed that the increase in all the studied independent variables and their interactions above a certain level led to a decrease in LDPE weight loss. Such finding indicate that the weight loss of LDPE is not necessarily increased with high levels of screws speed, operating temperature or concentration of SiO2; yet, it requires careful adjustment of these factors' levels in order to obtain the desired initial degradation percentage.
Table 3: Statistical analysis of measured responses for LDPE pretreatment
Fitting model
|
Factors
|
Coefficient
|
P-value
|
ANOVA
|
Crystallinity index (Y1)
|
Intercept
|
22.68
|
|
F = 207.43,
R2 = 0.9925, Model P-value ˂0.0001,
P-value of lack of fit = 0.41
|
X1
|
0.58
|
0.0017
|
X2
|
-3.52
|
< 0.0001
|
X3
|
0.44
|
0.0053
|
X1X2
|
1.56
|
< 0.0001
|
X1X3
|
0.88
|
0.0012
|
X2X3
|
-0.2
|
0.1976
|
X12
|
0.79
|
0.0023
|
X22
|
-1.85
|
< 0.0001
|
X32
|
-0.09
|
0.5442
|
Carbonyl index (Y2)
|
Intercept
|
0.86
|
|
F = 247.60,
R2 = 0.9937, Model P-value
= ˂ 0.0001,
P-value of lack of fit = 0.62
|
X1
|
0.13
|
0.0010
|
X2
|
-0.22
|
< 0.0001
|
X3
|
-0.22
|
< 0.0001
|
X1X2
|
0.13
|
0.0045
|
X1X3
|
0.19
|
0.0007
|
X2X3
|
0.51
|
< 0.0001
|
X12
|
0.31
|
< 0.0001
|
X22
|
-0.63
|
< 0.0001
|
X32
|
0.73
|
< 0.0001
|
LDPE weight loss (%) (Y3)
|
Intercept
|
64.18
|
|
F = 1461.89,
R2 = 0.9989, Model P-value
= ˂ 0.0001,
P-value of lack of fit = 0.52
|
X1
|
-1.74
|
< 0.0001
|
X2
|
-4.36
|
< 0.0001
|
X3
|
-2.05
|
< 0.0001
|
X1X2
|
-1.46
|
< 0.0001
|
X1X3
|
-2.57
|
< 0.0001
|
X2X3
|
-1.05
|
< 0.0001
|
X12
|
-1.22
|
< 0.0001
|
X22
|
-5.07
|
< 0.0001
|
X32
|
-1.43
|
< 0.0001
|
X1: Screws speed, X2: Operating temperature, X3: Concentration of SiO2, Y1: Crystallinity index, Y2: Carbonyl index, Y3: Weight loss (%) of LDPE
The adequacy of the proposed model to describe the crystallinity index, carbonyl index and weight loss of treated LDPE residues was evaluated and the results are demonstrated in Table 3. A sequential test was performed and the obtained quadratic model F-values (207.43 for crystallinity index, 247.60 for carbonyl index and 14611.89 for LDPE weight loss) were large compared to other model terms values in the equation. Thus, the proposed experimental systems for all responses can be modeled effectively. Based on the statistics test, high coefficients of determination were observed for all studied responses. The adjusted R2 values were calculated to be 0.9925, 0.9937 and 0.9989 for crystallinity index, carbonyl index and LDPE weight loss respectively).
Analysis of variance (ANOVA) was also applied to determine the significance of the model at a 95% confidence interval. A model is said to be significant if the probability value (p-value) is ˂ 0.05. The p-values demonstrated in Table 3 for crystallinity index, carbonyl index and LDPE weight loss indicated that these responses fitted the model well. From the lack-of-fit test the response showed a highly desirable non-significant lack-of-fit (p˃0.1) with p-values of 0.41 for crystallinity index, 0.62 for carbonyl index and 0.52 for LDPE weight loss.
3.2. Response surface analysis
Response surface graphical plots were generated between the responses obtained for LDPE pretreatment and the studied independent variables to estimate the effect of combinations of these variables on the studied responses. The 3-D and contour plots for crystallinity index, carbonyl index and LDPE weight loss (%) are demonstrated in Figures 1, 2 and 3. Figure 1 illustrates the dependence of the studied responses on screws speed and operating temperature. It can be observed that high levels of operating temperature and low levels of screws speed resulted in a significant decrease in crystallinity index. On the other hand, increased carbonyl index and LDPE weight loss percentage were detected at 320 to 370⁰C temperature range along all levels of screws speed. The values of both responses started to decrease at higher temperatures indicating the importance of adjusting the levels of temperature to obtain desirable degree of degradation. Thus, based on the obtained results, low screws speed and moderately elevated temperatures are required for LDPE oxidative degradation.
Figure 2 shows the dependence of the studied responses on screws speed and concentration of SiO2. The lowest crystallinity index was observed at low levels of both screws speed and concentration of SiO2 while increased values of carbonyl index were observed at low levels of SiO2 concentration and along all levels of screws speed. On the other hand, high concentrations of SiO2 led to elevated LDPE weight loss along all levels of screws speed where SiO2 acted as a catalyst in the LDPE degradation process. Therefore, it can be indicated that low levels of both SiO2 concentration and screws speed are able to produce desirable results for the crystallinity index and carbonyl index with respect to biodegradation. Nevertheless, high concentrations of SiO2 is also necessary to increase the rate of LDPE degradation during the REX process producing a residue of lower mass that can be more easily handled during biodegradation.
Figure 3 demonstrates the dependence of the studied responses on operating temperature and concentration of SiO2. It can be observed that the increase in operating temperature led to a decrease in crystallinity index of LDPE residues despite the concentration of SiO2 which indicates that the temperature has the upper hand in controlling the crystallinity of the treated LDPE samples. The same behavior was observed on LDPE weight loss where varying the SiO2 concentration didn’t show a significant difference in the degradation of LDPE upon interaction with operating temperature. Moreover, the highest carbonyl index and LDPE weight loss (%) were observed at a temperature range from 300 to 360 ⁰C where further increase in temperature did not show a profound increase in both responses.
3.3. Optimization of LDPE pretreatment via REX technique
All three responses were optimized simultaneously using BBD optimization. Pretreatment optimum conditions were chosen with the aim of attaining maximum initial LDPE degradation and enhancement of residual LDPE biodegradation post pretreatment. Based on the BBD results, maximum LDPE weight loss was observed with LDPE residues of high crystallinity index and high carbonyl index. Alternatively, based on literature review, enhanced LDPE biodegradation can be achieved through high carbonyl index and low crystallinity index residues [21]. Bearing such necessity in mind, the pretreatment conditions were adjusted to attain minimum crystallinity index and maximum carbonyl index and weight loss of LDPE, as shown in Table 4. A total of 40 optimized solutions were obtained. The selected solutions were determined according to their success to attain an acceptable desirability greater than 0.5 for the studied responses. Two batch experiments were carried out for LDPE pretreatment using the optimized conditions and the three responses were evaluated to validate the predicted model factors and responses. The response values (predicted and observed) for the optimized conditions are recorded in Table 4. The model was proven to be validated since a fine agreement existed between the predicted and observed results. This indicates the success of the BBD for the evaluation and optimization of the proposed LDPE pretreatment process.
Table 4
The optimized LDPE pretreatment process with observed and predicted response values
Independent Variable
|
|
Optimized level
|
|
Optimized 1
|
Optimized 2
|
X1: Screws speed (RPM)
|
|
50
|
50
|
X2: Operating temperature (⁰C)
|
|
380
|
390
|
X3: Concentration of SiO2 (wt%(w/w))
|
|
0
|
2
|
Overall desirability
|
|
0.65
|
0.66
|
Dependent variables
|
Desirability
|
Optimized 1
|
Optimized 2
|
Expected
|
Observed
|
Expected
|
Observed
|
Y1: LDPE crystallinity index
|
Minimize
|
19.77
|
20.86
|
16.70
|
18.33
|
Y2: LDPE carbonyl index
|
Maximize
|
1.47
|
1.02
|
1.05
|
1.04
|
Y3: LDPE weight loss (%)
|
Maximize
|
59.94
|
56.32
|
57.21
|
54.23
|
Moreover, gel permeation chromatography (GPC) was performed to LDPE virgin and optimized REX treated residues before undergoing bio-adhesion testing. Analysis was carried out using Tosoh EcoSEC HT-GPC 220 (UK) using an established protocol for assaying. GPC analysis results are presented in Table 5 as values of the weight–average molecular weight (Mw), the number–average molecular weight (Mn), and polydispersity index (PDI). As shown in Table 5, a significant reduction in the Mw and Mn masses of LDPE valued by 5 folds for Mw and 3 folds in the case of Mn upon treatment using the proposed technique. Noticeably, no ultrahigh-Mw polymer was detected in the residual solid indicating the occurrence of polymer degradation reactions. The PDIs of optimized treated residues were also lower in comparison with virgin LDPE.
Table 5
GPC analysis of pretreated LDPE
Sample
|
Mw (103 g/mol)
|
Mn (103 g/mol)
|
PDI
|
Virgin LDPE
|
128.4
|
18.1
|
7.09
|
Optimized 1
|
31.1
|
6.4
|
4.85
|
Optimized 2
|
23.9
|
5.8
|
4.54
|
3.4. Bacterial adhesion and biofilm formation on optimized pretreated LDPE
Pseudomonas species were previously shown to induce biodegradation of a variety of in treated plastics(s) [22]. Recently a consortium with Bacillus spp. was demonstrated to perform favourable biodegradation of the polyester polymer polyethylene terephthalate [6]. Here, P. aeruginosa was selected along with two environmental isolates of Bacillus sp. as opportunistic pathogens with substantial abilities to form biofilms [23]. In the current study, biofilms were successfully formed on pretreated LDPE samples; Optimized 1 and Optimized 2, after 14 days of incubation. Notably, adhesion and biofilm formation were 8- and 4-fold lower on untreated LDPE control samples (Figure 4[a]). Weight loss or change in the carbonyl and crystallinity indices of the LDPE samples during the 14 days of incubation biofilm attachment was not detected, indicating that while effective bio-adherence and bio-amenability has been demonstrated, evidence of biodegradation has not been detected. Interestingly, the consortium exhibited between 2-3-fold lower ability to form biofilms on all samples in comparison to the pure P. aerugionosa culture. This is in contrast to the similar study of Roberts et al, that used an environmental consortium of similar composition, and maybe associated with antagonistic effects amongst species [24]. The fact that Optimized 1 material showed a remarkable ability to adsorb basic, positively charged crystal violate (CV) stain [25], regardless of the presence of the microorganisms (Figure 4[b]), is evidence that the REX treatment induced beneficial surface changes which strongly diminish the bio-inert characteristics of LDPE, supporting a route towards effective sustainable post-consumer LDPE biodegradation.