3.1. Evaluation of the Model Fitting and the Statistical Analysis
Error! Reference source not found. illustrates the design matrix and the outcomes of experiments assessing the effectiveness of MB dye degradation in the microreactor over a treatment period of 10 minutes. Utilizing a Central Composite Design (CCD), the Design-Expert software developed a quadratic model based on the experimental data in
Table 2, focusing on two independent variables: voltage (X
1) and pH (X
2). The resulting models demonstrated a proficient capability in predicting the dye decomposition rate. Additionally, Analysis of Variance (ANOVA) was employed to evaluate the quadratic models, further confirming the significant correlations between the independent variables and the observed responses. The best-fit regression model, which was established based on actual factors for the response parameter, was determined by Design-Expert as follows:
Table 3 presents the results of the analysis of variance (ANOVA) that was used to test the significance of the developed models. With an F value of 29.59 and the corresponding p value of 0.0001, the model was considered significant according to the goodness-of-fit tests and there was a very low chance (0.02%) that the F value could occur due to noise. In addition, the insignificant lack of fit analysis results, with an F value of 4.27 and the corresponding p of 0.0972, indicated the adequacy of the model. The coefficients of determination (R
2) adjusted R
2 and predicted R
2 for decolorization rate were 0.9531, 0.9226, and 0.7384 respectively. The difference between adjusted R
2 and predicted R² is less than 0.2, meaning that the predicted R
2 is in reasonable agreement with the adjusted R
2. Besides, the linear correlations between the observed and predicted data for the decolorization rate were also obtained as shown in
Figure 2a. The R² of 0.9531 implied that the model of decolorization rate has a high correction and could explain 95.31% of the total variation. Coefficients of variation for decolorization rate (14.68%) indicated clear agreement between the experimental and model results. Moreover, the adequate precision value (15.11, which measures the signal to noise ratio) for the model was larger than 4.0, indicating that the signals of the model were all adequate. In conclusion, all these statistical results showed that the developed model was able to predict the experimental data and adequately describe the relationship between the variables and responses. To be more specific, this quadratic model was validated for describing the decolorization rate under different pH and voltage within the range used in this study.
Figure 2b presents a surface response plot illustrating the decolorization rate as a function of pH and voltage. These plots are instrumental in determining the optimal conditions for the response variable within the selected experimental ranges. Analysis reveals that the maximal decolorization efficiency, quantified at 94.0%, was achieved at an optimal pH of 4 and a voltage of 33 kV. The figure clearly showed a substantial enhancement in the decolorization rate with increasing voltage, underscoring its pivotal role in plasma treatment. Furthermore, comparing experimental runs 6 and 9 in
Table 2 indicated that the decolorization rate improved from 36% to 61% with a voltage increment from 27 kV to 33 kV. This phenomenon is attributable to the increased electric field intensity between the electrode and the solution surface, which significantly intensifies the formation of active species within the plasma streamers, thus facilitating their diffusion into the liquid phase.
Equation 3 presents a coefficient of -0.2627 for the pH term, signifying a substantial inverse relationship with the degradation rate. Such a coefficient suggests that an acidic environment enhances the rate of degradation more effectively than neutral or alkaline conditions. This correlation is also graphically illustrated in
Figure 2b. MB degradation was lower than 40% at a solution pH of 9, and it was increased to 94% at a solution pH of 4.
The enhanced degradation capability with decreasing pH is primarily due to the following reasons. First, the pH of the solution critically impacts the ionization potential of target pollutants, thereby affecting their reactivity and degradation. In aqueous environments, MB predominantly exists in its cationic form (MB
+). Under acidic conditions, MB transitions into a protonated form (MBH
2+), which alters its electronic structure. The increased protonation of the dimethylamino groups in acidic media enhances the susceptibility of the methyl groups to dissociation [
14]. This protonation potentially facilitates the oxidative degradation of MB, making acidic conditions more conducive to its breakdown. Second, the detrimental impact of the hydroxyl anion (OH
−) in alkaline conditions may interfere with the effective reaction of the hydroxyl radical (·OH) [
15]. In addition, the decomposition of hydrogen peroxide (H
2O
2) leads to the formation of hydroperoxide anion (HO
2−), which can act as a quenching agent for ·OH [
16]. It is important to note that the superoxide anion (O
2−) generated in Equation 5 is a relatively weak nucleophile and reducing agent, which can reduce MB to a colorless reduced form but cannot decompose the MB molecules.
Moreover, the equilibrium between H
2O
2 and ·OH is pH-dependent, being more favorable in acidic solutions. The electrons generated by plasma can interact with water to produce hydrogen radicals (·H), which can react with oxygen to form hydroperoxyl radicals (·HO
2) and superoxide anion radical (
) intermediates (Equation 6-7) [
17]. These intermediates can enhance the generation of H
2O
2 production (Equations 8–10) [
18]. Furthermore, in acidic conditions, H
2O
2 can decompose into ·OH (Equations 11), which possesses an exceptionally high oxidative potential (E
o) and ranking as the most potent among available oxidizing agents (only surpassed by fluorine). Thus, it can be concluded that a reduction in pH significantly enhances the degradation efficiency of MB.
3.3. Mechanism Analysis
Wastewater treatment through plasma-induced reactions is facilitated by three distinct methods, each influencing the design of the reactors used [
19]. The direct discharge approach involves immersing electrodes in water, creating a filamentary streamer discharge at high electric fields (around 1 MV/cm), which generates reactive oxygen species and produces OH radicals, hydrogen, and hydrogen peroxide. This method also induces shockwaves and UV radiation, aiding in the decomposition of organic contaminants and microorganisms. An alternative, the indirect discharge approach, generates plasma above the water surface, relying on the diffusion of plasma-produced species into the water, a process governed by Henry's law and requiring lower initiation voltage. Lastly, the bubbling method introduces plasma within injected bubbles in the water, effectively increasing the contact surface area between plasma and liquid and allowing for tailored chemical reactions, illustrating the diverse methodologies underlying plasma water purification systems. In this study, the indirect discharge method was used, since the gas phase plasma was more energy efficient [
20].
In this indirect discharge process, plasma is generated between the anode tip and the water surface, forming a bright streamer discharge. This plasma interacts with the water surface, causing slight movements or small ripples due to the electro-hydrodynamic effect, which leads to surface deformation [
21]. The primary mechanism for degrading pollutants in water through non-thermal plasma involves the production of a variety of reactive species after the collisions between accelerating electrons and neutrals. Primary reactive species with fleeting lifespans (1-3 μs) are formed in the gas phase immediately after the collision, including ionized neutrals and gas (M
+), excited neutrals and gas (M
*), N, O, atomic H, NO, and O
2*– [
22,
23,
24]. Then, some reactive species generated in the plasma process are subject to immediate radiative decay, while others would react with additional reactive species, neutral molecules, and water to form secondary reactive species like hydrogen peroxide (H
2O
2), nitrogen dioxide (NO
2), nitric oxide (NO), and ozone (O
3) in the surrounding air [
25,
26]. These secondary reactive species, once generated in the gas phase, would migrate into the liquid phase or other substrates, where the formed tertiary reactive species expand their lifetimes from milliseconds to several days. Tertiary reactive species are more stable, including O
3, H
2O
2, nitrate (NO
3-), peroxynitrite (ONOO
-), and nitrite (NO
2-).
Figure 4 visually summarizes the reactive species formation process from the discharge region through the gas phase to the target substrate.
In the plasma-treated substrate, the electrolytic dissociation of water is also facilitated by plasma-generated electrons, yielding H
2O
2 through Equations 13-14. Hydrogen peroxide is a stable oxidizing agent formed not only via direct electron impact with water molecules but also through the dimerization of hydroxyl radicals. The in-situ generation of H
2O
2 serves as a proxy for the presence and reactivity of hydroxyl radicals within the plasma-mediated process [
27]. Additionally, the peroxynitrous acid, a reactive nitrogenous intermediate, undergoes homolysis to yield ·NO and ·OH [
28], which are integral to the oxidative degradation pathway.
The reactive species produced can engage in various mechanistic pathways when interacting with organic dye molecules. These pathways include electrophilic addition, hydrogen abstraction, and initiation of radical chain reactions. Photochemical dissociation by ultraviolet (UV) photons, as well as electron and ion-induced fragmentation, contributes to the multitudinous routes of MB molecular decomposition. UV can also dissociate hydrogen peroxide molecules present in aqueous solution to further improve the formation of highly reactive hydroxyl radicals [
29]. This results in the disruption of conjugated chromophoric systems and the oxidative opening of aromatic ring structures, ultimately leading to the conversion into innocuous end-products like carbon dioxide and water.
The initial step of MB degradation is characterized by the homolytic cleavage of the N−CH
3 bond, which is the least energetically stable with a bond dissociation energy of 70.8 kcal/mol [
3]. This process yields methyl radicals that are subsequently oxidized to methanol (CH
3OH), formic acid (HCOOH), or formaldehyde (HCHO). The carbon-sulfur (C–S) bond and carbon-nitrogen (C–N) bond are the most active parts of the remaining structure, which means they are more likely to be attacked by reactive radicals or ozone due to their lower bond dissociation energies compared to other molecular bonds present [
30]. The degradation intermediates generated can undergo further oxidation reactions, leading either to complete mineralization or to the formation of less complex organic species, as shown in
Figure 5.
3.4. Practical Analysis
Evaluating the efficiency of AOP in wastewater treatment is very important and often characterized by kinetic rate constants. However, these constants alone do not encompass other aspects of AOP efficiency, particularly the operational costs. To fill this gap, Bolton et al [
31] introduced the Electrical Energy per Order (EE/O), defined by Equation (16). EE/O measures the electrical energy required per reactor volume to decrease a target contaminant’s concentration by one order of magnitude, a critical metric in evaluating the energy consumption, which constitutes a significant portion of AOP operating costs. This metric is especially relevant for scenarios with low initial pollutant concentration (C
0) and is a critical measure of operational costs and instrumental in effectively scaling up treatment designs and estimating costs. The calculated EE/O value provides a benchmark for assessing the energy efficiency of the plasma treatment relative to other AOPs. A lower EE/O signifies greater energy efficiency and cost-effectiveness in contaminant reduction.
where P
elec is system power (kW), V is volume of water treated (L) in time t (min), C₀ is the initial contaminant concentration (EEO is valid for low initial concentrations, typically < 100 mg/L) and C is the final concentration [
32].
For non-thermal plasma treatment process in this study, the total electrical energy utilized was 2.78 watts for treating a volume of 10 mL over 20 minutes, with a 50% duty cycle. The process successfully reduced the initial concentration of the contaminant from 20 ppm to 0.7 ppm. According to Equation (16), the EE/O = (0.00278 ∗ (20*50%) ∗ 1000)/(0.01 ∗ 60 ∗ log (20/0.7)) = 31.82 kWh/m3/order for an initial concentration of 20 mg/L.
The plasma treatment system, requiring no oxidants, has a cost advantage over other AOPs and is competitive in the market. A brief economic comparison of different AOPs, such as ultrasonication (U/S), O
3, and UV treatments, is provided in
Table 4. Notably, this moderate EE/O number is even lower than other existing efficient systems in low initial concentration.
The comprehensive data from various AOPs, as summarized in
Table 4, elucidate the relative economic and efficiency profiles of each method. The findings indicate that while standalone methods like U/S are cost-intensive, their integration with other AOPs can yield more economically viable solutions. For instance, U/S treatment alone exhibits a higher EE/O of 10964 kWh/m
3/order, which significantly diminishes to 989 kWh/m
3/order when combined with UV or O
3 treatments. This suggests that non-thermal plasma can also combine with other technologies to further enhance wastewater treatment efficiency.
The elevated total water treatment cost observed in this study is primarily attributed to the limited capacity of the employed reactor. Notably, the reactor operated without the introduction of any gases and featured a straightforward experimental setup, indicating substantial potential for future enhancements and optimization. Furthermore, the overall treatment efficiency could be further increased if the process was applied on a larger scale. Additionally, a study by Fahmy et al. [
37] provided valuable insights into optimizing the removal of Acid Orange 142 dye. They achieved an 88.87% removal efficiency in treating 100 mL of wastewater with a 20 mg/L dye concentration, using a 12.5 kV voltage and a 5 mm gap between the solution surface and the high-voltage electrode over 90 minutes. Their energy consumption under these conditions was 2.025 kWh, equating to 30.375 W/mL. In contrast, our study recorded a substantially lower energy consumption rate of 0.33 W/mL, while achieving a higher decolorization rate of 96.5%, indicating a more efficient process for dye decolorization. These results clearly showed the effectiveness of this reactor.
In comparison to other AOPs utilized for water treatment, the efficiency and cost of the current plasma-based technology are quite competitive. These findings suggest that plasma technology not only holds broad application prospects but also, when integrated with other AOPs, can lead to further cost reductions. The potential for cost-effective scalability and the ability to integrate seamlessly with other treatment methods make plasma technology a promising option in advancing sustainable water treatment strategies.