Evaluating a dielectric barrier discharge plasma process for the removal of cobalt from water and simultaneous synthesis of cobalt oxide catalyst

2.1. Materials and solution preparation

Cobalt nitrate, obtained from Thermo Fisher Scientific, USA, was utilized to prepare synthetic wastewater. It was dissolved in distilled water which was obtained from Walmart, USA. The distilled water’s pH and conductivity were measured as 5.3 ± 2 and 199 ± 5 µS/cm, respectively. After dissolving cobalt nitrate in the water, the mixture was stirred for 30 mins to ensure homogeneity. The resulting cobalt(II) concentration in the solution was 100 mg/L. Before conducting the experiments, the solution’s pH and conductivity were measured again, yielding values of 5.8 ± 2 and 323 ± 8 µS/cm, respectively.

2.2. Experimental setup and operation

Figure 1 illustrates the experimental setup used for plasma treatment in this study. The plasma power source was operated at a fixed frequency of 20 kHz. The reactor consisted of a quartz tube with inner and outer diameters of 3.5 mm and 5 mm, respectively. A stainless-steel needle (1.5 mm diameter) served as the high-voltage electrode, while a copper sheet (0.3 mm thick) was used as the ground electrode. The high-voltage electrode had a 3 mm tip length with an approximately 25 µm apex radius. A high-voltage electrode with the apex radius intensified the localized electric field, reducing the breakdown voltage for more efficient and stable plasma generation. Its sharp tip enhanced ionization, creating a well-confined, directed plasma jet ideal for applications such as wastewater treatment and material processing. The solution was held in a glass beaker, which also functioned as a dielectric barrier between the electrodes. The distance between the reactor nozzle and the solution surface was maintained at 1.0 cm throughout the experiments. To ensure uniform mixing during plasma treatment, a magnetic stirrer was employed at a constant speed of 80, while minimizing disturbances to the plasma jet. The selection of this low rotational speed was intentional, as higher speeds could induce a vortex formation in the solution, which could lead to fluctuations in the solution level, potentially altering the interaction between the plasma jet and the liquid surface. By maintaining a stable liquid surface, the plasma-liquid interface remained consistent, ensuring uniform plasma treatment conditions throughout the experiment.

Close

Figure 1.

Experimental setup of the DBD plasma reactor for cobalt removal and recovery. HV: High voltage; MFC: Mass flow controller; OD: Outer diameter; ID: Inner diameter; OSC: Oscilloscope.

Export to PPT

A stable and uniform plasma discharge was achieved using high-purity argon gas (99.99%), with flow rates ranging from 1.0 to 2.0 L/min, regulated by a mass flow controller. The plasma reactor was powered by a PVM500-Plasma Power Generator (Information Unlimited, USA), with the applied voltage ranging from 6.1 to 8.2 kV, corresponding to input powers of 30 to 50 W. A Watt meter (PN1500, Poniie Corporation) was used to monitor power delivery, and all electrical parameters were recorded with an oscilloscope (DPO 3034, Tektronix, USA) equipped with a high-voltage probe (P6015A, Tektronix, USA). The current was measured using Pearsion current monitor (Model 2100, Pearson Electronics, USA).

2.3. Analytical methods

Plasma chemistry in the discharge zone was analyzed using optical emission spectroscopy (OES, Ocean Optics HDX-UV–VIS), capturing photon emissions from plasma species with an optical fiber positioned 3 cm from the discharge zone. The formation of cobalt particles was further confirmed using a UV-vis spectrometer (Synergy HT, BioTek Instruments, Inc., USA).

The surface morphology and particle size distribution (PSD) were examined using scanning electron microscopy (SEM) and analyzed with ImageJ software. The elemental composition of the cobalt oxide powder was determined through energy dispersive X-ray spectroscopy (EDS) (X-max 80T, Oxford, UK). The solution’s pH and conductivity were measured using a PC850 Portable pH/Conductivity Meter Kit (PC850, APERA Instrument, USA). The cobalt removal efficiency ${\eta }_{CRE}$ and energy efficiency (EE) were calculated using the following Eqs. (1) and (2):

In these equations, $C_0$ represents the initial cobalt concentration (g/L), V denotes the total volume of the solution treated (L), ${\eta }_{CRE}$ is the cobalt removal efficiency (%), P is the power applied during the process (kW), and t refers to the treatment duration (h).

2.4. Experimental design

The statistical design and data analysis for this experiment were performed using Design Expert Software (version 13.0). The study examined the effects of three factors such as applied power (30, 40, and 50 W), treatment time (20, 25, and 30 mins), and gas flow rate (1.0, 1.5, and 2.0 L/min) on cobalt removal from water. A central composite design (CCD) was employed to evaluate the impact of these factors and their interactions on the response variables. Response surface methodology (RSM) was applied to develop polynomial models for predicting outcomes and to optimize the process. The selection of three independent variables was allowed for accurate estimation of a second-order quadratic model, facilitating identification of the optimal combination of parameters within the three-dimensional design space [47-49]. The experimental setup, including factor levels, has been summarized in Table 1.

A total of 20 experimental runs were conducted, including five replicates of the center point to accurately estimate experimental errors. A second-order quadratic polynomial equation was formulated to model the response as a function of the independent variables, as shown below.

In this study, Y represents the response variable, specifically the removal efficiency, while X₁, X₂, and X₃ correspond to the applied voltage, treatment time, and gas flow rate, respectively. The coefficients β₀, β₁, β₂, β₃, β_11, β_22, β_33, β₁₂, β₁₃, and β₂₃​ were determined from the experimental data and represent the effects and interactions of the variables within the model. The response surface function was used to analyze the relationships among the variables and to calculate the coefficients for Eq. (3).

The analysis of variance (ANOVA) was conducted using Design Expert Software to statistically evaluate the equality of means and examine the relationships between the control factors and the removal efficiency achieved using NTP. The model’s quality and accuracy were assessed using several statistical parameters, including the coefficient of determination (R²), adjusted R² (Adj. R²), predicted R² (Pre. R²), and adequate precision (Adeq Pre). Adeq Pre, which indicates the signal-to-noise ratio, is considered acceptable when it exceeds 4 [48,49].

The model’s reproducibility was evaluated using the coefficient of variation (CV), which represents the ratio of the standard error to the mean value of the response. A CV value of less than 10% is typically expected to ensure the model’s reliability and consistency [48].

3.1. Characterization of plasma discharge

Figure S1 presents the current-voltage characteristics of the plasma discharge at applied voltages of 6.1, 7.1, and 8.2 kV with a constant frequency of 20 kHz. Figure S1(a) illustrates the voltage and current waveforms, while Figure S1(b) displays the corresponding Lissajous figures. From Figure S1(a), each cycle of the applied voltage exhibits two distinct discharge current pulses, representing positive and negative breakdowns. During the positive voltage half-cycle, a negative current pulse is observed due to the phase shift, indicating negative breakdown. Conversely, the negative voltage half-cycle produces a positive current pulse, signifying positive breakdown. Notably, the current pulse from the negative half-cycle displays higher intensity, suggesting that positive breakdown is stronger than negative breakdown in this plasma discharge. The phase gap between voltage and current observed in Figure S1(a) signifies a substantial contribution of displacement current, a key characteristic of DBD and capacitively coupled plasma systems. This displacement current arises due to variations in the electric field within a dielectric or capacitive environment, as described by Maxwell’s equation $\left(I_d=\in \frac{dE}{dt}\right)$ [50,51]. Before plasma breakdown, the total current is predominantly displacement current, resulting in a noticeable phase shift between voltage and current waveforms [52]. However, once the plasma ignites, sharp pulses appear in the current waveform, corresponding to plasma-induced discharge current generated by electron avalanches and streamer discharges. These pulses indicate moments of ionization and charge transport, which drive plasma formation and energy dissipation [53].

Lissajous figures generated by the oscilloscope are a crucial diagnostic tool for analyzing the discharge characteristics of a DBD reactor, as they represent the relationship between applied voltage and transferred charge and provide a non-invasive approach to evaluating plasma generation efficiency, discharge modes, and reactor optimization. Before breakdown, the plot follows a linear path, indicating capacitive behavior dominated by the dielectric barrier. Once the breakdown voltage is reached, microdischarges occur, introducing charge transfer and forming a parallelogram shape, where the enclosed area represents the energy dissipated per cycle (Figure S1b). The discharge process progresses through pre-breakdown, plasma initiation, and extinction phases, with the Lissajous figure capturing key information, such as dielectric capacitance (1 μF) and power consumption being 19, 27, and 38 W, respectively, while the applied power was set at 30, 40, and 50 W. The discharge efficiency was then calculated to be 63%, 67.5%, and 76%, and the measured applied voltage was 6.1, 7.1, and 8.2 kV, respectively.

3.2. Modeling and statistical analysis for cobalt removal

The dataset used for the CCD combined with RSM, including both actual and predicted removal efficiencies, is presented in Table S1. The close alignment between the experimental results and the predicted values in Table S1 demonstrates the reliability of the developed model (Eq. 4). This confirms that the CCD-based model is effective in accurately predicting cobalt removal efficiency using NTP.

In this context, Y denotes the cobalt removal efficiency achieved through NTP, while X₁, X₂, and X₃​ correspond to the applied power, treatment duration, and gas flow rate, respectively.

The ANOVA results for Eq. (4) have been presented in Table 2, where the significance of each factor was assessed using F and P values. A high F value and a low P value are indicative of a significant effect [48,54-57]. As shown in Table 2, the model achieved an F value of 261.42, strongly confirming its significance, with only a 0.01% likelihood that this value could arise from random noise. Similarly, the model’s P value was 0.0001, well below the threshold of 0.05, further emphasizing its significance. Among the model terms, only X₁X₃, X₂X₃, and $X_1^2$ were deemed insignificant (p > 0.05).

The lack-of-fit F value was calculated as 2.00, indicating that the lack of fit was not significant compared to the pure error, with a 23.32% chance that such a value could result from noise. The R² was 0.9933, demonstrating a strong fit between the model and experimental data. The predicted R² (Pre. R²) (0.9622) closely aligned with the Adj. R² (0.9874), with a difference of less than 0.2, indicating consistency. The Adeq. Pre. value, which measures the signal-to-noise ratio, was 45.96, well above the desired minimum of 4, confirming the model’s reliability for navigating the design space. Additionally, the CV was 3.52%, significantly below the critical threshold of 10%, confirming the model’s high reproducibility.

Figure S2 shows the results of predicted versus actual values of the cobalt removal. The actual values were taken from the experiments (20 runs), whereas the predicted data were generated using Eq. (4). The correlation of the actual and predicted data for removal efficiency showed a highly linear relationship, indicating the excellent prediction of experimental data using the model.

3.3. Effect of power, treatment time and gas flow rate on cobalt removal efficiency

Figure 2(a) illustrates the combined effect of applied power and treatment time on cobalt removal efficiency. The applied power was varied from 30 to 50 W, while the treatment time ranged from 20 to 30 mins, with the gas flow rate maintained at 1.5 L/min. The graph clearly demonstrates that removal efficiency increased proportionally with both applied power and treatment duration. Higher power levels resulted in the generation of more plasma species, such as energetic electrons, hydroxyl radicals, UV radiation, atomic oxygen, and electric fields, which accelerated nucleation and enhanced removal efficiency. Additionally, extending the treatment duration allowed more time for nucleation, further improving efficiency. The maximum removal efficiency of 99% was observed at 50 W and 25 mins of treatment time, but increasing the duration to 30 mins did not yield any significant additional improvement.

Close

Figure 2.

Interacting influence of (a) power and treatment time, (b) power and gas flow rate, (c) treatment time and gas flow rate on the cobalt removal efficiency, and (d) regression curve of the sub-model for each variable.

Export to PPT

Figure 2(b) presents the combined effect of applied power and gas flow rate on cobalt removal efficiency, with the treatment time fixed at 25 mins. The gas flow rate was varied from 1 to 2 L/min, and the applied power was adjusted between 30 and 50 W. The removal efficiency generally increased linearly with higher applied power but showed a distinct trend concerning gas flow rate. Efficiency improved as the flow rate rose from 1.0 to 1.5 L/min, aligning with previous studies that indicate higher flow rates enhance discharge power by introducing more gas molecules capable of forming charged species [58]. However, when the gas flow rate exceeded 1.5 L/min and reached 2.0 L/min, the removal efficiency declined.

This phenomenon is further explained in Figure S3, which shows the calculated discharge power under varying flow rates. The data revealed that as the flow rate increased from 1.0 to 1.5 L/min, discharge power also increased at a constant applied power. However, further increasing the flow rate to 2.0 L/min caused a drop in discharge power. This decline likely occurred because the excess gas flow surpassed the ionization capacity of the fixed applied power, leading to a reduction in discharge power and, consequently, a decrease in cobalt removal efficiency.

Figure 2(c) illustrates the combined effect of gas flow rate and treatment time on cobalt removal efficiency, with the applied power held constant at 50 W. The results show that removal efficiency increased as the gas flow rate rose from 1.0 to 1.5 L/min and the treatment time extended from 20 to 30 mins. However, increasing the gas flow rate beyond 1.5 L/min to 2.0 L/min led to a slight decline in removal efficiency, likely due to reduced discharge power, as shown in Figure S3.

Figure 2(d) presents the regression curve for the sub-model, highlighting the influence of the independent variables on cobalt removal efficiency in the NTP process. Among the variables, applied power had the most significant impact on removal efficiency, followed by treatment time. In contrast, gas flow rate exhibited a relatively minor effect on the efficiency of cobalt removal.

3.4. Energy efficiency for cobalt removal

Figure 3 presents the EE for different combinations of applied power, gas flow rate, and treatment time. When the gas flow rate was set at 1.5 L/min, as outlined in Eq. (2), EE was found to be directly related to cobalt concentration, removal efficiency, and solution volume, while being inversely related to applied power and treatment time. With cobalt concentration fixed at 100 ppm (100 mg/L) and solution volume at 40 mL, EE was determined by the removal efficiency, applied power, and treatment time.

Close

Figure 3.

Interacting influence of (a) power and treatment time, (b) power and gas flow rate on EE.

Export to PPT

Figure 3(a) illustrates the combined effect of applied power and treatment time on EE. As treatment time increased from 20 to 30 mins, EE decreased. However, increasing the applied power from 30 to 50 W led to a rise in EE, even though EE is typically inversely related to power. The improvement in EE at higher applied power can be attributed to the significant increase in removal efficiency, which outweighed the moderate rise in applied power.

Figure 3(b) illustrates the combined effect of gas flow rate and applied power on EE, with treatment time held constant at 25 mins. The impact of applied power on EE was previously discussed in Figure 3(a), so here the focus is on the influence of gas flow rate. As the gas flow rate increased from 1.0 to 1.5 L/min, EE improved, which can be attributed to the corresponding increase in cobalt removal efficiency (see Figure 2b). However, when the gas flow rate was raised further to 2.0 L/min, a decline in cobalt removal efficiency was observed, resulting in a decrease in EE.

3.5. Kinetic modelling for cobalt removal by DBD

To study the kinetic modelling, treatments were run at 5 mins intervals up to 30 mins, and the treated samples were analyzed using inductive plasma coupled (ICP)-OES for the removal efficiency. The rate of removal efficiency of cobalt was evaluated using a pseudo-first-order kinetic model as follows (Eq. 5):

where C₀, C_t, k, and t represented the initial concentration of cobalt (100 mg/L), the concentration of cobalt at time t (mg/L), the pseudo-first-order reaction rate constant (min^-1), and time (min), respectively.

Kinetic models were developed for each applied power condition, as shown in Figure 4. A linear relationship between ln(C₀/C_t) and plasma treatment time for cobalt removal was achieved. The slopes of the linear regressions of ln(C₀/C_t) versus plasma treatment time were the kinetic rate constants ‘k’. The R² of 30 W, 40 W, and 50 W were 99.0, 97.9, and 95.7, respectively, which indicated a highly fitted pseudo-first-order kinetic model for this study. The magnitude of the reaction rate constant was found to follow an order from high to low according to applied power, i.e., 50 W> 40 W> 30 W. This was expected because, as discussed earlier, the removal efficiency increased with increasing applied power when the treatment time was constant. The reaction rate constants for 30 W, 40 W, and 50 W were found to be 0.0259 min^-1, 0.0636 min^-1, and 0.1332 min^-1, respectively.

Close

Figure 4.

Pseudo-first-order kinetic plot of ln(C₀/C_t) vs. time.

Export to PPT

3.6. Characterization and efficiency of cobalt recovery

Figure 5 shows the cobalt oxide nanoparticle formation at various treatment times. The applied power and gas flow were constant at 50 W and 1.5 L/min, respectively, in this study. The test solutions’ UV-vis spectra demonstrated an increase in the intensity of absorption bands at 370 nm. These spectra matched very well with the results from previously published studies [59]. The NTP-treated solution contained cobalt oxide particles, and the colloidal particle density increased with increasing treatment time. No particle formation was observed at 0 mins treatment time. But after 1 min of treatment, a small cobalt oxide peak at 370 nm was seen, according to Figure 2. When treatment time increased to 7 mins, the maximum peak intensity was observed, but further increasing treatment time to 9 mins produced no peaks because, under this condition, all the cobalt oxide particles were precipitated, thus no absorption band at 370 nm was found.

Close

Figure 5.

UV-Vis spectra of NTP treated cobalt solution at various treatment time.

Export to PPT

The surface morphology study of the recovered cobalt oxide powder for both pristine and calcinated samples was done using SEM analysis, which showed almost similar irregular particle shapes. Figure 6(a) and (b) show the SEM images of recovered cobalt oxide powder for the pristine sample and for the calcinated sample. There was no morphological change observed between pristine and calcinated samples. Since the particle growth was uncontrolled, irregular particle shapes were developed under plasma treatment conditions.

Close

Figure 6.

SEM image of (a) pristine and (b) calcinated recovered cobalt oxide powder; (c) XRD patterns of pristine and calcinated recovered cobalt oxide powder; (d) PSD of recovered cobalt oxide powder.

Export to PPT

The composition and crystalline structure of the recovered cobalt-oxide particles were analyzed before and after calcination using X-ray diffraction (XRD). The calcination process was conducted at 500°C for 5 h in an oven. Figure 6(c) presents the XRD patterns of both the pristine and calcined samples, which closely match the patterns of Co₃O₄ and CoO. Characteristic diffraction peaks for both the pristine and calcined samples were observed at 19.00°, 31.35°, 36.95°, 38.50°, 44.95°, 59.25°, and 65.25°, corresponding to the (111), (220), (311), (222), (400), (511), and (440) planes, respectively. These peaks were consistent with Co₃O₄ (JCPDS card no. 75-2480) [60,61]. Three additional peaks at 55.65° (422), 74.20° (311), and 77.65° (222) were present only in the calcined samples. The peak at 55.65° was attributed to Co₃O₄, while the peaks at 74.20° and 77.65° indicated the formation of CoO [60] due to the calcination process. Another notable observation was the increase in crystallinity of the sample after treatment at 500°C, as evidenced by sharper and more intense peaks in the XRD pattern. From a catalytic perspective, enhanced crystallinity is beneficial as it can increase catalytic activity and stability, as well as reduce the activation energy for reactions. Additionally, improved crystallinity offers better recyclability compared to amorphous catalysts [47].

Figure 6(d) showed the PSD of the cobalt oxide powder. As there was no morphological difference between pristine and calcinated samples, this PSD represented both samples. The size of the particles was calculated using ImageJ software, and it could be seen that the particle size varied from 0.7 to 8 µm with an average particle size of 2.5 µm, which was within the range of a typical commercial cobalt oxide catalyst.

EDS analysis was done to confirm the chemical composition of the recovered cobalt oxide powder. Figure S4 showed the EDS analysis spectrum, where Figure S4(a) was the EDS spectrum of pristine powder and Figure S4(b) the EDS spectrum of calcinated powder. In both cases, the weight percentage and atomic percentage of the elements were also presented. It was noticeable that the presence of cobalt in both cases was found to be more than 70%. The amount of oxygen in both samples was also found to be similar, at around 26%. Apart from these observations, the EDS also detected a negligible amount of sodium, silicon, and chlorine. The presence of sodium and silicon in the cobalt oxide powder was believed to come from the mortar and pestle, which were used to grind the recovered cobalt oxide powder before XRD and SEM analyses. While the source of chlorine in the cobalt oxide powder was unknown, the presence of carbon in the spectrum was understandable because the samples were prepared on carbon tape.

For this study, a series of experiments were conducted under optimal conditions to evaluate cobalt recovery after plasma treatment. To analyze the Co recovery rate, we assumed that all the recovered material is Co₃O₄, despite the presence of two small peaks corresponding to CoO in the XRD data. This assumption simplifies the analysis and is based on the predominance of the Co₃O₄ phase in the recovered material, as indicated by the XRD results. The cobalt recovery rate was determined based on an initial cobalt concentration of 100 ppm (100 mg/L) in a 1-liter solution. After plasma treatment, 72 mg of Co₃O₄ was recovered. Given that the molecular weight of Co₃O₄ is 240.8 g/mol, with cobalt contributing 73.4% of its mass, the actual cobalt content in the recovered solid was calculated to be 52.9 mg. Comparing this to the initial 100 mg of cobalt in the solution, the recovery rate was determined using the formula (Recovered Co / Initial Co) × 100, resulting in a cobalt recovery rate of 52.9%. Although the recovery rate is substantial, some discrepancies in cobalt mass balance during the plasma treatment process, filtration, and calcination may be due to incomplete precipitation, retention in the residual solution, or volatilization at high temperatures. Further optimization of these processing steps could improve the overall cobalt recovery efficiency.

3.7. Reaction mechanism of cobalt oxide precipitation with excited plasma species

Figure 7 displays the OES spectrum of the DBD plasma. The spectra were recorded to identify reactive species generated during plasma discharge. Only species with relatively long lifetimes and high intensities were detected, as transient species with shorter lifetimes and lower intensities could not be captured. For comparison, the plasma spectrum in air (without the liquid solution) was also recorded, with all measurements conducted at an applied power of 50 W.

Close

Figure 7.

Optical emission spectrums of the reactors taken at 50 W applied power during the experiments.

Export to PPT

Due to the presence of nitrogen in air and the atmospheric-pressure plasma environment, the spectrum was dominated by Ar and N species. In the case of plasma in air, NO species (A²Σ⁺- X²Π) were detected between 220 and 275 nm. However, in the presence of a liquid precursor, NO species and O radicals were absent, likely because the interaction between the plasma and liquid limited the formation time for these species. The spectrum included low-intensity N lines from the second positive system N₂ (C³Π_u−B³Π_g) between 302 and 435 nm [62] and an oxygen emission peak (3P⁵-3S⁵) at 777.03 nm [63-66], likely resulting from H₂​O fragmentation or O₂​ dissociation [67].

Prominent Ar emission lines (4P-4S) were observed between 690 and 980 nm [66,68], alongside low-intensity lines for OH, H_ꞵ, and H_α ​at 309.1, 486.1, and 656.3 nm, respectively. The OH radical at 313 nm overlapped with nitrogen peaks [69,70]. For plasma in air, H_α​ and H_ꞵ emissions were not detected, possibly due to insufficient water vapor near the plasma source or inadequate energy delivery to produce these lines. Vaporized water near the plasma interacting with the discharge likely contributed to H_α ​and H_ꞵ​ emissions in liquid-associated plasma, which were absent when water vapor was insufficient or energy input was inadequate.

The reactive species formed due to the energy gained from the collisions between accelerating electrons and neutrals are called primary reactive species, including electrons (e⁻), ionized neutrals and gas (M⁺), excited neutrals and gas (M^*), N, O, and atomic H, NO, and O₂^.–. These primary reactive species have a short lifetime. The secondary reactive species are formed when the primary species are transformed into other reactive species, such as H₂O₂, NO₂, NO₃, and O₃ in the ambient environment [71]. The possible reaction mechanism for the cobalt removal process was derived using OES based on the generated reactive species present in the plasma discharge zone. Figure 7 showed such reactive species at the operation condition of 50 W power with argon (99.99%) as a discharge gas. Ar is an ideal discharge gas for plasma jets at atmospheric pressure due to its low ionization energy (15.76 eV), which facilitates stable plasma formation with relatively low power input. Additionally, its strong optical emission in the visible and near-UV range allows for easy plasma diagnostics using OES. Compared to other noble gases like helium, argon is significantly more cost-effective and widely available, making it a practical choice for sustained plasma jet operation in industrial and research applications. As an inert noble gas, it also prevents unwanted chemical reactions and creates an isolated environment for excitation by separating elements such as oxygen and nitrogen from air. Among these generated species, hydroxyl radicals (^•OH) and hydrogen peroxide (H₂O₂) were identified as the major reactive species responsible for the formation of Co₃O₄ particles to remove the cobalt (II) ions in the solution by NTP plasma. The possible reaction pathways of the generation of OH radicals and H₂O₂ in the plasma region are shown below in Eqs. (6-9). [36,59,72,73].

Since hydroxyl radical is a strong oxidant, it plays a major role in the decomposition/degradation of pollutants owing to its high oxidation potential and high second-order reaction rate constants (10⁸–10⁹ Lmol^-1s⁻¹) in wastewater treatment [74,75]. In this study, dissolved Co²⁺ in the solutions was oxidized into Co³⁺ ions by the generated ^•OH in the plasma phase, followed by spontaneous co-precipitation of mixed Co^II/Co^III hydroxide nucleate particles. Then, Co²⁺ ions in the solution were adsorbed onto the nucleate particles to form CoOOH_(ad) by surface oxidation with the help of H₂O₂, as shown in Eqs. (10-12) [59].

In this stage, nucleates provided sites for continuous adsorption of CoII(ad), and the oxidation reaction rate depended on the production of H₂O₂ in the plasma phase. The CoOOH(ad) formed was incorporated into a solid CoOOH phase, leading to the growth of oxide particles, which could be slowly transformed to Co₃O₄. This process to form larger particles of Co₃O_{4 (S)} continued until Co²⁺ ions were depleted in the solution.

3.8. Comparison of the outcome with other reported work

Table 3 presents a comparative analysis of various plasma-based methods for cobalt and other metal removal, highlighting key differences in applied energy, removal efficiency, treatment duration, and EE. This study was conducted under fixed conditions, including a gas flow rate of 1.5 L/min, a treatment time of 25 mins, and an applied power of 50 W. Both continuous liquid-phase plasma discharge (CLPD) [34] and MW plasma [45] operated at 300 W, with treatment times of 25 and 30 mins, respectively. While MW plasma achieved a slightly higher removal efficiency (96.9%) than CLPD (93%), CLPD exhibited a significantly higher EE of 27,440 mg/kWh. The MW plasma’s EE remained unreported. Micro plasma [26], operating at 6 kV for 25 mins, could treat a notably higher initial concentration (589 mg/L); however, its removal efficiency and EE are unavailable. The DBD plasma [73] treatment for Zn, using an AC voltage of 6 kV and a DC voltage of 8 V over 10 mins, resulted in a relatively low removal efficiency of 29%. In contrast, the DBD plasma jet examined in this work demonstrated outstanding performance, utilizing only 50 W to achieve a remarkable 99% removal efficiency within 25 mins for a 40 mL solution containing 100 mg/L of Co. Despite its lower energy consumption, its EE (196.08 mg/kWh) was significantly lower than that of CLPD but remained superior to other reported methods. These findings suggest that the DBD plasma jet is a highly effective and energy-efficient technique for Co removal, offering an optimal balance between treatment efficiency and energy consumption.