Figure 3 shows the current and voltage waveform measurements. When the capacitor was discharged, the current decreased, and after a certain time, a sudden current drop and sharp rise in voltage occurred. The current drop and voltage rise resulted from an increase in the resistivity of the wire due to vaporization. The resistance increased several orders of magnitude, and the current in the circuit was interrupted. The inductive energy stored in the stray inductance of the circuit generated a high voltage, which was higher than the initial voltage of the capacitor. Once vaporization occurred, an arc discharge formed between the electrodes, and the current flowed through the low-resistance arc plasma. Thus, the wire was heated and vaporized by the energy deposited in the wire until the arc discharged.

Figure 4 shows the voltage and current waveforms under various capacitances. Under the same discharge voltage, the pulsed voltage was faster and higher with 30 μF than with 7.5 μF. The increase in capacitance resulted in a slight increase in current and energy deposition.

The energy deposited in the wire affected the size and shape of the particles (Table 3). As the energy deposited in the wire increased, the particle size decreased. Figure 5 shows the X-ray diffraction [XRD] patterns of silver nanoparticles generated through the EEWL. Only a silver peak was observed. The average size of the synthesized Ag nanoparticles calculated from the XRD patterns by Scherrer's equation is around 23.9 nm. Figure 6 shows the size distribution from TEM images and the morphology of the Ag nanoparticles at different capacitances. The charging time and duration of plasma can influence the formation of particles. The Ag particles were spherical in shape and the particle distribution approached a bimodal distribution, which may have resulted from liquid droplet formation-induced aggregation of particles 17 .

Figure 6b shows the coagulation of Ag nanoparticles, which was likely a result of three potential processes. First, agglomeration may have occurred during handling or processing due to the high specific surface area of the nanoparticles. Second, it may have occurred during the drying process of sample preparation. Finally, it may have occurred during condensation.

Figure 7 shows the effect of liquid type on the current and energy. Deionized water resulted in smaller particle sizes and good stability (Table 4). The plasma expanded in the medium due to the temperature difference between the plasma and the medium. A low viscosity led to a larger expansion in the plasma volume; thus, the particle size was smaller.

Defects were observed in the particle structure. Under non-equilibrium process conditions, the synthesized particles were imperfect and had defects, such as twins and dislocations. The UV/Vis absorption spectra of Ag nanofluids are shown in Figure 8. Maximum absorbance occurred at ~398 nm, similar to previous reports 18 .

Typical boiling curves for pure water and Ag nanofluids are shown in Figure 9. The boiling curves of Ag nanofluids showed no significant change due to the very low concentration. The increase in the CHF can be explained by the changes in surface properties 19 . Analysis of the wire surface showed that a silver layer built up during nanofluids boiling, as shown in Figure 10.

DOE was used to compare the discharge features and to optimize process parameters. EEWL is a function of many parameters, including wire diameter, length and material, electrical circuit properties, and ambient conditions. In this experiment, there were four parameters of interest that affected the size and dispersibility of particles: capacitance, wire length, liquid volume, and liquid type. Each parameter had two levels or values. The experimental parameters are summarized in Table 5. The experiments were set up using Minitab. The DOE used a 1/4 fractional factorial design of the full factorial design with two repetitions. This resulted in eight total experiments, which was less than the full factorial design of 16, as shown in Table 5. Designs with factors that are usually set at two levels assume that the effects of the factors are linear. Generally, a factor is only evaluated at the low and high points. Ideally, there should be one or more center points to improve the reliability. Further analysis can compare the measurements of the dependent variable at the center point with the average values of the design.

Figures 11 and 12 show the main effects plot, interaction plot, and cube plot for size and zeta potential to quantitatively assess the effects of the parameters of EEWL. In the cube plot, the mean responses of the factors are displayed on the corners of the cube. Low levels of the factors are to the left, front, or bottom and high levels to the right, back, or front of the cube. Figure 13 represents the residual plot for size and zeta potential. The normal probability plot did not show any unusual features.

The process was optimized when capacitance was 30 μF, wire length was 38 mm, liquid volume was 500 mL, and liquid type was deionized water. These conditions corresponded to a particle size of 49 nm and zeta potential of -39.15 mV, as shown in Figure 14. These predicted optimized parameters were confirmed by repeating the experiment, and the results for particle size and zeta potential were 118.9 nm and -42.5 mV, respectively. There were differences between the prediction and the experimental results. However, the experimental results showed the smallest and most stable Ag nanoparticles.