Low-dielectric layer increases nanosecond electric discharges in distilled water

Many applications of plasmas generated from electric discharges (EDs) in liquid have emerged,^1–3 including water purification,^4,5 nanomaterial synthesis,⁶ and biomedical application.^7,8 The non-thermal character of nanosecond discharges in water, created by pulsed high voltages with pulse-widths of several nanoseconds,^9,10 also makes them attractive for liquid processing.^11,12 To facilitate the ED in liquid, gaseous bubbles have been added,^12,13 with induced discontinuities in the dielectric permittivity $( ε )$ and density $( ρ )$ at the bubble-liquid interface. With gaseous bubbles, a discontinuity in $ε$ changes the spatial distribution of the electric field and causes locally augmented field intensity at the bubble-liquid interface.^14,15 Although the breakdown voltage can be lowered with gaseous bubbles,^12,16 the interaction between the plasma and liquid is limited because the discharge occurs in the gaseous bubbles. For this reason, for high processing efficiency, direct discharges into liquid are preferred.^17,18 Thus, researchers have sought to increase processing efficiency by minimizing energy consumption. Here, we report a simple method for improving ED in water. We add a layer of low-$ε$ dielectric material (n-heptane, $ε = 2$⁠) to the water $( ε = 80 )$⁠. The interface between the water and n-heptane caused by the immiscible nature of these liquids provides the intended discontinuity in $ε$⁠. We demonstrate that the discharge probability, the discharge volume, the injected charge, and the intensity of the electric field increase as the position of the interface approaches the tip of the anode. This is attributed to locally enhanced field intensity near the anode tip supported by a static field simulation.

Nanosecond discharges were created in distilled water using the experimental setup schematically presented in Figure 1. A pulsed power supply (FID, FPG-25-15NM) was used to create a positive high-voltage (HV) pulse (amplitude V_a = 10 and 15 kV, full-width at half maximum of 10 ns, rise time of 3 ns, and repetition rate of 1 Hz) to a tungsten pin (anode, diameter of 0.5 mm with a radius curvature of $∼ 50 μ m$ at the tip) using a coaxial cable, while a lower tungsten rod (cathode, diameter of 0.5 mm) was grounded. Further details on the voltage-current waveforms with discharge are available in a supplementary material. Two gap distances (d) between the electrodes were studied: d = 1.5 and 2.5 mm. Both electrodes were immersed in a layered liquid consisting of distilled water (⁠$ε ≈ 80$⁠, $ρ = 1$ kg/m³) on the bottom and n-heptane (⁠$ε ≈ 2$⁠, $ρ = 0.7$ kg/m³) on the top; the thickness of heptane layer was 4 cm. The voltage and current were measured with a HV probe (Tektronix, P6015A) and a current monitor (Pearson, 6585), respectively, and recorded by an oscilloscope (Tektronix, DPO-4104B). An ICCD camera (Princeton Instruments, PI-MAX2) was used to capture the spatial distribution of the light emitted from the discharge. A delay generator (BNC, 575) synchronized the gating of the ICCD and a LED lamp to the discharge event. The LED lamp was used to discriminate the interface between two liquids and the electrodes in ICCD images. As the synchronization at nanosecond scale requires delicate attention,¹⁹ we generated a reference square signal to trigger the high voltage. To determine the delay time of a second triggering signal to the ICCD camera, we increased the delay gradually with a fixed exposure time of the ICCD exposure (2 ns) until light emission was detected. In the present experimental configuration including cables and connections, the delay was 190 ns. Then, the exposure time of the ICCD was set at $1 μ s$ to fully accumulate emitted light from discharges. Meanwhile, since the lamp requires 5 ms to reach its saturated light intensity under 10 V input signal, a triggering square signal for the lamp was with 10 V (amplitude), 7 ms (width), and –6 ms (delay from the reference signal) ensured homogeneous background illumination during 1 ms. Noting that during this experiment, the whole electrical connections were kept unchanged and the obtained results were consistently reproducible.

The position (h) of the interface was systematically varied by controlling the volume of water using a microsyringe; we defined h as a positive when the interface was higher than the tip of the anode (Figure 1). We could generate EDs only in a range of 0 ≤ h (mm) ≤ 3 at V_a = 10 kV or of 0 ≤ h ≤ 7 at V_a = 15 kV with a gap of 1.5 mm between the electrodes (d) and in a range of 0 ≤ h ≤ 1.5 at V_a = 10 kV or of 0 ≤ h ≤ 4.5 at V_a = 15 kV with d = 2.5 mm. No ignition occurred when the interface was located below the anode (h < 0) at both tested voltages.

First, we experimentally studied the morphology of EDs at d = 1.5 mm and V_a = 15 kV. Typical resulting ICCD images for selected positions (h) of the interface are shown in Figure 2. When we decreased h to ∼3 mm, the spherical plasma region became bigger as indicated by the radial expansion of the plasma discharge volume in Figure 2(a–h). In addition to this increased plasma discharge volume, additional plasma filaments originating from the plasma sphere became more visible as h further decreased (Figure 2(i–m)). When h approached the tip of the anode, we observed an abrupt and significant change in the morphology of the EDs (Figure 2(n–o)). At h = 0.25 and 0 mm, the plasma volume expanded only in the water, resulting in a thin pancake-like plasma confined near the water-heptane interface.

Second, to characterize the stochastic nature of the discharge in water, we introduced the concept of discharge probability (DP), which is the percentage of successful EDs out of more than 200 applied HV pulses for a given interfacial position at a given d and V_a. Given that we could not generate EDs at either h > 7 mm or h < 0 (DP = 0) in the case of d = 1.5 mm and V_a = 15 kV, which means that we could not generate EDs in pure water nor in heptane with 15 kV in the present electrode configuration. As shown in Figure 3, as we moved down the position of the interface closer to the tip of the anode (h was varied from 7 to zero), DP increased curvilinearly and reached 100% at h ≈ 2 mm at V_a = 15 kV. When V_a = 10 kV, the maximum DP (= 98%) was achieved at h ≈ 0.25 mm. However, DP dropped suddenly when h approached the tip of the anode (h = 0) with DP = ∼10 % and ∼45 % for V_a = 10 kV and 15 kV, respectively, with localized luminous plasma volumes along the interface (Figure 2(o)). Figure 3 also indicated that we can increase EDs in pure water by simply adding a covering layer of n-heptane (called the low dielectric material below) slightly above the tip of the anode.

We used the injected charge (Q_injected) to quantify each discharge characteristic of a given condition. Q_injected was determined from the corrected absolute current profile by integrating it, which we deduced from the measured current by subtracting a current with no discharge (misignited) at the same h and V_a.²⁰ Figure 4 shows the resulting injected charges for d = 1.5 mm and V_a = 15 kV as a typical example. We found that when the interface position (h) approached the tip of the anode, Q_injected increased to reach a maximum (⁠$6.7 μ C$⁠) at h ∼1.25 mm and dropped suddenly with further decreased h. This finding agrees with the increased plasma discharge volume (Figure 2) and with the DP trend (Figure 3). Even in the range of h exhibiting DP = 100%, a maximum Q_injected exists, suggesting that there is an optimal location of the interface relative to the position of the tip of the anode to create optimal ED. Note that the current integration was conducted over $1 μ s$ due to its significance up to this time scale (see a Supplementary Material). However, when we narrowed the integration time down to 25 ns, Q_injected could be obtained as 22 nC and this is comparable to that reported by Marinov et al.⁹ as 20 nC for the same 25 ns integration.

To understand the role played by the position of the water-heptane interface, we simulated a static electric field (EF) using a commercial software (Comsol Multiphysics). Although a dynamic model should be a preferred approach to fully explain the experimental result, we believe that a static model is sufficient to explain the role of the interfacial position on the onset of discharges because it is primarily governed by field intensity based on previous literature.¹⁵ The exact experimental configuration was calculated in three dimensions. The spatial distribution of the EF for various interface positions in the y-z plane is presented in Figure 5(a) for d = 1.5 mm and V_a = 15 kV. As the position of the interface approached the tip of the anode from 7 to 0.05 mm, we observed two phenomena. First, the intensity of EF at the tip of the anode increased from 3.7×10⁸ to 8.3×10⁸ V/m (more than two fold) when h decreased from 7 to 0.05 mm (Figure 5(b)). For h > 10 mm, the effect of the low dielectric material became negligible. For instance, the EF for h = 10 mm was 3.3×10⁸ V/m whereas it was 3.2×10⁸ V/m in pure water (without the low-dielectric-material layer). Second, the volume of the high EF near the anode increased as the interface approached the tip of the anode. This expansion in volume most likely influenced the discharge dynamics shown in Figure 2. When the position of the interface was between the electrodes (–1.5 < h (mm) < 0), the intensity of the EF at the tip of the anode dropped significantly and became insufficient to ignite the discharge in the low dielectric material (see the image at h = –1 mm).

According to the simulation, DP should be maximal at h = 0, which was not the case in our experiment where a sharp decrease in DP occurred when h approached the tip of the anode. This can be partly attributed to the reduced water volume near the tip of the anode because the discharge occurred only in the water and not in the heptane as shown in Figure 2.

We were not able to generalize the relation between the experimentally obtained DP and numerically estimated E_anode for various gap distances and applied voltages, partly because the field intensity was very sensitive to the shape of the electrodes and the location of the interface. However, we could deduce a reasonable correlation between them. For instance, to obtain DP = 80%, the corresponding interfacial locations were h = 0.5, 3, 0.2, and 1.8 mm for the cases of (V_a = 10 kV, d = 1.5mm), (15 kV, 1.5 mm), (10 kV, 2.5 mm), and (15 kV, 2.5 mm), respectively; the estimated E_anode (×10⁸ V/m) for each condition was 4.3, 3.7, 4.4, and 4.7, respectively.

Our finding of the ED with the very different dielectric permittivities (low and high ε) of the two layered liquids can be conceptually understood as follows. First, the physical problem can be explained as the presence of static electricity at the interface of two dielectric media. We set E_x and E_y to be tangential components and E_z to be the normal component of the EF. According to classical electrostatics,²¹ at the interface of two dielectrics, the tangential components of the EF in both media should be matched, say E_x,water = E_x,heptane and E_y,water = E_y,heptane. Meanwhile, the density of the electric flux should be conserved in a direction normal to the interface such that $ε water E z , water$ = $ε h e p t a n e E z , heptane$⁠. Therefore, the normal component of the EF is intensified by the ratio of $ε water / ε h e p t a n e$⁠, i.e., 40 times on the water side.

In summary, by simply placing a covering layer of a low-dielectric material on distilled water, we could increase the EDs from the same geometrical configuration and under the same operating conditions in water alone that could not generate EDs. Addition of the low-dielectric layer increased the discharge probability from 0 to 100%. The plasma volume also increased when the position of the interface between the water and the low-dielectric material approached the tip of the anode. Practical processes that use plasma in liquids, such as water purification, can be improved by the addition of a layer of low-dielectric material. Moreover, we believe that the low-dielectric layer can be either solid or gas. In the case of gas, the setup should be carefully built to avoid discharges in the gas. The choice of the low ε material should depend on the permittivity of the liquid to be treated, because a larger difference in ε leads to greater plasma generation and thus to improved process efficiency.