Exploring streamer variability in experiments

The goal of this experimental investigation is to systematically explore di®erences in streamers under a large variety of conditions; this will form a basis on which theory can be tested and developed. Streamers are narrow, rapidly growing, weakly ionized channels. They can be created by applying a high voltage over a non-conducting medium such as air. Streamers are used in applications because highly reactive radicals are created in their ionizing front which are very suitable for cleaning purposes in water and gas (e.g. killing of bacteria, removal of phenol, NOx, SO2, °y ash, odor and tar). Also, in the ignition of so called high intensity discharge lamps streamers are found. Although streamers show up in many desired or undesired places, especially at sharp tips where the electric ¯eld is enhanced, not many people have ever seen their wonderful appear- ance due to their low light intensity and short duration. They typically look like lightning but then with many more branches, like a tree. You can try to observe them in nature as sprite discharges high up in the atmosphere or as St. Elmo's ¯re on ships. The chances of hearing them are higher. They make a distinct hissing or buzzing sound as sometimes can be heard near high voltage lines. This thesis focusses on the start and propagation of primary streamers. Parameters that are changed during the experiments are the streamer polarity (positive and negative), the elec- trode distance (10-160 mm) and shape (point-plane and plane-plane), voltage amplitude (1-96 kV) and rise time (12-150 ns), pressure (13-1000 mbar) and gas (air and nitrogen). An intensi¯ed CCD-camera with a time resolution of » 2 ns is used to photograph the discharge. Current and voltage are digitized on an oscilloscope. A streamer is called positive (cathode directed) or negative (anode directed) depending on the polarity of the applied pulse (chapter 7). Time resolved photographs show that the positive and negative primary streamer propagation is built up of four stages: 1) a light emitting cloud at the electrode tip that evolves into 2) a thin expanding shell from which 3) one or more streamers emerge that 4) propagate through the gap. Positive streamers go through stages 1 to 4 for voltages V ¸ Vinception (chapter 3). The negative streamer propagation needs a minimal critical voltage to go beyond stage 2 (chapter 7). This volt- age appears to be around the DC-breakdown voltage in our experiments. The di®erences between positive and negative streamers disappear with increasing voltage as shown for streamers in a 40 mm point-plane gap in air at 1 bar: ² 5 kV <V <40 kV (t VDC¡breakdown): The positive streamers are thin (0.2-1 mm1), bridge the complete electrode gap when V ¸ 20 kV and branch. Their velocity ranges from 0.1 to 1 mm/ns. Negative streamers remain as a cloud near the electrode tip. ² 40 kV <V <56 kV: Positive and negative streamers have a similar thickness (1-2 mm) only the negative streamers do not bridge the gap until a voltage of 56 kV is applied. The positive streamer velocity is 1-1.5 mm/ns. ² 56 kV <V <96 kV: Positive and negative streamers have a similar diameter (2-3 mm) and energy (20-50 mJ @ 74-90 kV). Negative streamers are » 20% slower than positive streamers which have a velocity of 1.5-4 mm/ns. The remainder of the summary is devoted to positive streamers since they are easier to create at sharp tips than negative ones because of their lower inception voltage. Positive streamers become gradually thicker and faster with increasing voltage (chapter 4). For ease of comparison they are labelled type 1 to 4 based on their diameter. In a 40 and 80 mm electrode gap and voltages between 5-60 kV in air these values are: ² Type 1 streamers are very thick with a diameter of about 2.5 mm, their velocity is just over 1 mm/ns and they carry currents of up to 12 A. Their current density is about 2.4 A/mm2. They are created when V > 40 kV. ² Type 2 streamers are thick with a diameter of about 1.2 mm, a velocity of 0.5 mm/ns and currents of the order of 1 A. Their current density is about 2.4 A/mm2. They are created when V t 40 kV (t VDC¡breakdown). ² Type 3 streamers are the thinnest streamers found. Their diameter is 0.2 mm, their velocity is » 0.1 mm/ns, their current » 10 mA and their current density is about 0.5 A/mm2. They are created when Vinception . V <40 kV. ² Type 4 streamers are late, they start to propagate after streamers of type 1 or 2 have crossed the gap, their diameter appears to be similar to type 3 streamers, their velocity and current could not be determined but are expected again to be similar 1The ¯rst value corresponds to the lowest voltage in this range, the last value to the highest voltage. to type 3 streamers. Type 4 streamers occasionally connect to the already existing streamer paths of type 1 or 2. The number of streamers and branches increases up to the moment that approximately the DC-breakdown voltage is applied. At higher voltages the number of streamers and branches decreases with increasing voltage. Type 1 or 2 streamers can branch into thinner ones (type 2 or 3) provided that the electrode gap is long enough. Type 3 streamers branch often but they keep (approximately) the same diameter. In our small point-plane gaps (· 40 mm) the streamer diameter usually remains constant. Only at voltages higher than 40 kV the diameter seems to be slightly thinner near the electrodes. A thin streamer can arise via branching but it can also directly be created when a voltage just above the inception voltage is applied or when an impedance is added to the circuit such that the voltage pulse has a long rise time (> 60 ns) and the streamers start before the voltage has reached its maximal amplitude. In general, it seems that diameter and velocity are related since a certain diameter has a certain velocity regardless of where the streamer propagates in the gap. The streamer properties depend on the local electric ¯eld as is usually assumed in numerical models. Di®erences between positive streamers in ambient air and nitrogen (N2, purity 99.9%) in a pressure range of 13-1000 mbar are investigated in a point-plane gap (chapter 6). Positive streamers in nitrogen are 1) thinner, 2) curlier, 3) more intense and 4) less di®use than streamers in air. They also 5) branch more resulting in 6) a shorter distance D between branching events and 7) they propagate further down the electrode gap. The branches that deviate from the main channel however 8) die out closer to this channel. The measurements are used to search for experimental evidence of the theoretical expecta- tion that lengths and times scale with inverse pressure p (chapter 5 and 6). These results can be extrapolated to sprites at an altitude of 80 km in the atmosphere where the pressure is » 10¡2 mbar. ² For the minimal diameter d the relation p ¢ d = 0.20 § 0.02 mm¢bar for air and p ¢ d = 0.12 § 0.02 mm¢bar for N2 is found. The estimated minimal value for sprites is 0.1-0.3 mm¢bar. ² The ratio D=d gives a value of D=d = 11.6 § 1.5 for air and D=d = 9.1 § 3.3 for N2. ² The experiments in air and N2 show that the streamer velocity v increases about 0.1-0.2 mm/ns with decreasing pressure while the scaling theory predicts that v is independent of p. Here it must be noted that the measurements are done on minimal, type 3 streamers and type 2 streamers. In a 10 mm plane-plane gap, discharges are initiated by focussing and shooting a laser on the top plate (chapter 7) when a voltage near the DC-breakdown voltage is applied. On the photographs only negative discharges that consist of a streamer and evolution to glow are seen; no positive discharges can be photographed except for sparks. This is unexpected since both polarities show a streamer peak in the current signal of similar amplitude (1- 10 mA for pressures of 100-1000 mbar). In the negative current pulse the evolution to glow is also visible while this does not exist for positive discharges. Perhaps the streamer light is too faint compared to the laserspot and only the glow is seen on the pictures. It must be noted though that primary streamers in a point-plane gap are not overexposed by laserlight when the laser is shot to the needle tip. Current and voltage evolutions in the plane-plane gap furthermore show that positive and negative streamers in N2 and air at di®erent pressures are created at a similar reduced electric ¯eld of 20 § 5 kV/(cm¢bar). This thesis gives insight into streamer start, propagation and branching behavior. Its measurements are in good agreement with various experimental results reported in the literature which makes the broad parameter study reliable and suitable for comparison with results from analytical theory and numerical simulations. The conclusion of this thesis is that there is one kind of streamer whose properties vary gradually with voltage, pressure and circuit impedance as long as measurements are done in one gas and with one polarity, where it must be noted that the di®erences between positive and negative streamers decrease with increasing voltage. When measurements are done in di®erent gases, di®erent minimal diameters, velocities, distances between branch- ing events, breakdown voltages, etc. are found. The experiments have veri¯ed that lengths scale with inverse pressure even when streamers are ignited at an electrode at (near) atmo- spheric pressure: conditions that according to theory will break the scaling law since the electrode is not scaled with pressure and the nitrogen states that emit the photons that are used for photoionization are quenched at pressures above » 40 mbar. A last conclusion is that streamers made in di®erent setups show similar patterns and diameters as long as the voltage rise time, peak voltage and internal resistance are similar.