UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ignition of arc discharges at high pressures Tait, Robert Niall 1988

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1988_A7 T34.pdf [ 2.76MB ]
JSON: 831-1.0085033.json
JSON-LD: 831-1.0085033-ld.json
RDF/XML (Pretty): 831-1.0085033-rdf.xml
RDF/JSON: 831-1.0085033-rdf.json
Turtle: 831-1.0085033-turtle.txt
N-Triples: 831-1.0085033-rdf-ntriples.txt
Original Record: 831-1.0085033-source.json
Full Text

Full Text

IGNITION O F A R C D I S C H A R G E S A T H I G H P R E S S U R E S By Niall Tait Sc. (Electrical Engineering) University of Alberta, 1986 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF PHYSICS We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August, 1988 © Niall Tait, 1988 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of - S A C S The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 DE-6(3/81) Abstract This report describes attempts to discover a method for starting a water walled argon arc at high pressure. It is demonstrated that addition of small amounts of acetylene to the argon causes a very small reduction of breakdown voltage at pressures of about 3 atmospheres. Droplets of distilled water and of KC1 solution are shown to effectively increase the breakdown potential of a gas. A calculation of static fields before arc igni-tion is presented, and alternative starting circuit tests are done. Parallel starting pulse injection is found inferior to the normally used series injection. An auxiliary discharge is found to aid starting slightly, and a drastic decrease in the breakdown potential due to removal of the water wall is discovered. n Table of Contents Abstract 1 1 List of Tables vi List of Figures v u Acknowledgement , x 1 Introduction 1 1.1 The Problem 1 1.2 The System 1 1.3 The Plan 2 2 Some Relevant Theory 6 2.1 Important Fundamental Processes 6 2.2 Theories of Breakdown 12 2.2.1 Townsend Breakdown 13 2.2.2 Paschen's Law 15 2.2.3 Streamer Theories 17 2.2.4 High Frequency Breakdown 19 2.2.5 Prediction of Breakdown Voltage 20 2.3 Conclusion 21 iii 3 Designing The Apparatus 22 3.1 Introduction 22 3.2 Tube Design 22 3.3 The Electrical System 24 3.4 The Gas System 26 4 Gas Additives 30 4.1 Penning Effect 30 4.1.1 Background 30 4.1.2 Theory 31 4.1.3 The Experiment 32 4.1.4 Conclusion 39 4.2 Liquid Sprays 40 4.2.1 The Experiment 40 5 Circuit Modifications 45 5.1 Introduction 45 5.2 Field Calculations 45 5.3 Parallel Pulse Injection 48 5.3.1 The Experiment 48 5.4 Auxiliary Electrodes 48 5.4.1 The Experiment • 51 5.5 Conclusion 51 6 Ignition Without the Water Wall 53 6.1 The Experiment 53 6.2 The Result 5 6 iv 6.3 Conclusion 58 7 Suggestions and Conclusions 59 Bibliography 61 A Early Designs 65 A . l The Rotating Tube 65 A.2 The Water Jet 67 B Starting Transformer 69 C Finite Element Program 70 C l The Finite Element Method 70 v List o f Tables 2.1 Ionization and Recombination Processes 12 4.2 Reactions for Metastable Argon 31 4.3 Reactions of Excited Impurities 31 v i List of Figures 1.1 Vortex Stabilized Arc 3 1.2 Standard Electrical System 4 2.3 Main Arc Components 7 2.4 Regions of the Paschen Curve 8 2.5 Ionization Cross Sections in Argon 10 2.6 Energy Levels and Ionization Mechanisms 11 2.7 Ionization Growth Characteristic . 14 2.8 Typical Paschen Curve 16 2.9 Space Charge and Field Distortion in an Avalanche 18 2.10 High Frequency Breakdown Characteristic 20 3.11 Tube Cross Section 23 3.12 Typical Waveforms for Current and Voltage 25 3.13 Breakdown Pressure and D.C. Voltage 27 3.14 Electrical and Gas Schematics . . 29 4.15 Energy/ion pair (W) for impurities in Argon 33 4.16 Probability of Ignition 35 4.17 Effect, of Penning mixtures on Paschen characteristic (slope estimated) 36 4.18 Breakdown Pressure 37 4.19 Electric Field Near a Dielectric Sphere 41 4.20 Spray Nozzle Design 42 vii 4.21 Probability of Breakdown vs. Pressure 43 5.22 Equipotentials in the Absence of a Grounded Tube 46 5.23 Equipotentials with a Grounded Wall 47 5.24 Alternative Starting Circuits 49 5.25 Typical Mercury Lamp Starter 50 5.26 Probability of Breakdown vs. Pressure 52 6.27 Success Rates for Various d.c. Voltages 54 6.28 Success Rates for Various d.c. Voltages 55 6.29 Representation of the Water Wall 56 6.30 Paschen Curves and The Water Wall 57 A.31 Rotating Tube Cross Section 66 A.32 Water Jet Schematic 67 vm Acknowledgement I w o u l d l ike to t h a n k D r . C u r z o n for ge t t i ng me s ta r ted on this p ro jec t and p r o v i d i n g m a n y he lpfu l suggest ions. A . C h e u c k loca t ed m y equ ipment and was general ly ind i spen-sible a r o u n d the l a b . J . Bosnia deserves t hanks for design advice a n d for the use of his shop, and E . W i l l i a m s assistance i n o b t a i n i n g glass tubes should be noted . L . D a S i l v a was he lpfu l as the p l a s m a physics group c o m p u t e r consu l tan t . In o rde r to cover anyone I m a y have neglec ted to men t ion by name , I w o u l d l ike to acknowledge the suppor t of the ent i re p l a s m a phys ics group. F i n a l l y , the efforts of D r . A h l b o r n i n g u i d i n g the course of t h i s research and the deve lopment of th i s thesis were espec ia l ly apprec ia ted . H i s charac te r i s t i c ded ica t ion and en thus i a sm were essent ial to the c o m p l e t i o n of this work . ix 4. Chapter 1 Introduction 1.1 The Problem Interest has revived in the study of electric arcs. Although known for almost 200 years, recent, commercial applications to gas heaters, light sources, and materials processing has sparked new study of the phenomenon. One area of investigation is in high power arc lamps, and it has resulted in this report. The topic, of inquiry is the starting of a vortex stabilized water walled arc. For high power radiative output, high gas pressure is required, and this discourages ignition. A solution is to start at low pressure(% 3 atm) and change to operating pressure(% 7 atm) after the arc is established, but this requires a large and expensive compressor. The objective is to find a method for starting the arc at full pressure, hence eliminating the compressor. 1.2 The System The lamp discussed is vortex stabilized, meaning it has a strong gas vortex which creates a pressure gradient and centers the arc. The arc vessel can then have a larger radius than is possible with wall stabilization, reducing heat flux to the wall and allowing quartz to be used as a wall material. For the high power lamp studied here (40-120 kW radiated power), external cooling is insufficient and a water film on the inside of the quartz tube is used to control the temperature. The water is injected with large spin so that it adheres 1 Chapter 1. Introduction 2 centrifugally to the wall (see Fig. 1.1). The arc vessel for this lamp is a 27 mm i.d. quartz tube with a 1.5 mm wall. The gas and water are injected by pairs of inlet nozzles which give an azimuthal velocity component. Electrodes are tungsten tipped copper, for slow erosion and high thermal conductivity. These are attached to brass tubing which delivers cooling water to the interior of the electrodes. The anode is cylindrical, with the corner rounded to discourage the arc from attaching to the edge of the electrode. The cathode is a 45° cone , a shape which stabilizes the arc by encouraging the cathode spot to attach at the center. Power is supplied by a 208 V d.c. supply with a current rating of 500 A (Fig. 1.2). A pulse to break down the gap is specified at 60 kV and 4 MHz [5, 24, 27]. 1.3 The P l a n To address the problem presented, three main approaches are to be considered. The first involves the gas, attempting to reduce the breakdown voltage by mixing impurities with the arc. medium. The next is to vary electric field configurations inside the vessel in order to achieve lower breakdown potentials. The final approach is to decrease sparking voltage by removing the circulating water wall. Work on additives to the gas begins with attempts to exploit Penning ionization in the device. This requires energy transfer from argon atoms in metastable states to molecules of an impurity gas, and its significance depends on the importance of colli-sions in ionization of the gas. Some background on the possible ionization processes is therefore useful. This information can also be applied to other efforts at changing gas characteristics, which include spraying of droplets of distilled and of ion filled water into the gap. The next experiments concentrate on how the circuit is configured, essentially on how Chapter 1. Introduction 3 deionized water Figure 1.1: Vortex Stabilized Arc apter 1. Introduction 400 A 450 A A B C 240 V Z4> current shunt t reflector 4 M 1.6 fi s tar t ing circuit Figure 1.2: Standard Electrical System Chapter 1. Introduction 5 the preionization is introduced. Both an alternative method of starting pulse injection and the use of an auxiliary d.c. arc are considered for starting the lamp, hence an understanding of both a.c. and d.c. gas discharges is enlightening. The remainder of the work is aimed at showing how the circulating water wall dras-tically increases the voltage required to break down the gap. This is done by measuring voltage-pressure characteristics with and without the wall. Of course an apparatus of versatile design is required, with facilities for varying gas mixture , pressure and electrode geometry. The apparatus must also have characteristics consistent with the lamp system to be modelled. Armed with a basic understanding of gas discharges and an adequately designed system, it is possible to proceed to experimental work. Chapter 2 Some Relevant Theory The arc of interest in this study is a high pressure, long gap discharge. It is a d.c. arc, but is started by a high frequency pulse. Such conditions are not well suited to theoretical study, but some mention of theoretical aspects is still deserved. An understanding of the basic, characteristics of gas discharges will help in explaining the success or failure of the experiments to follow. The description is brief and chiefly qualitative. Basic kinetic theory of gases is neglected completely, but any gas discharge book will give an adequate review [17, 22]. 2.1 Important Fundamental Processes In order to understand the important processes occurring in the arc, one must consider the main components involved (Fig. 2.3). These include electrodes mounted at gap d and held at potential difference V and a gas at some pressure p. For an arc to exist the gas must be ionized and switch from an insulating to a conducting state. The easier it is to ionize the gas, the easier it will be to start the arc. For a given pressure p and gap width d there is a voltage Vj, above which breakdown and hence ignition of an arc is possible. It has been observed experimentally that 14 is a function of the product pd only, a relationship which is known as Paschen's law. This experiment is concerned with the law in a region of high pd, and involves locating the curve by setting the voltage and lowering pressure until breakdown occurs (see arrow in Fig. 2.4). 6 Chapter 2. Some Relevant Theory ( anode potential difference V cathode Figure 2.3: Main Arc Components Chapter 2. Some Relevant Theory 6 Figure 2.4: Regions of the Paschen Curve Chapter 2. Some Relevant Theory 9 Basically ionization, and hence breakdown, is a result of a collision of a neutral particle with an electron, ion, other neutral, or photon. Also possible as a result of collisions is excitation, in which an orbiting electron of an atom or molecule is raised to a higher energy level than it occupied in the ground state. For a complete discussion of conduction and breakdown in gases, mechanisms of electron liberation at the electrodes should be included. For the long gap (d = 10cm) considered here, it was expected that electrode processes would not be significant, but variation of breakdown voltages as electrodes became contaminated indicates otherwise. The discussion therefore concentrates on gas processes but briefly mentions secondary processes at the cathode [17]. For a more complete discussion, see [20, 22]. For breakdown to occur, ionization must obviously exceed the loss of charged particles. Of the charged particles present in a gas, the electrons are accelerated most rapidly in the electric field. When a fast electron collides with a molecule or atom, and the electron kinetic energy is greater than the ionization energy of the molecule, a positive ion and two slow electrons can result. Ionization may also be produced by a much slower electron, with energy lower than the ionization energy, incident on a molecule already raised to an excited state. This process becomes likely only if the lifetime of the excited state is longer than the average time between collisions. A metastable state has a lifetime T % 10~2 seconds, as a direct transition back to ground state is forbidden, and is far more likely to participate in a reaction than a normal excited state with a lifetime r % 10~8 seconds. Of course ions are also accelerated by the electric field. If an ion and an atom collide with low energy, the collision is essentially elastic. At higher energies, an inelastic collision causing radiation or electron liberation can result. Experiments have shown ionization by collision with unexcited neutral atoms or positive ions is unlikely in pure gases [15, 31], howyever in mixtures of gases charge transfer is possible, the probability of occurrence Chapter 2. Some Relevant Theory 10 io1 io2 io3 10' io5 io6 Energy, eV Figure 2.5: Ionization Cross Sections in Argon Chapter 2. Some Relevant Theory 11 15.76 eV 11.72 eV 11.55 eV met ast able levels direct impact ionization multi step ionization Penning ionization 11.41 eV Argon Levels Acetylene Levels Figure 2.6: Energy Levels and Ionization Mechanisms being large if the difference in ionization energies of the two participants, is small. In similar conditions Penning ionization becomes likely, as described in Chapter 4. Another possibility is photoionization, where a molecule is ionized by a photon of energy greater than the ionization energy of the substance. The process can also occur in steps, especially in a high concentration of excited atoms. Cross sections for collisions of photons and other particles with argon are given in Fig. 2.5. The energy transfer paths for various ionization processes are given in Fig. 2.6. In competition with these processes to determine the fate of the discharge are a number of deionization reactions, including recombination or attachment and diffusion or drift. Chapter 2. Some Relevant Theory 12 Process Reaction electron impact ion impact photon impact energy transfer e~ + M —• 2e~ + M+ x+ + M -> z + + M + + e~ hu ^ M M + + e" M * + X -» M + A'+ + e~ recombination attachment c- + M + + X -> M + X e~ + M + X -+ M - + A r Table 2.1: Ionization and Recombination Processes Recombination of an electron and an ion results in release of energy to a third body or as radiation. Transfer to a third body is the most probable process, hence recombination is most likely at high pressures. Attachment occurs in a molecule with an unoccupied energy level in its outermost group. A colliding electron may then enter this level, creating a negative ion. Dissociation may also occur with attachment, and ions may be produced in excited states. Gases with this behavior include halogens, oxygen, sulphur, and various hydrocarbons. Probability of attachment is enhanced by the presence of a third body, so like recombination, it is more likely at high pressures. Diffusion and drift account for losses simply by aiding recombination at electrodes and boundaries, which act as a third body in the reaction. 2.2 Theories of Breakdown The ionization processes described dominate the losses for electric fields higher than the breakdown field at a particular pressure. The objective of this thesis is to obtain the lowest possible breakdown voltage at the full operating pressure. Several theories exist for anticipating this voltage. Chapter 2. Some Relevant Theory 13 2.2.1 T o w n s e n d B r e a k d o w n In breakdown by the Townsend mechanism an initial electron avalanche (rapid multipli-cation of free electron numbers by collisional ionization) is started by an external factor such as background radiation. This avalanche stimulates secondary avalanches, evolving into an unbroken chain of avalanches known as a self sustaining discharge. Because of the statistical nature of the processes involved, not every externally produced electron will trigger a self sustaining discharge. The electrons beginning the secondary avalanches may be produced by several dif-ferent processes. At the cathode they may be released by impact of positive ions from previous avalanches, by impact from photons, or by impact of metastable molecules whose long lifetime has allowed to diffuse to the cathode. The simplest approach to the growth of ionization is to consider only collisions of elec-trons with gas molecules. If ? i D electrons initially leave the cathode, and each undergoes a ionizing collisions per centimeter, Now n — n0 ionizing collisions must have occurred between the electrodes. Defining a secondary ionization coefficient 7 as the average number of secondary electrons per ionizing collision in an avalanche, then dn = an dx so that the number of electrons striking the anode at x = d. is; secondary electrons are produced at the cathode. Chapter 2. Some Relevant Theory 14 Figure 2.7: Ionization Growth Characteristic The first generation of secondary electrons have now been accounted for. These will in turn produce (^(e0^ — 1 ))7n0 secondary electrons. Ultimately the discharge will involve n = nc6*d(l + x + x2 + x3 + •• •) secondary electrons, where x — 7(e° l J — 1). If x < 1 the series converges to n — 1 - 7(e«< - 1) If x > 1 however, the series diverges and the corresponding current grows without bound. The criterion for the onset of a self sustaining discharge is then 7 ( e ° ^ - l ) > l This criterion is valid for any collisional processes as well as attachment, and for any form of ionization at the cathode. Processes causing secondary ionization elsewhere, such as photoionization, affect the validity, but it still provides a good approximation. Chapter 2. Some Relevant Theory 15 2 . 2 . 2 P a s c h e n ' s L a w Often associated with the Townsend criterion is Paschen's law. If ^  = f(^) a n c l 7 = 9(^) then the criterion includes only the variables Vj, and pd, . For any value of pd there is then only one value of Vb (Paschen's law, see also Fig. 2.8). In the lab one soon discovers that Vb is not well defined, and there is actually a range of values for which breakdown occurs with varying probability. To understand Paschen's law on an atomic, level, suppose each electron participates in an average of z collisions in crossing the gap d. If the electron mean free path is X — 1/no, where n is number density and a is cross section, then d = z • X p-d = (nkT)-(z)-( —) na p . d = — - z a . This indicates that if pd is kept constant, so is the number of collisions, z. The result of this is that the plot Vb against pd exhibits a minimum. For higher values of pd the mean free path is too short to produce electrons energetic enough to optimize the number of ionizing collisions. For values of pd below the minimum, there are not enough collisions to sustain an efficient discharge. The law loses validity at very high pressure or field strength, as shown in a number of experiments [22, p. 560]. In these situations as p varies but pd remains constant the curve will shift, indicating Vf, is no longer a function of pd only. Up to at least 10 atm. deviation from the law is only a few percent, and breakdown is still well characterized by the Paschen curve. apter 2. Some Relevant Theory ID CM O ro 0.0 10.0 20.0 30.0 40.0 50. P r e s s u r e * G a p , pd T o r r cm Figure 2.8: Typical Paschen Curve Chapter 2. Some Relevant Theory 17 2 . 2 . 3 Streamer T h e o r i e s Studies of the temporal growth of ionization began to cast doubt on the accuracy of the Townsend theory at high pressure 1 atm) [30]. The transit time of the ions required in Townsend's secondary process was a factor of 100 longer than the times measured. This led to the streamer theories, which require two main mechanisms; 1. photons from the head of the avalanche produce free electrons in the gas by pho-toionization 2. space charge produced in the avalanche causes sufficient field distortion to move these electrons towards the avalanche head and thereby generate further avalanches. The electrons in the avalanche can be assumed to have a spherical distribution, with the radius determined by diffusion. The rapidly moving electrons leave a tail of positive ions. This space charge configuration causes enhancement of the field between the anode and the avalanche head, but a reduction between the electrons and the ion tail (Fig. 2.9). This distortion of the field becomes important for electron numbers of 106, and streamer propagation is likely for 108 electrons per avalanche. It was believed that this theory was necessary to account for breakdown at high values of pd, but it appears Townsend theory can be made satisfactory. In a case where the cathode is a large distance from the head of a developing avalanche, cathode secondary processes are unlikely, and streamer theory gives a reasonable explanation. However, allowing photoionization as the dominant secondary process and including space charge in the Townsend breakdown criteria, the theories become essentially identical [17]. Chapter 2. Some Relevant Theory 18 Figure 2.9: Space Charge and Field Distortion in an Avalanche Chapter 2. Some Relevant Theory 19 2.2.4 High Frequency Breakdown The mechanisms discussed thus far have been for d.c. discharges. The arc ignition in this experiment, however, is triggered by a high frequency pulse. An a.c. discharge is basically the same as a d.c. discharge for low frequencies, that is, when the time for one cycle is much longer than the ion transit time. For higher frequencies, the breakdown voltage varies from the d.c. value. This can be explained by considering that the breakdown requires the production of electrons to be greater than the recombination losses. With an a.c. field, frequency increase reverses particles in their motion, preventing them from reaching electrodes and boundaries, and thus being lost. Production of electrons is increased, as particles undergo ionizing collisions during more than one traverse of the gap. Of course fewer ions then reach the electrodes to produce secondary electrons, but as frequency increases the rise in multiplication factor becomes dominant. As frequency is further increased, electrons oscillate with amplitude much smaller than the gap length, and the only loss is then diffusion. This balance between electron production and diffusion loss is the basis of a diffusion theory of breakdown [7, 10]. If the field is considered uniform at any instant, then the breakdown criterion is; vxne -f D e V 2 ? z f = 0 where u7 is the ionizing collision rate, ne is the electron density, and De is the electron diffusion coefficient. Solving the one dimensional problem for ne = 0 at x — dtd/2 for d the gap length, Note that = au in terms of drift velocity u and Townsend's primary ionization coeffi-cient a. As frequency continues to rise, oscillation amplitude shrinks, and energy transfer to Chapter 2. Some Relevant Theory 20 Figure 2.10: High Frequency Breakdown Characteristic the gas is decreased. At this point breakdown voltage begins rising, proportional to the frequency. The resulting voltage-pressure characteristic for high frequency exhibits two minima, the usual Paschen minimum, and also a minimum at the pressure for which electron amplitude becomes equal to the discharge length, and losses are suddenly reduced. How does a high frequency discharge switch to d.c? As the mechanism is essentially the same, it should be a matter of the h.f. discharge creating sufficient ionization to permit the slow and less efficient d.c. process to take over, without the gas returning to an insulating state. This is the breakdown process to be studied here. 2.2.5 Prediction of Breakdown Voltage Attempts to predict, breakdown voltage using any of these theories require knowledge of ionization coefficients, and this information is not generally available. The normal solution is to make simple assumptions regarding the breakdown criteria or coefficients, Chapter 2. Some Relevant Theory 21 but data in the region of interest is still necessary. The easiest method of prediction is to obtain a Paschen curve, so that for any pd a unique Vb can be found. The accuracy of these theories is reduced at higher pressure because electrons and ions approach ther-mal equilibrium, and many particle interactions involving van der Waals forces become important. 2 . 3 C o n c l u s i o n From this discussion a few points relevant to the experiment become clear. The success of changes to the gas will depend on increasing ionization by decreasing ionization potentials or increasing cross sections, while minimizing recombination and attachment. It also appears that at high pressure the high frequency breakdown will be much easier to create than the d.c. arc. This suggests some freedom as to the manner in which the preionizing pulse is applied. Chapter 3 Designing The Apparatus 3.1 Introduction To investigate breakdown at high pressure, it is necessary to first design a device to operate under the conditions of interest. It must be versatile enough to allow testing of gas and field characteristics as previously proposed, yet still simulate conditions of the lamp system accurately. 3.2 Tube Design As with many projects of this nature, the first few designs were aborted without producing many results. These attempts are described in Appendix A. The physical aspects of the design were simple to choose, with the exception of the method for generating the water wall. Dimensions of the tube and electrodes imitate the actual lamp, with a 2.7 cm diameter tube and 1.25 cm diameter electrodes separated by a 10 cm gap. Details such as electrode cooling were omitted due to the short oper-ating time of the experimental discharge. It was very likely that the water would have some effect on the starting characteristics, and several approaches were tried to obtain the best compromise between simplicity of operation and accurate representation of the characteristics of the water wall. The first attempt involved rotating the entire tube, taking advantage of centrifugal force to distribute the water around the wall. Inconvenience for gas filling, and an 22 Chapter 3. Designing The Apparatus gas water j t pyrex tube water Z3T plexiglass tube aluminum foil N water jets Figure 3.11: Tube Cross Section Chapter 3. Designing The Apparatus 24 impractical overall design caused this device to be abandoned. Another attempt used a pump and water jets to generate the wall, but it was noticed that the critical aspect was a film of water covering the metal strip that guides the initial discharge. Most experiments were then carried out with a water film along the bottom of the horizontal tube, and covering this foil (Fig. 3.11). This strip created some initial problems. The lamp has a reflector which acts to capacitively couple the gap. Initially the reflector was simulated by an aluminum bar, but arcing external to the tube resulted. It was soon found that a strip of aluminum foil held to the outside of the glass by electrical tape worked well. It should be pointed out that the tube used was pyrex, rather than the more expensive quartz, as thermal stress and absorption spectra were not a concern. Also, electrodes were brass, rather than tungsten tipped copper, but since electrode processes are not being studied, this should have no effect on results. The entire tube was placed in a plexiglass enclosure, in case of a catastrophic failure under pressure. 3.3 T h e E l e c t r i c a l S y s t e m If the growth of ionization is considered, it is found that the time required for a discharge to form is only the order of microseconds (Fig. 3.12). To study the initiation of break-down it is then necessary only to look at a few milliseconds at most. For this reason it was decided that large capacitors could be used to act as a power supply. To charge the capacitors, a high voltage supply of a few milliamps was connected. The starting pulse was supplied by a transformer manufactured by L.P. Associates of California. This was one of the transformers used in earlier development of the lamp system [27]. Other com-ponents were high voltage charging resistors, a high current dump resistor, and solenoid to dump the capacitor charge when power was off. Chapter 3. Designing The Apparatus 25 Chapter 3. Designing The Apparatus 26 The operation of the circuit (Fig. 3.14) is initiated by switching on the power supply, which activates the solenoid and allows the capacitors to charge. Charging voltage is monitored at the supply. Once the d.c. voltage is established, breakdown is encouraged by a pulse from the starting transformer. The pulse is triggered either manually by a switch or by a timer controlling a solid state relay. The timing was set for a 200 ms pulse with 20 seconds recovery time between pulses. Details of the starting transformer are given in Appendix B. Voltage levels remained fixed throughout the experiment, while pressure was varied. The idea was to set conditions such that breakdown would result at approximately the running pressure used in an actual lamp. Conditions would then be varied to obtain breakdown at much higher pressure. The desired effect is essentially a shift of the Paschen curve to a higher pressure, and the method used to look for this shift is to sample a range of pressures along a line of fixed voltage, and thus to find where the curve crosses this line. To obtain initial conditions similar to a lamp, the voltage level was chosen to be 1.5 kV so that the capacitors discharged for a tube pressure about 30 psig. The effect of changing this voltage was investigated by checking the number of successful discharges in 25 attempts with a variety of pressures and voltages (Fig. 3.13). 3.4 The Gas System Since there is no gas flow through in the system used, it was necessary to evacuate the tube to minimize contamination of the argon fill gas by air. A mechanical pump capable of reducing pressure to a few hundred millitorr absolute was used. Poly-flo tubing and the presence of water in the tube limited the possible vacuum to about 5 torr. This vacuum was considered sufficient, as some contamination is inevitable in a real lamp. The system for pressurizing the tube consisted of gas bottles from Linde (Union Chapter 3. Designing The Apparatus 27 o 0 . 0 10 .0 2 0 . 0 3 0 . 0 4 0 . 0 5 0 . 0 6 0 . 0 7 0 . 0 8 0 . 0 Gouge Pressure, pslg Figure 3.13: Breakdown Pressure and Bias Voltage Chapter 3. Designing The Apparatus 28 Carbide) connected to the tube via poly-flo tubing and valves. Of course two gauges were necessary, one vacuum gauge reading 0-20 torr and one pressure gauge reading 0-400 psi. More poly-flo was used to connect and isolate the two gauges, as pressurizing vacuum gauges severely limits their useful life (see Fig. 3.14). In summary, we now have a system that simulates the most important characteristics of the Vortek arc. lamp. With some modification the tube can be adapted to different geometry and circuit configurations, with gas pressures from a few torr to about 100 psig. It is possible to add gaseous and liquid contaminants, and to mount, additional electrodes. This system was used to study the variation of breakdown potential first for gas additives and then for circuit modifications. Chapter 3. Designing The Apparatus 29 supply 20 kV 30 mA 200 kfi 2pF, 8 kV 1 kn starting | transformer" trigger timed SCR -i to supply switch coupling strip gas arc vessel electr ical system argon acetylene mechanical pump ® - l — < & discharge tube © © vacuum gauge pressure gauge gas system Figure 3.14: Electrical and Gas Schematics Chapter 4 Gas Additives 4.1 Penning Effect 4.1.1 Background Gases in which metastable states form have long been known to have their breakdown voltages drastically reduced by the presence of impurities. In the 1920's Frans Michel Penning (1894-1953) at Philips Laboratories noted large discrepancies among data on breakdown voltages of rare gases previously reported by others. In 1927 he discovered that breakdown voltages of neon and argon were significantly decreased by addition of minute amounts of impurities. He concluded that the impurities were ionized by energy transfer from metastable neon or argon atoms that were present in the discharges. He based this on the fact that the effect was observed only for mixtures in which the metastable atoms of the parent gas have enough energy to ionize the impurities and that irradiation from a second lamp diminished the effect [16]. Penning's work was done at pressures of about 20 Torr, with atoms excited by low energy electrons. About 25 years later, William P. Jesse discovered a similar effect for high energy alpha particles in noble gases between 400 and 1200 Torr [13]. In measuring total ionization yield in He he discovered extreme sensitivity to even minute amounts of virtually any impurity. Soon after this discovery there was a great deal of interest in the microscopic mechanism. There is still no concensus on the path of energy transfer. It is likely that both resonance and metastable states play a role. 30 Chapter 4. Gas Additives 31 spontaneous radiation Ar* —> Ar + hu transfer by collison with argon Ar* + Ar —»• Ar + Ar + /ii/ and with other atoms Ar* + x --> Ar + X * Table 4.2: Reactions for Metastable Argon spontaneous radiation X* -•* X + hv dissociation X* -•4 A + B ionization X* -- X+ + e" Table 4.3: Reactions of Excited Impurities Work continued through to 1970, when A.E.D Heylen at the University of Leeds was studying electrical breakdown in gases. He found that argon-hydrocarbon mixtures can have a much lower sparking potential than pure argon [8]. These results inspired our experiments on the effect of argon-acetylene mixtures on starting of a high pressure water walled arc. 4 . 1 . 2 T h e o r y R e a c t i o n s The Penning effect is attributed to the collision of a metastable A* with B to leave A unexcited and B ionized. A* + B -> A + B+ + e~ This project is concerned with an arc in argon, therefore we are interested in the long lived excited states of argon whose fate is determined by competition between radiative emission due to collisions with ground state argon atoms and energy transfer due to collisions with impurities. The most likely reactions for argon are given in Table 4.2, and for the impurity X in Table 4.3, where a * indicates an electronic excitation of unspecified level. The contribution to total ionization due to energy transfer to A' is significant, only if ionization is much more likely than dissociation or radiation. Christophorou et.al. Chapter 4. Gas Additives 32 claim evidence that energy transfer to X from Ar is the dominant reaction of the first set [28]. As previously mentioned, the Penning effect is most noticeable for additives X of ionization energy slightly below the resonance and metastable states of the host. This is tabulated in Fig. 4.15 [12], which gives the energy required to produce an ion-electron pair, W, against ionization energy for a number of impurities in argon. Examination of the candidates exposes the fact that one of the most easily obtainable gases is likely to give the best results. Acetylene seems the ideal additive to use. Of course water vapour, which causes a slight Penning effect, is always present in a water walled arc chamber and will likely affect results. Also to be considered is the tendency of acetylene to dissociate at pressures above atmospheric, especially in the presence of some metals which catalyse the reaction [25]. Experimental work using both high energy a particles [12, 28] and low energy elec-trons [8] as ionization sources has shown a significant decrease in energy required to create an electron-ion pair in argon-acetylene mixtures as compared to pure argon. This should correspond to a decrease in breakdown voltage for the gas mixture even at several atmospheres pressure. Heylen showed drops in sparking voltage of about 70% at pressures of a few hundred torr. It is hoped that the effect is also present at higher pressure and in the presence of a water wall. 4.1.3 The Experiment Having constructed the apparatus described in Chapter 3 , the experiment is a matter of filling the tube with a gas mixture and lowering the pressure until breakdown occurs. Mixing the gas consists of first pumping the discharge vessel to about 5 torr. Acetylene is then admitted in the required amount, and the argon is added to bring total pressure, Chapter 4. Gas Additives w J ° r _ A _ T & 0 I > 4-,. ID _L3. 14 12 17 5 i i i i i i 11 Resonance levels t 3pb5s 3 Pi 14.09 eW, 3p55s 3 P a 14.26 eV-H Metastable levels 1 3p 54 5 3 P : 11.55 eV| 3pHs 3 P 0 H-72 eV, Resonance levels i 3p545 3 P 3 11.62 e V '-3p54s 3 P 0 11.83 eV| 10 12 Ionization Potential (eV) 14 16 ! Gas 4 (eV) W (eV) mixture W (eV) pure gas 1 Argon Ar 15.76 26.4 26.4 2 Acetylene C2H2 11.41 20.4 27.64 3 nButane 10.8 22.9 26.18 4 But ene C4Hs 9.72 22.2 27.1 5 Carbon Dioxide CO, 13.78 26.0 34.45 6 Carbon Monoxide CO 14.01 26.3 34.77 7 Cyclopropane C3HS 9.7 23.0 26.11 8 Ethane C2H6 11.6 24.4 26.7 9 Ethylene C2H4 10.5 23.9 28.0 10 Hydrogen H2 15.4 no effect 37.0 11 Krypton Kr 13.93 24.0 24.0 12 Methane CH4 13.12 26.0 29.26 13 Nitrogen N2 15.6 no effect 36.6 14 Oxygen o2 12.2 26.0 32.2 15 Propane C3H6 11.2 23.6 26.3 16 Propylene CzH(, 9.8 23.8 27.3 17 Wat er H20 12.6 25.2 37.7 Figure 4.15: Energy/ion pair (W) for impurities in Argon Chapter 4. Gas Additives i' 34 and hence acetylene percentage to the desired level. For 0.15% acetylene in argon, for example, 5 torr of acetylene would be added followed by 50 psig (3350 torr) of argon. Varying mixtures in this manner from 0 to 2% acetylene in argon, a few starting pulses were applied at 5 psi intervals decreasing from 50 psig until breakdown was visible. Regardless of the mixture, breakdown occurred at approximately the same pressure. However, at very low percentages of acetylene, it was noticed that the discharge always occurred, whereas at high acetylene content breakdown was seen for only a fraction of the trials. This observation indicates the Paschen curve is not as well defined as normally implied by theoretical descriptions. This led to a repetition of the experiment, with the number of successful breakdowns from 50 attempts recorded for each pressure in an attempt to find the pressure range over which breakdown could occur. This was first done manually, and then with the timer circuit, to improve uniformity of pulse length and recovery time. Statistical accuracy was limited by graduate student enthusiasm. The same method of data acquisition was used in later experiments. The results are presented in Fig. 4.16 and Fig. 4.17. The first plot shows the breakdown probability curve for different acetylene concentrations. By taking an 80% breakdown probability to correspond to the breakdown voltage Vf,, a point on the Paschen curve for each mixture can be located, as shown in Fig. 4.17 The data show some evidence of the Penning effect, although to a much lower degree than expected. For large concentrations of acetylene the breakdown voltage does tend to increase towards the value for pure acetylene. If we consider a breakdown characteristic following Paschen's law, we would expect small quantities of acetylene in argon to increase greatly the breakdown pressure for a fixed voltage in comparison to pure argon. This prediction is based on low pressure data, but as long as VJ, vs. pd is fairly linear, this should extend to a higher pressure region. Chapter 4. Gas Additives Figure 4.16: Probability of Ignition Chapter 4. Gas Additives 36 Figure 4.17: Effect of Penning mixtures on Paschen characteristic (slope estimated) Chapter 4. Gas Additives 37 o o IT) I CO Q_ D Rcetulene Percentage Figure 4.16: Breakdown Pressure Chapter 4. Gas Additives 38 There are many possible interpretations of the reduced Penning effect. The first sus-pect is the water vapour. Water does give a small Penning effect, as its ionization energy is slightly below upper resonance levels of argon. Since the argon-water collisions are more frequent than argon-acetylene collisions for the mixtures used, possible quenching of excited argon atoms by the vapour was considered. However, there appear to be no vibrational or electronic absorption bands in water that are close to the argon states con-sidered responsible for the energy transfer to acetylene. Other impurities in the system could hide the effect of the acetylene, since the breakdown voltage of what was believed to be pure argon could in fact have been a value reduced by Penning ionization of unknown impurities. The chemical properties of acetylene are the second suspect. As mentioned it can dissociate or form negative ions, reactions which are enhanced at higher pressures. The third suspect is the breakdown mechanism. There is no real concensus on which mechanism is dominant at this pressure. It is fairly certain that electrode processes lose importance for long gaps at high pressure. It is also possible that photoionization becomes more important relative to collisional ionization. This would reduce the Penning effect, but it should still show a dependence of breakdown pressure on acetylene content. The remaining factor that should be considered is the experimental method itself. More data would increase accuracy, but the large effects of interest here should have appeared in the results already gathered. The recovery time between shots was set by the recovery time of the electrical system, which should be significantly longer than the time required by the gas. Temperature change due to repeated shots was also consid-ered. I measured only 0.2°C change after 50 shots. From [3] I found a 10°C change affects sparking voltages in air only 2%, and discarded the notion of any error caused by temperature fluctuation. Chapter 4. Gas Additives 39 4.1.4 Conclusion The Penning mixtures used in this experiment showed a small increase in possible break-down pressure compared with pure argon, however this effect was much smaller than expected. I have suggested several possible causes of the reduction in the measured ef-fect without rigorous justification. The change in starting pressure resulting from these mixtures is not sufficient to make their use worthwhile, and possible causes indicate no method for improving the effect, therefore different methods of improving the breakdown characteristic must be examined. Chapter 4. Gas Additives 40 4.2 Liquid Sprays Reducing the breakdown potential requires increasing the ionization coefficient or elec-tric field, or decreasing the gap length or pressure. Since changing the pressure or gap is contrary to the objective of this work, and increasing the ionization coefficient through Penning ionization failed, the field is considered next. Virtually any text on electromag-netic theory has an example problem calculating the electric field near a dielectric sphere immersed in a uniform electric field. The result is an enhancement of the field near the dielectric, as field lines curve to meet the surface normally (Fig. 4.19). Could this field enhancement effect be applied to promote the breakdown of a gas? 4.2.1 The Experiment The logical choice for a dielectric to use in this system is water, since it is already present in the tube. Although it has been shown [6] that vapour mist dielectrics increase the electrical strength of insulating gases, the effect of a distilled water spray was tested. The effect of a spray of salt solution was also checked, encouraged by the evidence [4] that metallic vapour decreases the electrical strength of a host gas. The apparatus previously used was slightly modified to introduce the spray. The mist was generated by argon flowing through a nozzle containing a capillary tube to the water reservoir. Another tube to a stagnant point in the flow created the pressure difference required to move the water (Fig. 4.20). The flow rate was about 0.1 cc/sec. The device generated a spray with average droplet size estimated to be about lOOfim from condensation of droplets on the wall. Finer mist could be generated ultrasonically, but problems obtaining the equipment killed this idea. The spray was introduced to the tube through a hollow cathode, with flow rate controlled by a vent valve at the other end of the discharge tube. Figure 4.19: Electric Field Near a Dielectric Sphere Chapter 4. Gas Additives 42 spray of water droplets argon J" "V-reservoir Figure 4.20: Spray Nozzle Design Chapier 4. Gas Additives 43 Figure 4.21: Probability of Breakdown vs. Pressure Chapter 4. Gas Additives 44 The results presented (Fig.4.21) are for a 5 psi drop across the nozzle, and other conditions as previously; 1.5 kV bias, 10 cm gap. The distilled water spray reduced the possible breakdown pressure, as expected. A metal vapour mist was difficult to produce, so the experiment was repeated with a saturated solution of KC1. This made the mist conducting, as it now contained K+ and Cl~ ions. The puzzling result was that the spray now prevented breakdown at any pressure! The most likely explanation of the effect of the distilled water involves the affinity of small droplets for charged particles1. This probably removed a number of the free electrons required to begin breakdown. The effect of the KC1 solution has no obvious cause, although the droplets may again have been responsible for absorbing electrons. The solution also made the water in the tube slightly conducting, and by providing a current path it may have been responsible for energy losses large enough to inhibit breakdown. Altogether this sequence of experiments gave no significant reduction of the break-down voltage at high pressure and was therefore terminated in order to look into effects associated with the electric field distribution. Remember Millikan Chapter 5 Circuit Modifications 5.1 Introduction When considering the breakdown of a gap, the characteristics of the external circuit should be considered as well as those of the gas. Of course the field configuration is very important, and some knowledge of it is helpful in understanding the breakdown. Different approaches to designing the circuit can be taken, but it is difficult to compare designs directly as different power requirements may exist. As this study is concerned only with obtaining breakdown, any circuit not requiring additional complicated or costly elements is a candidate. 5.2 Field Calculations An attempt was made to calculate the electrostatic field for an electrode geometry similar to that used in the experiment. The program used was a slightly modified version of that presented in [21]. The field was calculated for two electrodes mounted in a grounded tube and with ro-tational symmetry. Plots of a half cross section of the tube are presented with (Fig. 5.22) and without (Fig. 5.23) the grounded metal tube outside the glass wall. Both include a water wall. The effect, of the grounded wall is to enhance the field both axially and radially. Other parameters such as presence of space charge and of the water were varied, but with minimal effect. The qualitative characteristics for each set of conditions could 45 Chapter 5. Circuit Modifications • c 0.00 0.01 0.02 R(m) 0.03 0.04 0.05 Figure 5.22: Equipotentials in the Absence of a Grounded Tube Chapter 5. Circuit Modifications 0.00 0.01 0.02 0.03 R(m) 0.04 0.05 Figure 5.23: Equipotentials with a Grounded Wall Chapter 5. Circuit Modifications 48 easily have been predicted without calculation. These characteristics include a stronger radial than axial field. This indicates initial breakdown is across the tube. Knowing the field distribution, it should also be noted that, placing the source of preionization in the region of highest, field will allow the most rapid evolution of a breakdown. However since this region is in the vicinity of the positive electrode the electrons produced in a preionization here will cause only a local breakdown. 5.3 Parallel Pulse Injection In order to create breakdown along a gap, it makes sense to apply a voltage pulse to the ends of the gap. But if the initial breakdown is across the tube, and the diameter of the vessel is much less than the length, it seems likely that ionization can be produced more easily by applying the pulse transversely. 5.3.1 The Experiment Once again the previously described apparatus was used, with the same gap and bias, but with the starting pulse applied to the foil strip attached to the glass (Fig.5.24). It appears that considerably more power is required to break down the gap with this method compared with the normal series injection, as breakdown could not be induced with 1.5 kV on the electrodes. The reasons may include a change in load due to the interference of the wall and water with this configuration, or perhaps simply more energy loss to corona discharge. 5.4 Auxi l iary Electrodes A fairly common practice with spark gaps and mercury arc. lamps is to include three (or more) electrodes. The extra electrode is used to trigger the gap. Generally the electrode Chapter 5. Circuit A^odificafions 200 kn 2 „ F coupling strip starting transformer r arc vessel -WvVAWr-200 kf) supply starting transformer auxiliary supply 30 kV coupling strip Figure 5.24: Alternative Starting Circuits Chapter 5. Circuit Modifications 50 R2 Figure 5.25: Typical Mercury Lamp Starter is introduced close to and parallel to the cathode. A discharge is set up between this auxiliary electrode and the cathode, generating a plasma. With this source of ionization present, it becomes easier to break down the main gap. A common mercury lamp starting geometry is shown (Fig.5.25), with the auxiliary gap firing initially, but extinguishing upon breakdown of the main gap due to the extra resistance. Chapter 5. Circuit Modifications 51 5.4.1 The Experiment For a high frequency starting pulse the mercury lamp starting arrangement is not prac-tical, as finding a resistor for both high frequency and high voltage is difficult. Most either act as spark gaps or heating elements. An alternative is to use another d.c. power supply for the auxiliary arc (Fig. 5.24). A high voltage supply was then used to put 7.5 kV at .3 raA on an electrode in the center of the cathode. The result was a small increase in breakdown pressure. The polarity of the main supply was then reversed so that the cathode was at -1500V and the auxiliary electrode at 7.5kV and .3 mA parallel to it. The result was an increase of starting pressure of only a few psi (Fig. 5.26). An increase in current would probably enhance the effect a small amount. 5.5 C o n c l u s i o n These attempts at changing the field configuration did not greatly reduce the break-down voltage. The experiments were then easily abandoned to pursue an interesting phenomenon which was observed when the foil strip was not covered with a water film. A discussion of the effect follows. Chapter 5. Circuit Modifications Gouge Pressure, PSIG Figure 5.26: Probability of Breakdown vs. Pressure Chapter 6 Ignition Without the Water Wall When investigating the ignition of an arc in a vessel with a water covered wall, it seems natural to examine the effect of that water on the starting of the arc. This was in fact quickly checked and the result indicated that without the water there was a dramatic decrease in the breakdown voltage. This inspired measurements of Paschen curves for the vessel with and without the water wall. 6.1 The Experiment In order to obtain the Paschen curve, it was necessary to measure ignition success vs. pressure curves as in previous experiments, but now for a number of voltages. The voltage range was 1.0 to 2.5 kV, with the upper limit set by capacitor ratings in the starting transformer and by the breakdown pressure considered safe for the tube. Gap length and starting pulse were identical to previous tests. The measurements with the wall were made with the reflecting strip along the bottom of the tube so that the water lying in the vessel formed a film between the foil and the electrodes. The measurements for no wall were made with the tube rotated 90° around the azimuthal axis so that the water no longer covered the strip (see Fig.6.29). Once the success vs. pressure curves (Fig. 6.27 and 6.28) are obtained, the Paschen curve is found by choosing a reference success rate so that for each voltage a pressure can be found. 53 Chapter 6. Ignition Without the Water Wall 54 o 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 Gouge Pressure, pstg Figure 6.27: Success Rates for Various d.c. Voltages Chapter 6. Ignition Without the Water Wall o Gouge P r e s s u r e , p s i g Figure 6.26: Success Rates for Various d.c. Voltages Chapter 6. Ignition Without the Water Wall 56 electrode water reflector tube wall tube wall electrode -reflector water water wall no wall Figure 6.29: Representation of the Water Wall 6.2 The Result From Fig. 6.30 there is a significant shift in the Paschen curve when the water wall is removed. The shift is even greater when the tube is dried out. A small change should be expected as the water affects the capacitance of the tube. The large change observed can be verified by considering the initial spark as a surface discharge with the foil as a guide. Then according to [35], breakdown voltage for a surface discharge on pyrex decreases with increasing humidity until the 100% level is reached. A film of water then forms on the surface, and the breakdown voltage jumps by as much as a factor of four. Although known for at least 30 years, it appears there has been very little work on the subject. A possible explanation of the effect lies in the properties of water and pyrex. The glass is a hard high melting point substance. By comparison, a water film evaporates easily, and when an arc begins to form on a water surface the liquid vaporizes and interrupts the discharge by cooling the gas and acting as an electron sink. Chapter 6. Ignition Without the Wafer Wall 57 o to LO -X oo CD cn o —> LT) IT) o -L success 20'/. success [ § • No Woter Wall A 80'/. success + 20'/. success _ j i i 0 . 0 1 0 . 0 2 0 . 0 3 0 . 0 4 0 . 0 5 0 . 0 Pressure, pslg 6 0 . 0 7 0 . 0 8 0 . 0 Figure 6.30: Paschen Curves and The Water Wall Chapter 6. Ignition Without the Water Wall (' 58 6.3 C o n c l u s i o n The phenomenon described here has some obvious potential applications. A possible solution to the problem presented in this work can now be identified. The normal result of starting the lamp without cooling water is extreme thermal stress and explosion of the tube, but it is likely that by starting at low current, or by injecting water just following the starting pulse, the lamp could be started in the absence of the wall and hence at full pressure. The effect may also have applications to high voltage insulation, as a flow of distilled water over a ceramic bushing should greatly increase it's insulating effect. This obviously will be limited by contamination, but could be applied in sealed systems such as laboratory power supplies. This speculation logically leads to suggestions for further work. C h a p t e r 7 Suggest ions and C o n c l u s i o n s Several interesting questions have emerged from this investigation, as is commonly the case with research work. These suggest further work to be done on starting of the lamp, as well as study of other areas. Of immediate interest for starting of the lamp is ignition without the cooling water vortex. An estimate should be made of the thermal stress that can be supported by the tube, and how long after ignition the fracture point will be reached in the absence of cooling. If the water can be introduced within this time, and the quenching effect does not. itself cause fracture, then it should be possible to start the arc by injecting water just following the starting pulse. This effect, of a water film on a solid surface might also be investigated as a means of improving insulation values of materials in certain cases. Wet insulators generally have reduced electrical resistance due to surface dust and the conductivity of water, but the reverse is true for a continuous film of distilled water on a clean surface. A study of the conditions, such as the importance of the deionization of the water, that improve the insulation value should be done, and from this possible applications can be identified. From a pure physics viewpoint, the reduced Penning effect, observed here is interesting. Although it can likely be shown to result from impurities in the tube preventing a good reference measurement, for pure argon, or from dissociation of the acetylene, it may indicate that at high pressure collisional energy transfer loses significance relative to photoionization and other mechanisms. 59 Chapter 7. Suggestions and Conclusions 60 These investigations are left for the future. For now it has been shown that the use of Penning mixtures and of mists of distilled water or salt solution show no promise for reducing the breakdown voltage in the arc lamp. Parallel starting pulse injection and the use of an auxiliary discharge as a plasma source have similarly shown little potential. The approaches most likely to meet success are to increase circuit voltages, which is discouraged on account of cost and safety requirements, or to consider work on starting without the water wall as discussed above. Bibliography [l] Birks, J.B., and J.H. Schulman, (1959), Progress in Dielectrics, Volume 1. John Wiley and Sons Inc., New York. [2] Cobine J.D., (1958), Gaseous Conductors, Dover Publications Inc., New York. [3] G'.R.C. Handbook of Spectroscopy, (1974), J.W. Robinson, ed., C.R.C. Press, Cleve-land. [4] Falkovskii, N.I., I.V. Bozhko, S.R. Troitskii, and N.I. Glazkov, (1985), Study of the Critical Voltages of a Discharge in Gases with an Easily Ionizablc Additive, High Temperature, V.23, no.2, pp 196-201 [5] Gettel, L . E . , (1980), A Comparative Study of D.C. and A.C. Vortex Stabilized Arcs. Plasma Physics Group, U.B.C., Ph.D. Thesis. [6] Harrold, R.T. , (1986), Physical Aspects of Vapour Mist Dielectrics, I .E .E .E. Trans, on Industry Appl., V.1A-22, no.l, p.63. [7] Herlin M.R., and S.C. Brown, (1948), Phys. Rev., 74, p.291. [8] Heylen, A . E . D . , (1970), Maximization of Argon Hydrocarbon Penning Mixtures,J. Phys. D.: Appl. Phys., V.3, p. 789. [9] Heylen, A . E . D . , (1983), Optimum Argon Hydrocarbon Penning Mixture Ionization Formula, Int. J . Electronics, V.55, no.2, pp. 259-264. 10] Howatson, A . M . , (1965), An Introduction to Gas Discharges, 2nd ed., Pergamon Press, Toronto. Bibliography 62 [11] Hoyaux, M . F . , (1968), Arc Physics, Springer Verlag, New York. [12] Hurst, G.S., T . E . Bortner, and R .E . Glick, (1965), Ionization and Excitation of Argon with Alpha Particles, Journ. of Chem. Phys., V.42, no.2, p.713. [13] Jesse W.P., and J. Sadouskis, (1952), Alpha-particle Ionization in Mixtures of the Noble Gases, Phys. Rev. 88, pp 417-418. [14] Jones, G.R., and M . T . C . Feng, (1980), The Physics of High Power Arcs, Rep. Prog. Phys., V.43, p.1416. [15] Kessel Q .C. , and B. Fastrup, (1973), in Case Studies in Atomic Physics, V.3, North Holland, Amsterdam. [16] Kruithof, A .A . , and F . M . Penning, (1937), Determination of the Townsend Ioniza-tion Coefficient a for Mixtures of Neon and Argon, Physica IV,no.6, p.430. [17] Llewellyn-Jones, F., (1957), Ionization and Breakdown in Gases, Methuen and Co. Ltd., London. [18] Llewellyn-Jones, F. , (1967), Ionization Avalanches and Breakdown, Methuen and Co. Ltd., London. [19] Loeb L .B . , and J .M. Meek, (1941), The Mechanism, of the Electric Spark, Stanford University Press, California. [20] Massey H.S., and E . H . Burhop, (1952), Electronic and Ionic Impact Phenomena, Clarendon Press, Oxford. [21] McAllister D., J.R. Smith, and N.J. Disirens, (1985), Computer Modelling in Elec-trostatics, Research Studies Press Ltd., Letchworth, England. Bibliography 63 [22] Meek, J .M. , and J.D. Craggs, (1978), Electrical Breakdown of Gases, John Wiley and Sons, Chichester. [23] Nakanishi, K., L.G.Christophorou, L . G . Carter, and S.R. Hunter, (1985), Penning Ionization Ternary Gas Mixtures for Diffuse Discharge Switching Applications, J. Appl. Phys., V.58, no.2, p.633. [24] Neilson, J.B., (1981), An Investigation of a Vortex Stabilized Arc, Plasma Physics Group, U . B . C , Ph.D. Thesis. [25] Nieuwland J.A., and R.R. Vogt, (1945), The Chemistry of Acetylene, Reinhold Pub-lishing Corporation, New York. [26] Papoular, R., (1965), Electrical Phenomena in Gases, Iliffe Books Ltd., London. [27] Pearson, J.B., (1985), Aspects of Energy Transport in a Vortex Stabilized. Arc, Plasma Physics Group, U . B . C , Ph.D. Thesis. [28] Reinking, G.F., L . G . Christophorou and S.R. Hunter, (1986), Studies of Total Ion-ization in Gases/Mixtures of Interest to Pulsed Power Applications., J. Appl. Phys., V.60, no.2, p.499 [29] Reitz J.R.. F.J. Milford, and R.W. Christy, (1979), Foundations of Electromagnetic Theory, 3rd ed., Addison-Wesley, Massachusetts. [30] Rogowski W., (1928), Arch. Elektrotech., 20,p.99. [31] Rudd M . E . , and J.H. Macek, (1973), in Case Studies in Atomic Physics, V.3, p.47, North-Holland, Amsterdam. [32] Sakuntala, M . , (1965), Discharges in Potassium Seeded Argon at Elevated Temper-atures, Br. J. Appl. Phys. 16 pp. 821-832. Bibliography 64 [33] Segerlind, L . G . , (1984), Applied Finite Element Analysis, 2nd edition, Wiley, New York. [34] Silvester, P.P., and R.L. Ferrari, (1983), Finite Elements for Electrical Engineers, Cambridge University Press, Cambridge, England. [35] Starr, W.T. , (1956), Corona Properties of Insulating Materials, Electrical Manufac-turing, June 1956, p. 132. [36] Trump J .G. , F .J . Safford, and R.W. Cloud, (1941), Trans. Amer. Inst. Elect. Engrs., 60, p.132. [37] Von Engel A. , (1965), Ionized Gases, Clarendon Press, Oxford. Appendix A Early Designs A brief description of advantages and disadvantages of early tube designs is included to avoid the repetition of any mistakes. A . l The Rotating Tube This design was intended to create a water wall by spinning the tube and hence creating a water cylinder by friction and centrifugal force (Fig. A.31). The shaft, seal is used so that the inner tube can rotate while the outer tube remains fixed, allowing the pressure gauge to remain stationary. The pressure measurement then requires the same pressure in both tubes. The rotation is caused by an electric motor driving an o ring belt and pulley, which makes speed adjustment, possible through alteration of pulley sizes. The belt isolates the motor from the high voltage on the electrodes, with electrical contact made through the bearings. This design presented many drawbacks when it was first implemented. The water wall tended to be uneven due to tube vibration, partly due to a poor point bearing and lack of rigidity in the structure. The bearings also made poor electrical contact at times, and the balls corroded when appreciable current was passed. The equal pressure inside and outside the central tube meant that breakdown was likely in both regions, and tracking problems resulted. All these difficulties could have been overcome, but it was decided that a device using a water jet would be simpler to use. 65 Appendix A. Early Designs 66 drive pulley r o j ] e r bearing electrical contact point bearing electrical contact gas/vacuum connection he T 1 \ a* r fi 1 CQ n p-i_ its III J If* n _/ -4, tr tfe! -TSJ 1 rotating shaft seal Garlock Pk Mechanical Seal 70066 0003 Figure A.31: Rotating Tube Cross Section Appendix A. Early Designs 67 J centrifugal pump jet assembly tube r e s e r v o i r drain Figure A.32: Water Jet Schematic A.2 The Water Jet The water jet system consists of a centrifugal pump pumping distilled water from a reservoir to a chamber with four small holes at the top of a vertical tube. The holes provide a tangential velocity and some azimuthal velocity which is aided by gravity. The water is returned to the reservoir through a drain at the bottom of the tube. There is a tube with a valve connecting the airspace in the reservoir with the gas in the gap in order to keep the pressures the same. The valve allows some control when airlocks cause the tube to fill with water. Appendix A. Early Designs 68 This system is an improvement as there are no moving parts except the pump rotor. The disadvantages include the fact that disassembly can dump a lot of water, and that it is still more complicated than is really necessary. A pool of water along the bottom of a horizontal tube was found to have the same effect on breakdown as the moving wall. The tube constructed for this is the one used for the experiments presented in this thesis. Appendix B Starting Transformer TRIGGER S W I T C H 115 V A C TR, Cr; - r - r S.G. 500 pF, 30 kV Cc 0.005 MFD, ikV L] R F INDUCTOR SG SPARK GAP 60Hz T R A N S F O R M E R T R 2 R F T R A N S F O R M E R 69 1' Appendix C Finite Element Program C l The Finite Element Method T h e f ini te element m e t h o d is o u t l i n e d here as i t was a p p l i e d to t h i s problem. T h a t is, for an a x i s y m m e t r i c e lec t ros ta t ic p r o b l e m us ing l inear three node t r i angu la r e lements . 1 T h e e q u a t i o n of interest is Po i s son ' s equa t ion ; Y-(eV</>) = -p or i f we assume no space charge, Lap lace ' s equa t ion . For a n a x i s y m m e t r i c p r o b l e m , we are interested i n c y l i n d r i c a l geometry, w i t h no 6 dependence . Because e w i l l d e p e n d on the densi ty of t he gas, i t w i l l vary rad ia l ly . E v e n for pressures of several a tmospheres , it is ve ry close to e c , so the r a d i a l dependence was ignored . T h e p r o b l e m to be solved is then ; r dr dr dz2 e w i t h b o u n d a r y cond i t ions of b o t h D i r i c h l e t t ype , <b = f(r, z) on electrodes, and N e u m a n n type , | ^ = g(r, z) o n d ie lec t r ic boundar ies and the axis o f ro ta t ion . T h i s is the d ie lec t r ic b o u n d a r y c o n d i t i o n on the d i sp lacement f ie ld , a ' F o r a general discussion see [33, 34] 70 Appendix C. Finite Element Program 71 where u is any free surface charge o n the d ie lec t r ic . Fo r th is p r o b l e m o n l y p o l a r i z a t i o n charge is cons ide red , so 6$ Sn = 0 . T o comple t e t he f o r m u l a t i o n of the p r o b l e m , we use the energy w i t h i n a n element; W — J ^(eE2 + p<t>)dV + ^ j <7<}>dS A we igh ted r e s idua l m e t h o d c o u l d also be used [33]. N o w to o b t a i n the p o t e n t i a l d i s t r i b u t i o n we requi re tha t i t m i n i m i z e the s tored field energy. It mus t therefore sat isfy dW = 0 T o cons t ruc t the a p p r o x i m a t e s o l u t i o n , the p r o b l e m reg ion is broken i n t o t r i angu la r e lements . W i t h i n each e lement it is assumed tha t the p o t e n t i a l is a p p r o x i m a t e d by (f> = a + bx - f cy w i t h a,b, a n d c de t e rmined by the requi rement t ha t the p o t e n t i a l assume the vertex values at the vert ices . a b c T h e p o t e n t i a l is then g iven by 1 yi = 1 V2 <f>3 1 X3 2/3 <f>(x,y) = 1 x y 1 xl yi - i " 4>i ' 1 y2 <j>2 1 X3 ys . ^ 3 . or 1=1 Appendix C. Finite Element Program 72 where the a , are l inear funct ions of pos i t ion on ly . T h e energy i n t eg ra l , neglect ing any space or surface charge, is then W = ^ I \V4>\2dS 3 3 Def in ing i=l j = l Sij = J Va, • Vcv. dS W = ^<f>TScp T h e m a t r i x S is eas i ly eva lua ted for a t r i a n g u l a r e lement [34]. T h e element matr ices c a n then be superposed to generate a m a t r i x represen t ing the sys tem of equat ions for al l nodes i n the p r o b l e m . T h e p r o b l e m to be solved can now be w r i t t e n dW 34>k %»,]a^  * V sfp . Srf Spp where f a n d p denote free and presc r ibed po ten t ia l s . O n l y di f ferent ia t ion w i t h respect to free po ten t i a l s con t r ibu tes , so Sff Sfr 4>f o T h i s a l lows the p o t e n t i a l so lu t ion to be w r i t t e n — SjjSfpCpp cp = as suming nons ingu la r mat r ices . T h i s is the a p p r o x i m a t e s o l u t i o n as a set o f n o d a l values, bu t the so lu t i o n is a c t u a l l y k n o w n everwhere. Appendix C. Finite Element Program 73 T h e r e was one t r i ck used i n the n u m e r i c a l c a l c u l a t i o n . N o t i c i n g tha t the m a t r i x S is s y m m e t r i c and b a n d e d meant t h a t o n l y the u p p e r ha l f b a n d w i d t h needed to be s tored. T h i s r educed the storage i n w h a t is a l ready a f a i r l y efficient, and accura te m e t h o d . P lo t s presented i n th is w o r k ( F i g . 5.22 and F i g . 5.23) were generated us ing a s l igh t ly mod i f i ed vers ion of the con tour a p p l i c a t i o n p r o g r a m genera ted by I S S C O ' s D I S S P L A C o d e b o o k . 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items