Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Switching characteristics of electrodeless gas-filled bulbs immersed in both A.C. and unipolar, uniform… Richardson, Michael J. 1992

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

Item Metadata


831-ubc_1992_fall_richardson_michael_j.pdf [ 2MB ]
JSON: 831-1.0085434.json
JSON-LD: 831-1.0085434-ld.json
RDF/XML (Pretty): 831-1.0085434-rdf.xml
RDF/JSON: 831-1.0085434-rdf.json
Turtle: 831-1.0085434-turtle.txt
N-Triples: 831-1.0085434-rdf-ntriples.txt
Original Record: 831-1.0085434-source.json
Full Text

Full Text

SWITCHING CHARACTERISTICS OF ELECTRODELESS GAS-FILLED BULBS IMMERSED IN BOTH A.C. AND UNIPOLAR, UNIFORM ELECTRIC FIELDS By Michael J. Richardson B.Sc. University of British Columbia, 1990  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE  in THE FACULTY OF GRADUATE STUDIES PHYSICS  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA June, 1992  © Michael J. Richardson, 1992  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 department  or  by  his  or  her  representatives.  It  is  understood  that  copying  my or  publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of  Physics  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  July 2/1992  ii  Abstract A more complete understanding of the switching characteristics of external electrode gas filled discharge bulbs is needed for many electrical devices (e.g. AC plasma display panels).  When a glass (or pyrex) bulb filled with gas (e.g.  neon) is placed in an electric field, it can emit light in the form of a pulse.  Whether or not a discharge (light pulse) occurs  depends on the magnitude of the electric field in the bulb and the various bulb parameters (geometry, pressure, gas, etc.).  The  field inside the bulb is made up of the field created by the external electrodes, and the field created by any wall charges deposited inside the discharge vessel.  The bulbs used in this  thesis were of two varieties; spherical bulbs (1 Torr of Ne), and spherical Ne-Xe (5 Torr) bulbs with getters attached to one end spoiling their overall symmetry.  The spherical, ungettered bulbs  have an isotropic response to the applied field, but possess long time lags (up to 2 seconds) before breakdown occurs.  The  duration of the time lag can be reduced by increasing the frequency of the applied field.  These bulbs also experience an  aging process which causes the AC threshold field (the minimum AC field required for breakdown) to increase over time so long as the bulbs are not exposed to any discharges.  The gettered bulbs  do not possess this aging effect but are unique in that breakdown in an AC field only occurs when electrons and ions are accelerated in a specific direction.  Furthermore, the gettered  iii bulbs only exhibit long time lags if they are initially exposed to a low field (too low to induce breakdown) for at least a few seconds.  Without the occurrence of time lags, the effects of  wall charges (created from previous discharges) can be studied. This thesis presents the theoretical and experimental results which explain these characteristics for both bulb types. The geometry of the gettered bulbs and the relative orientation of the applied field is described along with the necessary apparatus for producing the various electric fields. Finally, the characteristics of the two types of bulbs are compared. Suggestions for improving their switching performance are also advanced.  iv  Table of Contents  Abstract  ii  List of Tables  vii  List of Figures  viii  Acknowledgements  xi  1. Introduction  l  2. Electrodeless Breakdown: General  3  3. Production of Discharges  6  3 . 1 The Breakdown Mechanism  6  3.2  P a s c h e n ' s S c a l i n g Law  10  3.3 S t a r t i n g P o t e n t i a l s 3 . 3 . 1 S t a r t i n g P o t e n t i a l of G e t t e r e d B u l b s  14 . . . .  15  3 . 4 Breakdown i n a n I n s u l a t i n g V e s s e l  19  3 . 5 S t r o n g a n d Weak Breakdown  22  3 . 5 . 1 S t r o n g Breakdown  22  3 . 5 . 2 Weak Breakdown  24  V  4.  3.6 Wall Charge Leakage  26  3.7 Time Lags  28  3.7.1 Statistical Time Lag  29  3.7.2  30  Formative Time Lag  3.8 Ionization in Gases  32  3.9 The Penning Effect  34  Experimental Results  35  4.1 Introduction  35  4.2 Experimental Apparatus  37  4.2.1 Introduction  37  4.2.2 Instrument Used to Measure the Light Pulse Emission  37  4.2.3 Apparatus for Generating Uniform Timevarying Fields 4.3 The Bulbs  40 45  4.3.1 Gettered Bulbs  45  4.3.2 Ungettered Bulbs  47  4.4 Electrical Breakdowns in an A.C. Field 4.4.1 Ungettered Bulbs  48 48  4.4.2 Time Constants Involved In Breakdown Formation for the Ungettered Bulbs  59  4.4.3 Resetting a Dormant Bulb  62  4.5 Breakdown In Unipolar Electric Fields  68  4.5.1 Introduction  68  4.5.2 Apparatus Used to Produce the Unipolar Electric Field Pulses  69  4.6 Behaviour of the Ungettered Bulbs in the Unipolar Fields  75  4.7 The Probability of Breakdown Versus the Repetition Rate of Unipolar Pulsed Electric Fields 4.8 Gettered Bulbs in an A.C. Field  78 84  4.8.1 Bulb Conductivity and its Effect on Breakdown  85  4.8.2 Breakdown in the Gettered Bulbs  86  4.8.3 Unipolar Fields and Gettered Bulbs  90  4.8.4 Wall Charge Freezing  91  4.8.5 Gettered Bulb Time Lags  98  4.9 Evolution of Discharges in an A.C. Field for Gettered Bulbs  103  5. Summary and Conclusion  108  Appendix A.l  114  Ionization Coefficients for Electrons  114  vii  List of Tables Table 1 Breakdown parameters of different gases  13  Table 2 Breakdown parameters for various materials  13  Table 3 Time and reset constants for the ungettered bulbs.  . 66  Table 4 Relative time lags of getter tube and bulb discharges with varying unipolar voltage pulses applied  99  Table 5 Values of the starting potential coefficients, A and B, for various gases  116  viii  List of Figures Figure 1.1 Paschen curve - breakdown potential (minimum threshold value) as a function of pd  12  Figure 1.2 The typical gettered bulb and its various elements  15  Figure 1.3 Evolution of the various fields to which the electrodeless bulb is subjected  21  Figure 4.1 Apparatus used for detecting optical pulses. . . . 39 Figure 4.2 Parallel plates used with gettered bulbs Figure 4.3 Parallel plates used with ungettered bulbs.  42 . . . 42  Figure 4.4 Device used to produce the AC fields  44  Figure 4.5 Types of bulbs used; a) gettered and b) ungettered  46  Figure 4.6 Typical oscilloscope trace showing the light pulses (breakdowns) superimposed on the applied field. . 48 Figure 4.7 Applied field versus number of breakdowns/halfcycle  50  Figure 4.8 Threshold field versus total time that the bulb remains dormant  53  Figure 4.9 Threshold field versus dormancy time for a bulb with a higher oxygen content  53  Figure 4.10 Graph used to determine the time constant of the increasing threshold field for the clean bulb Figure 4.11 Graph used to determine the threshold field time  56  ix constant for the dirty bulb  56  Figure 4.12 Device used to count the number of breakdowns.  .  63  Figure 4.13 Resetting the threshold field for the clean ungettered bulb with breakdown exposure  65  Figure 4.14 Resetting the dirty ungettered bulb with breakdown exposure  65  Figure 4.15 Graph used to determine the reset constant k for the clean ungettered bulb  67  Figure 4.16 Graph used to determine the dirty ungettered bulb's reset constant k  67  Figure 4.17 Device used to produce a three-sided, unipolar voltage pulse with a flat top  70  Figure 4.18 Device used to trigger the spark gaps to initialize their firing  73  Figure 4.19 Device used to produce the voltage spikes or ramps shown as output  74  Figure 4.20 Device used to produce positive and negative voltage steps across the plates Figure 4.21 The dimensions of a typical voltage spike.  74 ...  78  Figure 4.22 The probability of a breakdown occurring on each electric field pulse versus the pulse frequency  80  Figure 4.23 Typical series of voltage spikes applied to the plates at a high frequency (14 Hz) along with the photomultiplier output  81  Figure 4.24 Device used to count the number of electric field spikes in a given interval  83  X  Figure 4.25 The positions that the optic fibre must be in to enable the recording of breakdowns in the tube and bulb independently  87  Figure 4.2 6 The occurrences of breakdowns in the bulb relative to the applied AC field  89  Figure 4.27 Device used for producing low voltage (up to 2 kV) three-sided unipolar voltage pulses  93  Figure 4.28 Dimensions of the unipolar pulse created by the krytron circuit  94  Figure 4.2 9 Typical oscilloscope trace showing the breakdown time delay relative to the pulse  94  Figure 4.3 0 Graph which shows how the wall charge freezing depends on the magnitude of the voltage pulse  96  Figure 4.31 Oscilloscope trace showing the relative time lags for gettered breakdown  98  Figure 5.1 Typical oscilloscope trace showing the time lag as defined relative to the step pulse  100  Figure 5.2 Initial "freezing" voltage versus formative time lag  102  Figure 5.3 Pictorial evolution of gettered bulb breakdown in an AC field  104  xi  Acknowledgements  The author is indebted to F.L Curzon for his help in the development of the physical models and assistance with the planning and execution of the experiments used in this thesis. Gratitude is also extended to A. Cheuk and J. Bosma for their contributions in developing the experimental apparatus. The author is thankful to Z. Shen for providing the initial stimulus for a series of experiments which involved the "aging" of ungettered, electrodeless, discharge bulbs, and to B. Ahlborn and M. Lefrancois for generously offering any needed advice and use of their computer facilities. Finally, the continuous encouragement and support received from my family (and Marni Bell) will always be appreciated.  1  2. Introduction Electrical discharges occurring in a gas are often maintained by electric fields that are applied to electrodes inside the vessel holding the gas.  This type of discharge is  known as an Internal Electric Discharge.  Fluorescent lights and  neon signs are among the more common examples of such discharges. Another less common type of discharge occurs when the electric fields are applied to electrodes which are outside the discharge vessel holding the gas.  This type of discharge is appropriately  referred to as an External Electric Discharge (or EED). Since the discharge tubes themselves are insulators, and the electrodes are external, they can become electrically charged. We refer to these internal charges as "wall charges". The presence of these wall charges can dramatically modify the breakdown conditions of the gas inside the bulbs.  Good  insulators can remain charged for time periods much longer than the lifetime of the excited species of the discharge gas.  It is  therefore possible that the wall charges may influence the breakdown characteristics of repetitively pulsed discharges (a property generally not considered with internal electrodes since the charges are removed by the internal electrodes). Common examples where these EED characteristics occur are: 1)  AC plasma display panels consisting of many, tiny,  closely packed glass discharge tubes arranged in an array making up the screen.  The gas discharge television system makes use of  2 the intense emissions of ultraviolet light that occurs when short duration voltage pulses are applied to the external electrodes across the Xenon-filled tubes.  The ultraviolet light is then  absorbed by the appropriate phosphors to create the various coloured pixels of the television image. 2)  Instruments which can measure the electric fields at  frequencies of 50-60 Hz.  These meters are based on breakdown of  gases inside dielectric envelopes (bulbs).  Glass bulbs filled  with Neon emit pulses of light whenever they are exposed to electric fields of high enough magnitude to cause gaseous breakdown within.  The number of light pulses emitted per second  is proportional to the magnitude of the electric field.  Meters  such as these have several desirable characteristics for their operators.  They are readily portable.  The sensors are non-  metallic (greatly reducing the chances that flashover to the operator might occur), and, if the bulbs are spherical, they can be placed anywhere in the field without affecting the output (i.e. orientation of the sensor with respect to the field does not make any difference to the readings on the meters). 3)  Small gas bubbles which get trapped in the insulation of  conducting cables during the manufacturing process.  Since these  bubbles are subject to electric fields, discharges can occur within them.  Over time many discharges can harden and crack the  insulation creating a pathway for an arc to travel from the conductor.  This is ultimately responsible for failure in many  high voltage cables.  3  2. Electrodeless  Breakdown:  General  It has been known for about eighty years that electric fields can make a gas conducting. experiments were made bya Tesla1, Thomson3.  Some of the earliest Wiedemann  and Ebert2,  and  The starting (or maintenance) potential was measured  as a function of the nature and pressure of the gas used, the frequency of the applied field, and the separation of the electrodes.  Another important factor in determining the  behaviour of the "switching" characteristics of the gas in the field is the geometry of the apparatus (electrodes and vessel). Experiments have been performed with the electrodes positioned inside the discharge vessel (such as in fluorescent lighting and Neon signs), discharge).  as well as outside (the so-called electrodeless The effect of discharge vessel shape on the  discharge characteristics have also been studied with designs ranging from spherical bulbs to cylindrical tubes containing multiple anodes4 (commonly found in plasma display panels).  The  results obtained for this thesis concentrate on the properties of the electrodeless discharges that occur within spherical glass bulbs. The physics of the electrodeless discharge is very different from that of the internal electric discharge primarily because the electrodes and the gas can no longer complete the electrical a  . These references were cited in Francis, Phenomena in Gases", p 95 (see references)  G.,  "Ionization  4 circuit.  Instead, the discharge vessel walls (which interrupt  the current flow) play a major role.  The electrodeless discharge  can also take different forms depending on the frequency of the applied field5.  At the lower frequencies (less than 100 Hz),  the gas in the bulb becomes an insulator between successive discharges and wall charges play a dominant role. As the frequency is further increased, products of the discharge in one half cycle remain in the gas to assist the subsequent discharges, thereby decreasing the starting field (the field below which the gas will not breakdown).  These products may be resonance photons  or metastable atoms, as in the rare gases, or ions.  At still  higher frequencies the electrons are swept even a shorter distance by the field during each half cycle.  When their  oscillation amplitude becomes less than the length of the vessel, energy losses at the walls are greatly reduced and the minimum breakdown field decreases again (the vessel walls no longer play a dominant role).  Eventually, as very high frequencies are  reached, the electrons no longer collide with the gas molecules as often and the discharge threshold increases rapidly. The higher frequency regime has been studied mainly because of its application to plasma display panels.  The low frequency  regime, however, received only a little attention in the 1950's then quickly vanished from most of the popular plasma texts. Harries  and von Engel6  7 8  did most of their work on  electrodeless breakdown at 50 Hz; their motivation arose from a need to understand the deterioration of insulation used in  5 electrical equipment. For example, the processes involved in forming the polyethylene around electrical lines creates small gas pockets which support electrodeless breakdown, eventually leading to the destruction of the insulation.  They studied the  breakdown mechanism by observing current pulses produced with each discharge.  This work led to a basic explanation of why the  current pulses occur.  However, their work was only done with  uniform single phase sinusoidal fields.  6  3. Production 3.1  The Breakdown  of  Discharges  Mechanism  At low field frequencies the mechanism for breakdown is the same as it is in DC.  It is the Townsend discharge that governs  this breakdown mechanism.  This type of discharge has been  studied extensively because of the important role it plays in the formation of arcs between metal electrodes.  Townsend formulated  the following theory9. Consider N electrons propagating through a gas in a vessel which is exposed to an electric field.  Each of these N electrons  will be accelerated by the field and will produce on average m new electrons plus m positive ions.  The m positive ions will  travel more slowly (due to their greater mass) back towards the cathode side of the vessel.  When these m positive ions (plus the  associated particles in the form of metastable particles or quanta in the form of photons) release just one secondary electron when they bombard the interior of the vessel, the process is considered self-perpetuating thus sustaining the discharge.  If y is the number of secondary electrons generated  at the cathode per ion created in the discharge vessel by electron-atom collisions, then the self-sustaining condition sought is  7 i77Y=l  (3.1)  where m is the ion multiplication factor.  The increase in the  number of free electrons dNe along dx in the direction of the field is proportional to the number of electrons Ne at x and an electron ionization coefficient a  (ion pairs produced per  electron over a unit distance). dN={aN)dx Thus the number of new electrons (Ne)  that are created after a  distance d is given by Ng=N0(ead-l) where N0 is the initial number of electrons in the gas at location x=0  (due to ionization by cosmic rays, radioactive  impurities, or background radiation).  Note that if we assume an  avalanche starts with one electron, the first electron, unlike the rest, is not accompanied by a new ion.  The number of new  electrons is equivalent to the number of new ions.  The number of  secondary electrons is simply given by Y(ead-1) If the process is continuous (ie. the condition for selfsustenance given by equation 1.1 is met) then the number of electrons after many avalanches will be l+Y(eod-l) +v2(ead-l)2+. . .=  i— (l-Y(eod-l))  If the denominator of the expression on the right vanishes, then complete breakdown is obtained and all the gas is ionized. Hence for complete breakdown we require <3-2>  <xd=ln(l+l/Y)  This expression is commonly known as the Townsend breakdown criterion.  It is valid for uniform fields and constant Y  at  medium and low pressures provided that p is large enough to warrant the use of a (i.e. ad >> 1 and X << d where X is the mean free path for electron-atom ionizing collisions).  Several  important observations can be made10: 1.  a depends on the amount of energy that the electron  acquires during a mean free path.  If the electron's energy  is below the ionization energy of the atom with which it collides, then the total number of electrons will not increase (no new electrons will be emitted into the gas). Hence below a certain electric field, the value of a will tend to zero and the Townsend criterion will not be satisfied.  This field at which the Townsend criterion is  just satisfied will be termed the threshold field ET for avalanche growth.b 2.  the Townsend criterion dependence on y is logarithmic  and therefore weak.  Its value is primarily governed by the  properties of the material at the site of the secondary  b  see Appendix  A.l  9 emission (ie. the vessel walls). 3.  The value of a is usually the most important parameter  for controlling  ET.  10  3.2  Paschen's  Scaling  Law  We can assume that if the secondary emitter surface for a given discharge vessel is uniform and constant, y is also constant.  Thus ad is also constant.  Furthermore, since a is a  function f  of the energy gained per mean free path EjX  (where A.  is the mean free path of particles in the gas), we write a=  — A  Since ad is a constant, it follows that cons tan t= But, X=constant/p  (where p denotes the gas pressure), therefore Vgg^Ejd^  where V^ function.  ^— d  (pd)  (3.3)  is the threshold voltage and <|) is just another This result is known as the Paschen Scaling Law.  examining equation (3.3) we see that if pd  Upon  is held constant, we  can lower the threshold field ET by increasing the dimension d of the vessel. constant  In other words ET varies inversely with d for  pd.  One can deduce the shape of the function <j) for a single uniform gas from basic physics.  At very high values of pd  number of molecules creates many collisions.  the  Since the mean free  11 path between collisions is greatly reduced, the electrons need a greater threshold field ET in order to obtain enough energy to ionize.  On the other hand, at the lower pd values there are so  few molecules that a large field is again needed to start breakdown.  Thus the function <J) must have a minimum somewhere in  between these two extremes.  In fact, the graph of V ^ versus pd  resembles an asymmetric parabola where the minimum is typically broad as sketched in Figure 1.1 . The following tables11 show typical values for the parameters involved in equations (3.3) and (3.2) for a few gases. Inert gases are more desirable because of their long, stable lifetimes and low chemical reactivity.  Neon and Argon are  commonly used because of their low V™ values.  12  Paschen minimum.  _,  ,  !  ,  p.  T  I  Figure 1.1 Paschen curve - breakdown potential (minimum threshold value) as a function of pd.  13  gas  Vgg (volts)  pd minimum (Torr mm)  neon  23 0  4  argon  22 0  1.5  air  350  6.0  Table 1 Breakdown parameters of different gases.  material  Y  pyrex  very low (less than 0.001)  metal electrode  0.01  photomultiplier coating  0.3  Table 2 Breakdown parameters for various materials.  14  3.3 Starting  Potentials  For Electrical  Breakdown  For ordinary gases at low and medium pressures, breakdown will occur such that the multiplication of electrons produced during avalanches goes to infinity and external energy sources are no longer necessary for initial electron production.  The  minimum field required for this process is termed the threshold field ET.  This value can be found experimentally by slowly  increasing the potential difference until breakdown occurs consistently with each change in field. required to cause breakdown is denoted  The potential difference V^.  To find an expression for the breakdown field in terms of the gas, its pressure, and the vessel dimensions (d), we utilize equation (3.2) .  If instead we write (a/p) (pd)  for ad and  substitute equation A.(1.1) (from appendix A.l) for a/p, obtain for the starting field12 ET_ B p C+ln {pd) with  we  15 or B[pd)  „ _ m  C+ln(pd)  V^j, as a function of (pd)  (3.4)  traces out the familiar Paschen  curve as depicted in Figure 1.1  3.3.1  Starting  Potential  of Gettered  Bulbs  Figure 1.2 The typical gettered bulb and its various elements. A gettered bulb consists of a spherical bulb connected by a long tube to a receptacle which contains a metal plate or wire (Figure 1.2 shows the various components of this type of bulb).  16 This wire has a high affinity for oxygen when it is heated and ensures that the oxygen content of the gas in the spherical bulb is very low.  Since the inner surface geometry of the gettered  bulbs is not symmetric for both field directions when the getter tube is parallel to the field, we can no longer assume that y (secondary emission coefficient) is constant. this variability of Y  to  We would expect  affect the breakdown potentials V^  opposite field directions.  for  It is the hole in the bulb at the  junction with the getter tube which decreases the chance of secondary emission from the vessel walls in this direction. Since Y would be higher for field directions which can accelerate electrons towards the getter tube we expect V ^ to be reduced in this direction. First we assume that there must be a component of the bulbs inner surface which is perpendicular to the direction of the field.  Hence the entire lower hemisphere of the bulb opposite to  the getter can be a source of secondary electrons.  If the field  direction is such that these secondary emissions can be accelerated towards the getter tube, then they can effectively assist in electrical breakdown of the gas.  We will refer to the  lower hemisphere's secondary emission coefficient as yx.  On the  other hand, the hemisphere of the bulb which contains the getter tube has a surface area which is much less than the lower hemisphere due to the hole leading to the getter.  Thus the  secondary emission coefficient, which we will denote as y2, will  17 be greatly reduced for this half of the bulb. Y will increase the value of V^  This decrease in  for field directions which  accelerate electrons from the getter side of the bulb towards the base.  Since the hole in the upper hemisphere constitutes about a  third of the upper surface area we can approximate the coefficients as Y^lY!  <3.5)  If VKB1 is the starting potential for field directions which accelerate electrons from the base of the bulb towards the getter and Vmz  is the starting potential for oppositely directed fields  then equation (3.4) states vmi_  C2+ln(pd) q + ln(pd)  VKB2  (3.6)  where, C22 =ln[. . " . ,) ln(l+l/Y2) and  q=ln[. 1  * . .]  lnd+l/Yi)  In equation (3.6) we have assumed that the gas pressure and content remains constant.  Hence Ax (see appendix A.l), p, and  18 the ionization potential V± are all constant for this analysis. We also assume that the vessel dimension d (where d is the diameter of the bulb plus the length of the tube in the field) is the same for both field directions so long as the field remains parallel to the tube.  Since A=l/X1  and B=Vi/k1  it follows that  these parameters are also held constant. For ordinary gases A is typically of the order of 10 {cm Torr)'1  (see table 5 in appendix A.l) .  Typical values of y  for bulbs made of a pyrex insulator are around 10"3 . pyrex gettered bulb of pressure p=5 Torr  Thus for a  and dimension d=7 cm  equation (3.6) reduces to (upon substituting in equation (3.5)) KBI - _ g 7 7KB2  or VKB2 is about 97 % of VKB1 for a typical pyrex gettered bulb with a decrease in y of a third in a given field direction. We find (experimentally) V  1370  Vm  1400 V  KBI_  V„  9  8  i.e. experimentally we find good agreement with the above prediction.  The experiments are described in section 4.8.3.  19  3.4 Breakdown  in an Insulating  Vessel  In a normal interior electrode discharge vessel, breakdown starts according to Townsend's theory and continues as a steady discharge with the ionized gas (plasma) completing the electrical circuit that exists outside the vessel.  The discharge will  continue provided that the DC voltage across the gap of the electrodes remains high enough.  However, in electrodeless  discharges the breakdown is interrupted by the insulating vessel walls.  Thus, when exposed to a continuous DC field, the gas  breaks down, charges created separate, and an internal field which can cancel out the applied field is set up.  This internal  field serves to reduce the net field inside the bulb and the discharge is quickly extinguished.  Instead of a continuous  steady discharge, only a single pulse discharge occurs. At any particular point in time, the net electric field ET inside the bulb is made up of two components13, EA and EB.  EA  is the electric field that is applied externally to the vessel via the electrodes, while EB is the field created by the separation of charges to the interior surface of the vessel walls.  The total amount of charges separated, and hence the  magnitude of EB, depends on the influence of ET on the plasmas which are produced with an electrical breakdown of the gas. These charges separate in such a way as to cancel out the  20 externally applied field EA.  Therefore EB is antiparallel to EA  immediately after the time of breakdown (see Figure 1.3 (b) ) . It follows then that EX=EA+EB  (3.7)  If E0 is the breakdown field strength of the gas inside the bulb (slightly lower than ET if a discharge has occurred within the last .05 seconds which can seed a subsequent discharge), then whenever \ET\  (the magnitude of ET)  reaches E0 in linearly  polarized fields, breakdown of the gas will occur.  If this  condition is satisfied, breakdown will evolve in accordance to the conventional Townsend theory previously outlined.  The  evolution of EB and Ex in the presence of a sinusoidal (alternating) applied field EA is shown in Figure 1.3 c.  The  dotted line represents the field created by the bulb's wall charges EB,  and the saw-toothed wave shows how Ex gets reset to  zero after each breakdown.  c  Graph obtained from Friedmann, D., Frequency Electric Fields Using Electrodeless p 21 (see references)  "Measurement of Low Breakdown of Gases",  21  Figure 1.3 Evolution of the various electrodeless bulb is subjected.  fields  to  which  the  22  3.5 Strong  and Weak  Breakdown  As already described, the charges produced at breakdown separate and set up the internal field EB which cancels the applied field. E  A  <  Moreover, since EB is produced by the effects of  \EB\ ^ S l e s s than or equal to \EA\ .  The separated charges  are held in place by polarization forces at the inner surface of the vessel shell between breakdowns. these charges as being immobile. subsequent breakdowns.  For now we will consider  Thus EB can only be changed by  The additional charges arriving at the  vessel surface after breakdown can either neutralize those charges already there, or increase their concentration. It is convenient to distinguish between two types of breakdown that can occur during these external electrode discharges.  Depending on the degree of ionization following  breakdown, either a strong breakdown or a weak breakdown will be achieved.  3.5.1  Strong  Breakdown  A strong breakdown will be achieved if the conductivity of the partially ionized gas is large enough such that the internal field Ez is rapidly brought to zero following breakdown.  23 That is ET=EA+EB=0  (3.8)  Recall that the evolution of EB (space charge field in the bulb) for strong breakdown is the stepped waveform that is shown below the time axis in Figure 1.3  (a). The applied field, EA,  in this  case has a sinusoidal waveform and the vertical lines represent the pulsed optical emission which accompanies breakdown of the The value of EB is constant in between optical pulses in  gas.  accordance with the assumed immobility of the charges which have adhered to the vessel's inner surface.  As already mentioned, Ez  is shown as the saw-toothed waveform that gets reset to zero every time it reaches a magnitude of E0 causing breakdown. Changes in ET between breakdowns correspond to changes in EA (since we have assumed EB to be constant during these times). If E'A denotes the largest value of EA, breakdown, fB,  then the frequency of  will be given by the expressiond fB=2[2EA/E0]fA  where fA  is the applied field frequency, EA is the magnitude of  the applied field, and the square brackets requires that  d  (3.9)  refer to Electric Fields references)  Friedmann, D., Using Electrodeless  "Measurement of Low Breakdown of Gases",  2EA/E0  Frequency p 23 (see  24 be rounded down to the nearest integer.  If we denote the average  number of pulses occurring per cycle of the applied field by n then n=2[2Ei/E0]  (3.10)  A step in the magnitude of EB will occur each time E'A increases by E0/2 • One would expect large amplitude optical pulses to accompany strong breakdowns along with little variation in pulse amplitude for successive discharges since conditions inside the bulb will be the same for the production of each subsequent breakdown (pulse).  3.5.2  Weak  Breakdown  If the degree of ionization following breakdown is too low, the total separation of ions and electrons will produce a space charge field EB which is weaker than EA. internal field, Ez, direction as EA.  Hence a residual  will remain in the vessel in the same Since Ez is not reset to zero by a breakdown,  the change in EA required to make \EX\ equal to the breakdown field will be smaller than the value of E0 for  strong breakdown.  Due to the statistical fluctuation in Ex from one breakdown to the next, one expects that the step function for fB versus E'A  25 will be smoothed oute.  That is, fB will now be approximately  linearly related to E'A . Since the change in ER between pulses is now smaller, one expects weaker optical pulses to occur.  e  refer to figure 3 of Friedmann, D., Frequency Electric Fields Using Electrodeless (see references)  "Measurement of Low Breakdown of Gases",  26  3.6  Wall  Charge  Leakage  So far we have seen that a key feature of EED's is the role played by the wall charges which line the interior surfaces of the discharge vessel.  Even in discharge devices which support  internal electrodes, we find that these wall charges also play a significant role.  Thus a wide range of gaseous electronic  devices can benefit from EED studies of wall charge behaviour. So far we have assumed that the wall charges are immobile and are held tightly on the vessel walls between breakdowns. Hence we have assumed that these charges do not leak away and that any change in Ez is the same as the change in EA. Suppose now that wall charges leak away between successive discharges14.  If t1 is the time of one discharge occurrence,  and t2 is the time of the next one, then EB at time t2 will be smaller than it would be with a perfect insulator which supports no wall charge leakage.  Thus one would have to wait a little  longer for EA to increase enough to compensate for the damping of the wall charge field occurring between times t± and t2 . Since the field inside the bulb depends on the charge carrier motion on or in the bulb material, it is clear that field changes depend on the bulb geometry as well as its orientation with respect to the applied field.  The diffusion rate (A-1, sec"1)  of a spatially uniform, switched-on, field through a prolate  27 spheroidal shell, has been related to this geometry of the bulb's shell along with the orientation of the field with respect to the shell15.  The shell is assumed to be an isotropic conductor  whose fields are described by Poisson's equation. The calculation involves describing the bulb geometry in spheroidal coordinates.  The bulb is assumed to be a shell of  infinitesimal thickness so that the dimensions of the conducting surface are small compared to the wavelength of the applied field.  The surface resistance of the shell is assumed spatially  isotropic and made of an ohmic conductor. By solving Poisson's equation for the electric field at the surface of the bulb subject to the constraint of charge conservation, an equation describing how the wall charge leakage is related to our previously defined electric fields can be constructed as follows dEA/dt=dEI/dt+k0EI  (3.11)  where A^1 is the time constant for the decay of the wall charges when EA=0 .  Equation (3.11) must be solved subject to the  condition that ET is reset to zero every time |£'J|=£'0 . It should also be kept in mind that the conductivity for glass bulbs can be dramatically increased16 by orders of magnitude as a consequence of environmental moisture.  Increases  in temperature T can also make it possible for charge carriers to escape from traps through thermal excitation thus increasing its conductivity as  (-1/T).  28  3 . 7 Time  Lags  Before an electrical discharge can be established in a gas, two conditions must be simultaneously satisfied.  Firstly, there  must exist at least one suitably located free electron in the gas.  Secondly, the applied electric field must be of sufficient  strength (greater than ET)  and duration to ensure that the  initial electron can produce the necessary sequences of avalanches that cause breakdown.  Without the presence of an  initial electron a discharge cannot develop even if the electric field is greater than ET (the breakdown field of the gas).  Thus  one must wait until an initial electron is liberated by some means before discharge can be observed. Free electrons which are produced naturally in the atmosphere (due to the arrival of cosmic rays, local radiations, or by the ultraviolet radiation from the sun) are found in very low concentrations of the order of 100-500cm-3 f.  The rate of  production of these electrons is also found to be quite low at approximately 10c/n"3s"lf.  Furthermore, once a free electron of  this type is liberated, it can quickly become attached to electronegative oxygen effectively eliminating it as a source for initiating a discharge.  1  refer references)  to Morgan,  C,  As a result, there is a small  "Irradiation  and Time Lags",  p 655  (see  29 probability of there being a free electron present when an overvoltage impulse is applied to an ordinary spark gap whose electrodes are exposed to the atmosphere. elapse before gap breakdown can occur. all gases.  Hence some time will  This time lag occurs in  Clearly, breakdown is always preceded by a waiting  period which is known as the total time lag of breakdown, ttot. This lag is comprised of two well defined parts; or statistical time lag ts  the initiatory  and the formative time lag t£.  tt  is  the time taken for a full breakdown to occur following initiation of the process.  The total time lag (ttot) is given by the  expression  3.7.1  Statistical  Time Lag  The statistical time lag is the period of time which elapses between the instant the electric field (which is greater than the static breakdown field ET)  is applied and the arrival of a free  electron which initiates the breakdown.  If the electron is  liberated by natural causes this lag may be as long as several seconds but is usually about 10~2s.  The liberating processes are  statistical in nature and the time lag is inversely related to the rate of electron liberation.  Thus ts  can be minimized with  30 the use of external irradiators or with exposure to powerful illumination. Since we know that the most effective place to liberate electrons is at the cathode side of the vessel (where they will then suffer maximum amplification as they traverse the greatest distance in route to the anode), it is useful to consider the relationship between the statistical time lag and the electron emission rate from this position.  Given an initial rectangular  voltage impulse of sufficient amplitude (greater than V^g - the starting potential) applied to the field generating electrodes of the vessel at timefc=0, it can be shown17 that, for fc^fcf  where ~t is the average time lag occurring over a number of trials N, and n is the number of trials in which breakdown has not occurred by time t .  3.7.2  Formative  tf  is the formative time lag.  Time Lag  After our electron has been liberated at a suitable position, preferably near the cathode side of the vessel to ensure maximum amplification, it undergoes exciting and ionizing collisions, initiating a succession of electron avalanches antecedent to breakdown of the gas.  The time taken for this  order of events from the end of the statistical time lag to the  31 onset of breakdown is the formative time lag tf.  The duration  of the formative time lag is primarily governed by the secondary ionization processes which are responsible for avalanche initiation.  This lag can range anywhere from nanoseconds to  hundreds of milliseconds depending upon the nature and pressure of the gas, the vessel materials, and the amount that the applied field exceeds the static breakdown field (tf  can be minimized by  applying a field whose strength greatly exceeds ET) . Typical results18 (Fisher and Bederson, 1951; Morgan, 1956; and Davies, Llewellyn, Jones, and Morgan, 19 63) s of formative time lag measurements tf  as a function of percentage over-  voltage (the applied voltage above the minimum voltage required to achieve static breakdown) shows an inverse, hyperbolic relation between the two.  The results show that tf  is extremely  large at lower over-voltages (less than 2% over-voltage). Secondly, the formative lag times of monatomic gases (Helium for example) are several orders of magnitude longer than those in H2 and air.  9  As quoted by J. Dutton Fields", (see references) .  in  "Spark  Breakdown  in  Uniform  32  3.8 Ionization  in Gases  For a gas molecule to become ionized, a collision with another particle is required.  If the net change in energy  transferred to the internal energy of the molecule is greater than its ionization energy, the molecule will be ionized (an electron will be ejected into the system)19.  For example, if  the incident particle is an electron whose kinetic energy is greater than the ionization energy of the target molecule, an inelastic collision may increase the internal energy of the molecule enough to eject another electron.  The degree to which  the primary electron penetrates the atom depends on its initial direction and speed as well as the number and configuration of the atomic electrons in the atom.  These factors will determine  how much the incoming electron is repelled from its original path causing various degrees of scattering.  The probability of  ionization will therefore depend on the momentum exchange, the ionization potential, as well as the polarizability (atomic electron configuration) of the target atom. So far we have assumed that electrons collide with atoms which are in the ground state.  In discharges, however, the  ionized gas often contains a number of excited atoms which require less additional energy to become ionized.  In this case  the slower and more numerous electrons now have the ability to ionize the gas thus facilitating a full discharge.  Unfortunately  33 these excited states are short lived and require a great many electrons present to aid in the discharge (i.e. they do not retain the energy put into them for very long).  34  3.9 The Penning  Effect  At energies encountered during discharges, ionization by positive ions or by unexcited neutral molecules is unlikely since their transferable energy is low.  Even ionization of excited  atoms is improbable since they are so short lived.  As a means to  decrease the threshold field required for breakdown, one must consider the Penning effect20. When a mixture of gases is used (say Ne and Xe) such that the majority constituent has metastable states which can retain their energies for extended times, the threshold for breakdown can be effectively lowered.  Because the metastable atoms are  able to retain the energy put into them, this energy can be transferred to the minority constituent upon collision21.  If  the excitation energy of the metastable level exceeds the ionization potential of the minority component, then this energy transferred can lead to ionization.  The net effect is that more  electrons are used for ionization and a is raised (see section 3.1) . Other possibilities for lowering the threshold include doping the gas with a radioactive source (say Cobalt-60). radioactive emission collides with the atoms providing more "seed" electrons which can contribute to the growth of ionization.  The  35  4. Experimental 4.1  Results  Introduction  This section provides a detailed description of the experiments which support the theoretical models previouslyoutlined.  All of the experiments were performed in the  laboratory using parallel plate capacitors with a variety of potentials applied to them.  The experiments which were carried  out on the ungettered bulbs reveal that their breakdown characteristics are sensitive to impurities which are present in the gas.  To study the characteristics of "clean" bulbs, similar  experiments were also performed on gettered bulbs (assumed to be virtually free of impurities).  The asymmetry of the gettered  bulbs also introduced exciting new AC breakdown peculiarities. The discovery of such breakdown characteristics opened up an entirely new line of research necessary to explain the AC breakdown pattern observed.  Data for both the gettered and  ungettered bulbs were obtained with plates machined to accommodate the specific bulb geometry. Single, unipolar field pulses were utilized as a means of simplifying the interpretation of the AC field results.  It was  assumed that each instantaneous AC pulse breakdown could be modelled with a single, unipolar pulse breakdown allowing us to achieve a more controlled study of the breakdown mechanism. Surprisingly, the AC results could not be reproduced with these  36 single pulses but exciting new results emerged unique to such fields. In an effort to filter out any adverse electrical noise caused by the high voltage circuits, a photomultiplier proved to be the most desirable way of detecting when electrical breakdown had occurred in a bulb.  The optical emission accompanying the  breakdown was recorded by displaying the photomultiplier signal on the oscilloscope. A collection of six major experiments are combined to give the physical models presented in order of occurrence. gettered and ungettered bulbs are compared for each. projects were as follows:  Both the The  1) The basic phenomena of breakdown in  uniform time-varying fields.  2) How periods of dormancy affect  the breakdown phenomena and how prolonged discharge exposure revives the bulbs back to their original states.  3) The  application of single voltage pulses to establish the presence of wall charge freezing.  4) The appearance of a lag-time occurring  between field application and breakdown.  5) Investigation of a  mechanism for reducing the above phenomena through the comparison of time lags of both dormant and revived bulbs.  6) An attempt to  explain the evolution of a discharge which occurs in an asymmetric bulb (such as a bulb with a getter). Each project described below consists of an account of the experiments and the results.  The results are then interpreted in  terms of the physical models which best describe the observations.  37  4.2 Experimental  4.2.1  Apparatus  Introduction  The experimental apparatus consists of two main parts:  a  device for generating electric fields and a device used for detecting and monitoring the electrical breakdowns.  Electric  fields are produced with two primary circuit configurations.  One  generates a uniform field whose magnitude varies sinusoidally with time in a fixed direction while the other delivers a wide variety of unipolar pulses of various "shapes".  The breakdown  detection system will be described first, followed by descriptions of the field generation equipment.  4.2.2  Instrument  Used to Measure  the Light  Pulse  Emission  The flashes of light emitted at breakdown are conveyed from the bulbs to an RCA IP21 photomultiplier (9 stages with a rise time of less than 2 ns)  by a vinyl clad glass fibre bundle (2 mm  in diameter and 1.5m long) which was given added rigidity in the electric field by passing it through a hollow tube made of insulating material (30 cm long, 1 cm in diameter).  The  photomultiplier is operated with its cathode at -900 V with respect to ground.  The signals are viewed using a storage  oscilloscope (Tektronix Inc., Model 468 Digital Storage  38 Oscilloscope).  The output from the oscilloscope can be easily-  fed into a digital signal counter in order to record the number of light flashes. All of the following experiments were performed at 293±4 Kelvin at less than 70% relative humidity.  Figure 4.1 shows the  apparatus configuration used to detect the pulses of light from the bulbs.  PARALLEL PLATES  •t o  OPTICAL  FIBER PHOTOMULTIPLIER  BULB  o OSCILLOSCOPE -^*A»y**Y  Light signal displayed on oscilloscope.  Figure 4.1 Apparatus used for detecting optical pulses.  40  4.2.3  Apparatus  Fixed  for Generating  Uniform Time-varying  Fields  in a  Direction  All of the electric fields that will be described are generated between the plates of a parallel plate capacitor.  The  plates consist of sheets of polished aluminum or brass with rounded corners serving to reduce corona at higher field strengths.  The gettered and ungettered bulbs have separate plate  configurations which serve to accommodate their different geometries while minimizing any electric field alterations. The capacitor plates for the gettered bulbs plates consist of square 2 6 cm by 2 6 cm sheets of aluminum mounted horizontally on 1.2 cm thick lucite sheets.  The plates are separated by  vertical lucite slabs (negligible conductivity) a distance of 9.5 cm so that they are completely enclosed by insulation for safety. All sides of this lucite box are machined to ensure uniform separation over the aluminum surfaces.  The upper plate is  customized to remove the bulb's getter from the field by means of a machined 6 mm wide channel through which the getter tube passes.  With the bulb in place the bottom of the channel is  sealed over with aluminum foil to keep the field uniform by encircling the getter tube with aluminum.  The optical fibre is  passed through a lucite tube (2 cm in diameter) into the centre of the plates where the bulb is mounted.  The negligible  conductivity of the fibre and the tube does not influence the  41 charge distribution of the plates nor does it alter the uniform distribution of the applied field (see Figure 4.2 ) . The parallel plate capacitor for the ungettered bulbs is made from two circular brass sheets (of radius 14 cm) mounted horizontally on sheets of .6 cm thick wood.  The upper sheet is  supported by eight vertical lucite rods (1.2 cm in diameter, 10 cm high) mounted in a circle surrounding the plates.  Three of  the sides are also enclosed using .6 cm thick plywood for safety (the front is left open for easy access to the bulb).  The  upper/lower plate separation is vertically adjustable by means of a brass bolt connecting the centre of the plate to the top wooden sheet.  Three nylon bolts positioned equidistant from the centre  allows the upper plate to be aligned correctly with the lower one.  The plates are parallel (±2 mm) and the separation is  normally set to 6.3 cm.  The optical fibre sensor is passed  through a non-conducting tube (1.2 cm in diameter) mounted on a sheet (30 cm long, .6 cm thick) of lucite containing a horizontal hole used as a nest for mounting the bulb in the middle of the field.  The non-conducting properties of this mount serve to keep  the applied field unaltered (see Figure 4.3 ) . The above capacitors are powered by a 60 Hz AC signal from a 50:1 transformer (Hammond Manufacture CO.) fed by a Variac (0-120 V 60 Hz Ohmite MFG. C O . ) . The voltage applied across the plates is recorded with a potential divider connected to a Digital Volt Meter (Fluke and Philips, Model 87 -True RMS Multimeter).  The  magnitude of the field can be easily adjusted using the Variac.  42  Gettered  Aluminum Plates  Tfo Pulse Generator p ^ ^ ^ ^ ^ ^  Lucite Box Figure 4.2 Parallel plates used with gettered bulbs.  Figure 4.3 Parallel plates used with ungettered bulbs  43 The apparatus is capable of generating fields up to 3 0 kV/m and normally operates at fields of about 15 kV/m.  Figure 4.4 depicts  the configuration used to generate these fields.  44  120 U.A.C. 60 Hz  o  TRANSFORMER  UARIAC  OSCILLOSCOPE i  PARALLEL PLATE CAPACITOR  i  (1000)R  1  D.U.M.  UOLTAGE DIUIDER  Figure 4 . 4 Device used t o produce t h e AC f i e l d s .  45  4.3 The  4.3.1  Bulbs  Gettered  Bulbs  The gettered and ungettered bulbs share the same appearance with one obvious exception.  On one side of the gettered bulbs  there exists a small (6 mm in diameter) hole that exits into a cylindrical tube which joins the bulb to its getter.  The getter  is enclosed by cylindrical bulb made of the same material as the lower part of the bulb (pyrex or soda glass).  Lining the inside  walls of the getter is a thin, metallic conducting surface.  Two  wire leads are connected to the inner conducting rings and exit the getter to be accessible outside the bulb.  The leads allow us  to connect the metallic getter to the equipotential surface lying closest to the getter.  This minimizes any field alterations near  the spherical part of the bulb that the getter may cause because of its conducting nature.  The getter tube allows one to locate  the bulb in the field while the getter remains isolated outside. The bulbs are filled with a Neon-Xenon mixture (2 % Xenon) to utilize the Penning effect so as to reduce ET.  After the  bulbs are filled to a pressure of 5 Torr the getter's metallic rings are heated to the point where it can actively adsorb impurities existing in the gas such as oxygen.  Figure 4.5 (a)  depicts a typical gettered bulb used to obtain data for this thesis.  46  Figure 4.5 Types of bulbs used; a) gettered and b) ungettered.  47  4.3.2  Ungettered  Bulbs  The ungettered bulbs are similar to the gettered bulbs except that the getter tube is replaced by protruding nipple (see Figure 4.5 (b)). These spherical bulbs are all about 4.0 ± 0.2 cm in diameter and are made of pyrex.  The bulbs are initially  evacuated to different levels and then filled with Neon.  Since  these bulbs possess no getter to help clean out unwanted impurities in the gas, it is the strength of the initial vacuum that determines their "quality".  Bulbs that were originally  evacuated to 10"5 Torr are considered "clean" bulbs while others evacuated to 10~3 Torr are "dirty" . During experiments these bulbs are positioned with their nipples (see Figure 4.5 (b)) parallel to the plates and in line with the equipotential surfaces. this way  The nipples are maintained in  in an effort to minimize any differences in path length  that they may cause.  It is stated in Friedmann22 that the  threshold field can vary as much as 19 % when the nipple deviates from pointing horizontally to pointing vertically due to the extra path length the nipple can provide.  48  4.4 Electrical  4.4.1  E  Breakdowns  Ungettered  in an A.C.  Field  Bulbs  Optical Pulses (breakdown)  A* r\  /  - < * • - " "  - Eo  \\  "X  /  I 1  /  /  y \  v  Time  V  \  Applied Field ~  \  \  \  /  1r Figure 4.6 Typical oscilloscope trace showing the light pulses (breakdowns) superimposed on the applied field. When a sinusoidal voltage is applied to the parallel plates, the number of pulses of light (breakdowns) occurring on each rising/falling edge of the voltage waveform is found to increase linearly with the magnitude of the applied field.  This  relationship is observed by simultaneously viewing the voltage  49 across the plates and the output from the photomultiplier on the oscilloscope.  The oscilloscope is triggered from the channel  receiving the voltage applied to the capacitor plates in an effort to view a stationary sine-wave and its corresponding light pulses.  The addition of the two signals shows the light-pulses  superimposed onto the applied field.  Figure 4.6 depicts a  typical trace viewed on the scope for the ungettered bulbs. Clearly, breakdowns for these bulbs occur on both the rising and falling edges of the sine-wave.  The gettered bulbs create a  similar trace except that the breakdowns viewed in the bulb are found to occur only on one edge (be it the rising or the falling edge).  This phenomena will be described in later sections in  much more detail.  For the ungettered bulbs, the light pulses on  any given rising or falling slope are separated such that the applied field (amplitude EA) between successive pulses.  changes by a constant magnitude Breakdowns occur whenever Ej=EQ.  (EQ) The  value of E0 depends on a combination of the bulb's pressure, diameter, and gaseous content.  The graphs of EA versus the  number of pulses per half-cycle of the applied voltage waveform are presented in Figure 4.7 for the two ungettered bulbs as well as a gettered bulb.  The light pulses are extinguished once the  peak to peak field values are reduced below E0 . values of EA greater than E0/2  However for  the pulse emission is maintained.  While pulses do in fact appear for these field values, their probability of occurrence diminishes the closer the field is  E-applied vs. No. Breakdouns/Half-cycle  —i—  12 4 6 8 10 No. Breakdouns/Half-cycle •  Ungettered/clean  +  Ungettered/dirty  X  Gettered/clean  e r r o r i n f i e l d = +- 5 percent  Figure 4.7 Applied f i e l d v e r s u s number of  breakdowns/half-cycle.  51 decreased to E0/2 .  The number a breakdowns occurring per half-  cycle in the ungettered bulbs can be predicted by equation (3.10).  For the gettered bulbs we need to divide this equation  by two. It is also observed that the AC field can be increased slowly from zero to peak to peak values much greater than E0 without any breakdown occurring.  Once a discharge occurs it can  "seed" subsequent discharges at lower field values, since the internal make-up of the bulb has changed to facilitate further breakdowns.  This value of EA that first causes the pulses to  occur is called the threshold value  ET.  For the ungettered bulbs ET is found to increase in a logarithmic fashion over time provided no breakdowns have occurred over this period.  We call the period over which the  bulb remains idle (i.e. exposed to no electrical breakdowns) the dormancy time tD.  Studies show that for longer values of  tD,  higher threshold values ET are required to break down these bulbs.  Ultimately ET reaches a limiting value dependent on the  properties of the bulb and its content. The internal quality of the bulbs used greatly affects the outcome of the threshold values that are achieved after each dormancy time.  The amount of oxygen remaining in the bulbs prior  to their inflation with the appropriate gas exerts a significant influence on ET.  Bulbs with a larger oxygen content are termed  52 "dirty", while those with a low oxygen content are labelled as "clean" bulbs. As explained earlier, the ungettered bulbs possess a value of ET which is found to increase logarithmically with  tD.  Ultimately, with very long dormancy times, ET achieves an upper bound [E0)  which depends on the type of bulb.  For the "dirtier"  bulb, the curve rises to a much higher threshold value over a longer dormancy time.  The "cleaner" bulb, however, reaches its  somewhat lower threshold value much more quickly.  Consistent  with the above results, the "cleanest" gettered bulbs do not exhibit any variation in their thresholds over time.  The graphs  of ET versus tD for the ungettered bulbs are shown in Figure 4.8 and Figure 4.9 on the next page. Data for these figures were obtained by testing the threshold values necessary to break down the bulbs consistently (at least once on every complete cycle on the AC voltage) after they have been dormant for a varying length of time.  Having  recorded this value, the field EA is subsequently increased with the variac to a value high enough to obtain at least ten pulses on each rising and falling edge of the slope (as viewed on the oscilloscope and predicted by Figure 4.7 shown previously).  The  field is then held here for approximately one minute thus exposing the inside of the bulb to at least fifty-thousand pulses of light from the breakdown plasmas.  This is discovered to reset  the bulb effectively to its time ^=0 state (as predicted by  53  Threshold Field vs. Time Dormant (Ungettered/clean bulb) 13 E,j=  12T) ,—t  OJ  s\  12700 U/m  11  w  "O  10 1 UJ 0 T) _ t  O  9  r  i—  *~s  y  Ul  7  QJ  -E L = 6020 U/Tn  10 time  15 20 (hours)  25  e r r o r i n E--fields = +- 10 percent  •  30  Bulb A2  Fignre 4.8 Threshold f i e l d versus t o t a l time t h a t the bulb remains dormant.  T h r e s h o l d F i e l d v s . Time Dormant (Ungettered/dirty bulb) 20-  e \ z>  18-  N  Eu= 20000 V/m  16XI  ^T  c 14aj  i UJ  o .c  W 3 O  121086-  W-^  E, = 6400 U/m I  •  10 error  "  1  20  1  30 time  •-1  40 (hours)  i n E - ^ i e l d = +-10 percent  •  1  50  T  60  70  Bulb B2  Figure 4.9 Threshold field versus dormancy time for a bulb with a higher oxygen content.  54 equation (4.3) described in section 4.4.3 on resetting the bulb). The above process is then repeated for a new tD thus mapping out the graphs for both the ungettered bulbs.  55  It is found that the increase in ET over time can be described empirically with an exponential equation of the following form  ETU) = iE0-EL) [l-e-pt°]  Where ET(tD)  +EL  (4.D  denotes the time behaviour of the threshold field  and EL and Ea are the smallest and largest values that ET can achieve (which depends on properties of the gas).  Notice that  ET(tD=0) =EL and ET(tD=<*>) =Ea in agreement with Figure 4.8 and Figure 4.9 previously shown.  The constant p"1 denotes the time  taken for a given bulb to show a substantial increase in ET.  If  we define a threshold ratio as Q = [ET (t^) -EL] / [EV-EL]  (4.2)  then the slope of the graph In{1-Q) versus t should give us this time constant p. The graphs of In(1-0) versus t for both ungettered bulbs are depicted in Figure 4.10 and Figure 4.11 on the next page. We find that the cleaner ungettered bulb has a slope of -0.12 ± 0. Olii"1 which corresponds to a time constant of P"1 = 30000s.  The time constant of the dirty bulb is much larger  at p"1 = 116000s. Since the bulbs containing the purest gas (i.e. the gettered bulbs made of pyrex) do not exhibit any variation in ET over  56  l n ( l - Q ) v s . Time Dormant <Unget t e r e d / c l e a n b u l b ) -0.5_1  1  • • ^ •*L ^^fc^  -Slope = - . 12 +- . 0 1 _^  "  ™^^^_  ~ -1.5o -2-  m  ^ ^ ^ T  ~ -2.5-3-  -3.5-  .  c)  5  10 time  15 20 (hours)  ^ " ^  25 •  30  Bulb A2  Figure 4.10 Graph used to determine the time constant of the increasing threshold field for the clean bulb.  l n ( l - Q ) v s . Time Dormant <Ungettered/dirty bulb) 0-  -0.5i  ^ * ^ - « ^  £  3  m  ^ ^ - S l o p e = - . 0 3 1 +- .001  -l-  s-\  o  .1-1.5KS  c  ~  -2/.. D ~l  C)  i  10  20  30 time  40 (hours)  50  60 •  70  Bulb B2  Figure 4.11 Graph used to determine the threshold constant for the dirty bulb.  field time  57 time, we can assume that this phenomena is caused by the foreign impurities which are present in the ungettered bulbs (also made of pyrex).  According to Townsend's breakdown criterion, the only  way that ET can be varied is through changes in the parameters a or y (refer to equation (3.2)) . If a = f(pd)  , and the vessel  dimensions d remain constant, then any changes in this parameter must be due to a change in the gas pressure or its ionization potentials.  Since these changes are attributed to the impurities  present in the gas, any reasonable change in a must be a result of this impurity being "pumped" out of the system in a large enough quantity so that the gas pressure and/or nature can vary. This would require a very large amount of the foreign particles to be present in the gas originally (i.e. the impurity could be adsorbed by the inner walls of the vessel over times of order p - 1 ).  On the other hand changes in y can only be a result of a  change in the properties of the vessel walls which affect secondary emission.  If a change in the secondary emission  parameter y occurs over times of the order of p"1, then we would expect changes in ET to occur for both the AC and single, unipolar fields.  However, small amounts of impurity present in  the gas does not affect the value of ET for single unipolar field pulses over times tD.  The fact that the ungettered bulbs  exhibit increases in ET over times tD when exposed only to an AC field suggests a breakdown phenomena which is not readily  58 explainable using the Townsend model (e.g. absorption/desorption of impurities which can change the breakdown parameters a and Y) . The purpose of the rest of the study of these dormancyeffects is to get as close as possible to obtaining a parametric description of the associated phenomena involved.  59  4.4.2  Time Constants  Ungettered  Involved  In Breakdown  Formation  for  the  Bulbs  If we assume that only the impurities are adsorbed onto the vessel walls, and we wish to describe this effect using the Townsend breakdown model, then we must also assume that it takes times approximately of the order P"1 for this adsorption to affect ET.  Using the ideal gas law P = NkT with N in units of  molecules per cubic centimetre, one can easily deduce that a typical gas at 1 Torr of pressure contains JV=4*1016 cm'*. According to Von Engelh, the mean free path A for neon at a pressure of 1 Torr (0°C) is 12*10~3cm.  Given that the radius of  the bulb is at the most 2.5 cm, the average number of collisions that a typical particle receives before it reaches the wall is given by N = (r/k)2=  (  2  • 5 cm 3  ) *43000  12 xl0~ cm  (i.e. the average particle undergoes about 43000 collisions before it reaches the vessel walls). The time between collisions for the gas in the bulb is related to the collision frequency fc  h  (see  refer to von Engel, references)  "Ionized  and is given by  Gases",  table  3.1 p.  27,  1955.  60 T = l/fc = \/v=  12xl0 3f " f =3xlQ-7g 40000cm/s  where v=40000cm/s is the approximate acoustic velocity for neon at 1 Torr (slightly greater than the speed of sound).  The  average time taken for particles to reach the vessel walls (43000 collisions later) is given by t waJJ =TxiV c = 0 . 0 1 3 s  Hence the average time taken for any particle (impurities included) to reach the vessel walls is of the order 10-2s.  Thus  the number of particles arriving at the vessel walls each second is jllrWfc^n * 2xl020s-1 Given that it takes times of the order of 10~2s for particles (impurities) to collide with the vessel walls, combined with the fact that there are 2*1020 collisions with the walls each in second, it is unlikely that changes in the Townsend parameters, a and y, as a result of absorption/desorption of impurities out of the gas and onto the vessel walls, is the mechanism responsible for the increases observed in ET over tD (recall that the time constants involved in this process, p_1, are of the order of days).  This would suggest that the AC breakdown  mechanism is governed by parameters external to those described by Townsend.  Furthermore, if changes in a and y were  responsible for the changes in ET, then we would expect a  similar dormancy effect to occur when the bulbs are exposed to single, unipolar field pulses. However, as we will see, results indicate that these dormancy effects are only observed when the bulbs are immersed in AC fields.  62  4.4.3  Resetting  a Dormant  Bulb  Initial attempts to reset the dormant bulbs back to their ^=0 state were made by heating the bulb with a hair dryer for long periods of time (up to an hour) in an effort to dislodge the trapped impurities through thermal agitation. proved to have no effect whatsoever on ET.  These attempts  We can conclude that  the increase in thermal kinetic energy did not significantly modify any possible adsorption/desorption mechanisms which might occur with impurities. As mentioned above - a dormant bulb can be reset to the tD=0 condition if a large number of breakdowns occur within the bulb in a time which is small compared to values of tD during which ET increases significantly.  In other words, the duration  of time for which the bulb is exposed to light-pulses is much smaller than the time required for the physical properties of the bulb to change.  Conveniently, the breakdown provided by the 60  Hz AC circuit satisfies this constraint. To record the number of pulses required to decrease the threshold value, the photomultiplier output was fed into both the oscilloscope and a signal counter as depicted in Figure 4.12 on the next page. The field is first applied to a previously dormant bulb whose threshold value at the time is near its upper limit. The  63  PARALLEL PLATES  4 c  OPTICAL  FIBER  ) y  I  PHOTOMULTIPLIER  BULB  1  SIGNAL COUNTER  O OSCTI  i nsrnPF  Figure 4.12 Device used to count the number of breakdowns.  64 magnitude of the 60 Hz field is slowly brought up to its threshold where the value of ET is recorded.  Every pulse of  light that is emitted at this point is recorded by the counter. After a significant number of breakdowns have occurred, the field is reduced to zero where the process starts all over again.  The  subsequent threshold values are observed to decrease exponentially with the total number of pulses the bulb is exposed to.  The results of this phenomena for both the ungettered bulbs  is presented in Figure 4.13 and Figure 4.14 (note that the gettered bulbs due not exhibit any change in threshold over time). The resetting of these bulbs can be described mathematically as a function of the number of breakdowns N with an exponential of the following form; ET (N) = (Ea-EL)  e~kN+EL  (4.3)  where the constants Eu and EL are exactly the same as those described previously in equation (4.1).  As expected, this  equation represents a reversal of the processes that led to increases in ET where now we have the conditions that ET(N=0) =EU and ET{.N=<x>) =EL.  The constant ic'1 represents the rate at which  the bulb can be reset with successive discharge exposure.  If we  once again define a threshold ratio Q as given by equation (4.2), then plotting a graph of slope which is equal to the constant  versus  will provide a  of (4.3).  The graphs are  65 E—threshold dP O i—i  •a •—(  CD 4-1  1 W  Breakdowns  14  + & >  *""*'  v s . Total W o .  15  13 12 T3  c  11  o  10  10 03 EH  9  TJ i—1  O  8  rn en  7  E-i  6  X3  n .£3  500 1000 1500 2000 Total N o . Break Downs (+- 5%)  2500  Clean Bulb  Figure 4.13 Resetting the threshold field for the clean ungettered bulb with breakdown exposure.  E-threshold ae>  24  i—i  2 2 -d  i  -  V  v s .  U=  Total  22180  Wo.  Breakdowns  V/m  20 - \  +  E-  >  *"-**  en  T3  T3 C  I—I  CD  CD -~H 4-1 H  3  o EH  c3  18 -  \  ^Sc i  16 14 -  O  i  10-  t  *  3\^  ^^c  S  8 -  V/m  \  y*  JS 00  CD  L = 6380  • SI  12 -  T3  i—l  E  N  s  J  ^~~~\-*-^ J_  '**-[•  c]  l  *  ^  T  6 -  i  C)  1000 1=1  i  i  2000 3000 T o t a l No. Break  Dirty  i  4000 Downs  1  1  5000 6000 ( + - 5%)  7000  Bulb  Figure 4.14 Resetting the dirty ungettered bulb with breakdown exposure.  66 given in Figure 4.15 and Figure 4.16 complete with slopes. Table 3 presents the various constants used to characterize the internal quality of the ungettered bulbs. and k'1  Lower values of P"1  are attributes of a bulb whose discharging gas (neon in  this case) is free of impurities.  Such a bulb will rise to Ev  much more quickly than "dirtier" bulbs. Bulb  P"1 (sec"1)  k~x  clean  30000  400  dirty  116000  1900  (N'1)  Table 3 Time and reset constants for the ungettered bulbs.  67 l n ( Q ) v s . T o t a l No. Breakdowns 0* -0.5 -  %_-.  , Slope  = -.00260  +-  .000O7  -1 ° \ J * Of  a  Roll-off  at  N=1250  -1.5-2-  XT  \  /'  Vj—I  1  -2.5 -3-3.5 C)  1  1  •  1  —ci-  1  2500  500 1000 1500 2000 T o t a l N o . B r e a k Downs (H— 5%) n  Clean  Bulb  Figure 4.15 Graph used to determine the reset constant k for the clean ungettered bulb.  ln(Q) v s . T o t a l No. Breakdowns 0.5Slope  0 * ^ v x.  = -.00052  +-  .00004  Roll-off  at  N=4100  _, s '  -0.5 ^ ~ \  -1 c  1—(  -1 .5 -2 -  ^\"  t  •  ^L^_  -2.5 -  •  -3 ) C  1000 2000 3000 4000 T o t a l N o . B r e a k Downs a  Dirty  5000 6000 ( + - 5%)  70 0 0  Bulb  Figure 4.16 Graph used to determine the dirty ungettered bulb's reset constant k.  68 4.5  Breakdown  4.5.1  In  Unipolar  Electric  Fields  Introduction  The equation which describes the behaviour of the internal field Ej- of the bulb under the influence of the external applied field EA is given by dEA/dt = dEI/dt+kaEI.  The simplest way to  study this equation is to examine the breakdown time lags obtained with a variety of elementary-shaped unipolar voltage pulses which create the breakdown requirements.  Simple  arrangements of capacitors and resistors which discharge through spark-gap switches enable us to produce a wide variety of pulse shapes with known slopes dEA/dt  given by RC time constants.  By  changing the various resistors that make up the circuit, we can easily create pulses with square, trapezoidal, triangular, or even rectified sine-wave shapes.  The breakdown time lags with  respect to the applied voltage pulse proves to be a useful way to study the effects of wall charge depositions within the bulb along with the various constants that describe the above equation. Results show that the behaviour of the gettered bulbs exposed to these fields is much more stable and predictable than some of the ungettered bulbs.  We will see that the requirements  for gaseous breakdown for some of the ungettered bulbs not only depends on the external change in field dEjdt  (thus affecting  69 the internal field Ex),  but also on the length of time under  which the bulb is exposed to these new fields. We also examine a way to reduce this lag time between the field change and the discharge it produces.  4.5.2  Apparatus  Used to Produce the Unipolar Electric  Field  Pulses  The apparatus consists of three main components.  First and  foremost are the parallel plates (Figure 4.2 and Figure 4.3 depicts the plates used for both types of bulbs), across which the voltages are applied.  The other two components are the  circuits which produce the various voltage pulses.  One circuit  is a simple RC configuration coupled with a set of spark gaps which switch on non-concurrently thus defining the positions of up to three discontinuities in dEA/dt.  The gaps are externally  triggered and provide the attached capacitors with a resistive path through which they can discharge and create the rising and falling edges of the desired pulse.  The resistors through which  these capacitors discharge can be readily exchanged to introduce different slopes to the applied field.  The advantage of using  spark gap switches is to allow the capacitors used to discharge very high voltages (up to 16 kV) quickly and cleanly across the plates with minimal electrical noise.  Other switches can often  present us with unwanted (and uncontrollable) voltage fluctuation during their activation.  Figure 4.17 shows the arrangement of  SPARK-GAP 1  V\Arp o R.  HIGH D.C. VOTAGE (VARIABLE) R  R-,  PARALLEL PLATE CAPACITOR  SPARK-GAP TRIGGER (DELAYED)  v W - o  <J  B  SPARK-GAP 2  Figure 4.17 Device used to produce a three-sided, unipolar voltage pulse with a flat top.  71 spark gaps and capacitors used to produce a high voltage three-sided unipolar voltage pulse. Notice that both parallel plates are initially held at the same high voltage (maintaining a net field of zero between the plates) before one plate is shorted out through a capacitor thus defining the first rising edge of our pulse. The final piece of apparatus is the circuit used for triggering the spark gaps described above.  This circuit once  again relies on the discharging of capacitors through externally triggered switches.  The switching in this circuit is performed  by two krytrons triggered at different times with the output from separate channels of an adjustable delay unit.  The krytrons are  tiny glass cylinders (1.5 cm high and 1 cm in diameter) which contain a radioactive emitter used to facilitate the breakdown of the gas it contains.  Inside these bulbs are four electrodes.  One electrode known as the "keep-alive" (or KA on the circuit diagram) is held at a constant high voltage thus maintaining a highly mobile plasma source within the krytron used to initialize a quick breakdown across the anode and cathode (A and K on the diagram respectively).  Another electrode called the "grid" (G on  the diagram) is connected to the output of a delay unit which is stepped-up with a 3:1 low voltage transformer.  The "grid" acts  as a trigger that completes the discharge across the gap between the anode and cathode, when a voltage spike from the delay unit reaches it.  When triggered by this delay unit, the krytrons fire  and surrender a spike of voltage to the connected 5:1, high  72 voltage, step-up transformers which feeds into the trigger pins of the spark gaps.  The configuration of the circuit described  above is presented in Figure 4.18 on the next page. The above two circuits work together to provide a wide variety of electric field pulse shapes. Notice that by simply connecting one plate to a point behind a spark gap (points A or B of Figure 4.17 - the square pulse generator), we can achieve a voltage spike if we trigger this gap alone.  Furthermore, voltage  pulses with flat tops are achieved by connecting one plate to the capacitor side of one spark gap, while the other plate is connected to the capacitor side of the other gap (refer to Figure 4.17 - points C and D).  A "positive" step pulse is  defined as a change in the net field from zero to its final value \E\ . Conversely, a "negative" step pulse occurs if the net field between the plates changes from |i?|*0 to zero.  The field change  across the plates is "positive" if both plates are initially held at the same voltage, and one plate is then discharged through its capacitor (i.e. switch closed in Figure 4.20 to achieve a "positive" step pulse).  Similarly, if one plate is initially  held grounded (with respect to the other) while the other is discharged from a high voltage, the field change across the plates is "negative" (i.e. switch open in Figure 4.20 to obtain a "negative" step pulse).  Figure 4.19 and Figure 4.20 depicts some  of the systems used for generating the wide variety of unipolar pulses shown as output.  R:  rWV  3:1  TO DELAY  LOW VOLTAGE TRANSFORMER  (Ch. 1)  5i1 HIGH VOLTAGE TRANSFORMER  TO #1 SPARK GAP (trigger)  3:1 LOW VOLTAGE TRANSFORMER  v\/v<l  5:1 HIGH VOLTAGE TRANSFORMER  TO DELAY UNIT (Ch. 2)  TO #2 SPARK GAP (trigger)  HIGH D.C, VOLTAGE  Figure 4.18 Device used to trigger the spark gaps to initialize their firing.  74  CONNECT HERE FOR VOLTAGE' RAMP  k  SPARK GAP  ^CHARGE  HIGH D . C , VOLTAGE  CONNECT HERE FOR VOLTAGE SPIKE  vW  PARALLEL  PLATES  SWITCH •«D  ISCHARGE  ^± ( B U L B )  KRYTRON TRIGGER CIRCUIT  Figure 4.19 Device used to produce the voltage spikes or ramps shown as output.  • I  Pot. Dlff. • *  HIGH D.C. VOLTAGE (VARIABLE)  PARALLEL  switch  wv lh  SPARK-GAP TRIGGER  C=*  O  PLATE CAPACITOR  1/  I'  SPARK-GAP  Net field change:  t  switch open  spark sap breaks — i down 0 0  d  switch closed  Figure 4.20 Device used to produce positive and negative voltage steps across the plates.  75  4.6 Behaviour  of the Ungettered  Bulbs in the Unipolar  Fields  During initial attempts to determine exactly the number of breakdowns required to reset the dormant ungettered bulbs, single unipolar pulses were applied to the plates with the intention of creating a single breakdown within the bulb.  The idea was that  the total number of breakdowns that the bulb was exposed to would be exactly equivalent to the number of pulses which were applied. To my surprise, however, the application of a single unipolar square pulse did not cause a breakdown every time even though the internal field Ex changes by a value greater than E0 .  To the  contrary, some of the ungettered bulbs exhibit no breakdowns whatsoever during the times that large field changes occur.  The  only field pulse conditions that consistently cause breakdown of these bulbs is when the external field is held at its final value for longer than .5 seconds.  That is, after the initial change in  field had occurred, the light pulses (breakdowns) appear with a lag time of at least .5 seconds.  This breakdown lag time is  apparent for both "positive" and "negative" (as previously defined in section 4.5.2) step field changes. Due to these extremely long lag times, the position of the breakdown relative to the start of the unipolar field pulse can not be viewed on the oscilloscope simply because the time base of the scope is not long enough.  Thus the oscilloscope can not be  triggered with the voltage pulse, but must be triggered with the  76 output from the photomultiplier in order to view these breakdowns.  On top of this, limitations in the delay unit makes  the longest possible pulse width only around .2 seconds.  Given  these constraining conditions, square unipolar field pulses (as produced with the circuit shown in Figure 4.17 previously) were applied to the plates with the output from the photomultiplier triggering the oscilloscope.  Square, unipolar, field pulses of  magnitude .12 MV/m (five times the breakdown threshold for these bulbs), and width .2 seconds, still do not provide sufficient field conditions for breakdown to occur within the bulb.  That  is, lag times can not be dramatically decreased (to below .2 seconds) with large increases in the field strength.  However,  when the plates are subject to step pulses, both "positive" and "negative", breakdowns are consistently viewed on the scope at much lower values of the external field.  In fact, for negative  steps in the applied field, lag times of up to 3 seconds are quite common. These lag times can be significantly decreased with the presence of radioactive source near the ungettered bulb. Specifically, when a 70 \iCi  drop of  60  Co is used as an emitter,  the lag times with the applied unipolar pulse are reduced by a factor of ten and regularly occur around 40 msec after the start of the step.  Lag times of this order allows the oscilloscope to  be triggered with the field pulse and enables one to view the light pulse relative to the step.  To minimize any field  alterations, the 2 mm disc of Cobalt is mounted onto a solid  77 lucite rod (1 cm diameter) and then capped with a plastic lid in an effort to impede current flows occurring from the bulb to the Cobalt.  Finally, to confirm that the change in lag times are a  result of a physical change within the bulb, and not simply due to the change in electric field caused by the presence of the metallic Cobalt,  the same lucite rod is positioned once again  into the field with a drop of silver solder replacing the Cobalt. As expected, the breakdown lag times return to values back up to the neighbourhood of at least .5 seconds. It is found that the lowest value of the step change occurring across the plates that regularly causes breakdown within the bulbs is 22 kV/m (same for both the positive and negative steps).  Moreover, this unipolar threshold value does  not change with the time that the bulb remains dormant as it did in the AC field.  78  4.7 The Probability  of Breakdown Versus the Repetition  Unipolar Pulsed Electric  Rate of  Fields  The following experiment involves attempts to breakdown the ungettered bulbs by applying a voltage "spike" across the plates (see Figure 4.21 for a description of this field pulse). Consistent with the above observations concerning the time lags, only the longer duration field "spikes" successfully achieve breakdown on a regular basis.  These spikes are wide enough for  the field to stay above the threshold value for a time exceeding the bulb's lag time, thus meeting the breakdown conditions.  Electric field  spikes with discharging time  R C discharge of capacitor  constants RC greater than 5 seconds are successful in producing internal breakdown in the bulb with a high probability, while spikes with  „. . _„ ml ,. . ,. Fxgure 4.21 The dimensions of a typical voltage spike.  time constants less than one second have low probabilities of causing breakdowns. Although these faster pulses do not cause regular breakdowns when applied alone, the probability of obtaining a breakdown per spike increases markedly when these field pulses are applied more frequently to the plates.  In fact, the probability of a  79 breakdown per field pulse rises to almost 100% for spike frequencies greater than 3 0 Hz and drops to nearly zero for frequencies less than 5 Hz. The probability that a discharge will occur is dependent on the conditions that exist within the bulb prior to the change in the applied field EA.  If remnants of a previous discharge are  left over in the gas, then the concentration of ions and electrons are such that there is a greater probability that they are located in positions which are most favourable for avalanche growth into a full discharge.  Favourable electrons are those  which can travel the greatest distance through the gas and are capable of creating the most ionizing collisions necessary for breakdown.  If the concentration of favourable electrons is low,  then the formative time lag increases while we wait for secondary processes (ejection of electrons from the walls of the bulb) to develop. Figure 4.22 shows the probability of a discharge occurring per spike (field pulse) as a function of the applied unipolar field frequency.  The voltage spike used for this graph possesses  an RC discharge time constant of 2 5 \is.  This graph suggests  that recombination of the ionized plasma takes times of the order of l/7Hz = 0.14 s.  That is, for field spike frequencies greater  than 7 Hz, the previous discharge does not completely recombine before the application of the next spike and can seed subsequent breakdowns with favourable electrons. Data for Figure 4.22 are obtained upon feeding the output of  80  Prob. of Breakdown/Pulse vs. Pulse Freq (obtained from clean, ungettered bulb) dp :  ±  90-  _  80-  '-T  dP  w  70-  5  60-  1i  "^ 50o  40-  S 30m 20° 10•  X!  o  £  _^Jf^ 5  o-)  (  10 15 20 25 Freq. (Hz.) (+- 10%)  30  35  Figure 4.22 The probability of a breakdown occurring on each electric field pulse versus the pulse frequency.  81  Voltage Spikes Vr  / ^ / ^ / ^ / ^ / ^ M f r ^ / * * * * ^ / ^ / r t r t 1 /l"******!/**!/*  Breakdown Figure 4.23 Typical series of voltage spikes applied to the plates at a high frequency (14 Hz) along with the photomultiplier output. the photomultiplier into both a digital pulse counter as well as the oscilloscope.  Figure 4.12 shows this configuration.  By  counting the number of light pulses that occur in a sixty second interval versus the number of electric field spikes that occurs in this same interval, we can compare the two frequencies and easily determine the probability of a breakdown per field spike. Figure 4.24 depicts the arrangement of the digital pulse counter and the oscilloscope used to count the number of voltage spikes. The configuration of the voltage spike generator is depicted in Figure 4.19 where the DC voltage supplied to the capacitor is much greater than the breakdown voltage for the particular spark  82 gap used.  In this way, if the voltage across the capacitor is  increased, the breakdown point of the spark gap will be achieved more quickly thus increasing the frequency of the spikes.  Since  the spark gap breaks down on its own at the higher frequencies, the krytron trigger is not required.  The trigger is required,  however, for the lower pulse frequencies where the DC voltage is no longer above the spark gaps breakdown potential. Figure 4.23 shows the probability that voltage spikes applied at a frequency of 14 Hz to the parallel plates has a 79 % chance of breaking down the gas on every spike (field change).  83  1  D . C . VOLTAGE HIGH D . C . SPIKE VOLTAGE *"~ GENERATOR  L  PARALLEL PLATES  CV* ^—-^  ¥  VOLTAGE DIVIDER A A  l| -  —s/VV^  A  /\ (10 D0)R  R  1  I  SIGNAL COUNTER  o OSCILLOSCOPE  Figure 4.24 Device used to count the number of electric field spikes in a given interval.  84  4.8 Gettered  Bulbs  in an A.C.  Field  Since the gettered bulbs and ungettered bulbs differ in geometry, it is not surprising that these bulbs behave differently when exposed to the same alternating fields. When the gettered bulbs are placed into a spatially uniform alternating field whose instantaneous direction is parallel to the getter tube, breakdown is discovered to occur when the applied fields are changing in one direction only.  That is,  breakdown in an AC field only occurs in the bulb if the direction of electron travel is from the "base" of the bulb towards the getter "tube" (see Figure 4.5 for a definition of getter tube and base).  If we assume that electrons are mainly ejected from the  inner surface of the bulb itself and not, in general, from the getter (or its neck), then the largest surface area capable of contributing to the concentration of secondary electrons is the base of the bulb when ejection occurs towards its getter. Conversely, when the field changes in such a way that the electrons travel back towards the base of the bulb from the getter side, the electron ejecting surface is considerably diminished and ET increases to a value greater than our applied field EA.  The total area of the surface which can eject  favourable electrons into the system is decreased by approximately 30 % because of the hole which leads to the getter  (the hole of the neck has an area of *30/nm2 while the bulb's hemisphere supports a surface area of «100ffim2).  Such a loss of  available secondary electrons decreases the value of the secondary emission coefficient y  an<  ^ is bound to increase ET.  This reasoning could explain why we get a single breakdown occurring on one edge of the alternating field, but could not explain why we get multiple pulses on one edge only because the difference in ET for both directions is only a few percent (see calculations in chapter 3.3.1).  In order for N (multiple)  pulses to occur the magnitude of EA (the applied field) must have a value of ~NET.  These values of EA are also most  certainly much higher than ET required to break down the bulb in the opposing direction.  One possible reason for this phenomena  could be that the internal field of the bulb Ex is being held at zero by some external (to the bulb) source of ions and electrons serving to keep the plasma inside the bulb alive.  4.8.1  Bulb Conductivity  and its  Effect  on Breakdown  The conductivity of the bulb (be it the gas inside or the vessel material itself) greatly affects the value of ET.  If the  conductivity of the bulb's shell remains high throughout increases in EA, then the anode and cathode sides of the vessel can charge themselves to the point where the field created by  86 this vessel wall charge separation cancels the applied field. Hence, the internal field Ex stays very close to zero and a discharge (the creation of ions and electrons) is not required (or possible) to reset Ex to zero.  Similarly, if the ions and  electrons in the bulb plasma are maintained at high concentrations for the duration of one edge of the AC field, then the conductivity of the gas remains high, the ions and electrons separate to the cathode and the anode respectively cancelling out EA and a full discharge cannot develop (EX=ET is never achieved). The mechanism(s) which are responsible for this phenomena led to the study of breakdown in the getter tube of the gettered bulb: this tube connects the bulb to the vessel which contains the getter.  4.8.2  Breakdown  in the Gettered  Bulbs  In order to study the relation between the breakdowns occurring in the bulb as opposed to those which occur in the getter tube it is necessary to position the optic fibre (which transmits the light pulses created at breakdown to the photomultiplier) at both of these locations.  We will refer to  measuring the breakdowns in the bulb as optic fibre position A and measuring breakdowns in the getter tube as position B. Figure 4.25 shows the optic fibre position in relation to the gettered bulb.  getter  bulb-  optic fibre  Position B Position A  parallel plates  Figure 4.25 The positions that the optic fibre must be in to enable the recording of breakdowns in the tube and bulb independently.  88 Figure 4.2 6 shows the light pulse output (breakdowns) from the photomultiplier for the various field/getter arrangements when the optic fibre is at position A (see Figure 4.25 ). As expected, breakdown in the bulb itself only occurs during the times when the field is accelerating electrons towards the getter hole.  With the optic fibre reading the light pulses from the  tube at position B (Figure 4.25 ), breakdowns can be observed on both the rising and falling edges when the bulb is oriented as in Figure 4.26 (a) and (b). Figure 4.26 (c) shows that when the getter neck is perpendicular to the field direction (horizontal to the parallel plates), neither the bulb nor the tube will break down when the field changes in either direction (EA brought up to 90000 V/m) .  For this particular arrangement, the surface  areas for favourable electron ejection is the same in both directions and the path length d of the tube is considerablyshortened.  Thus, if the gas is at a constant pressure p such  that decreases in the value of pd leads to large increases in Vm,  then we must be to the left of the Paschen minimum (see  Figure 1.1 ) since the tube will not breakdown in this orientation.  That is, the probability of breakdown is also being  affected by the size of the path length along the applied field, from one part of the bulb to another. in more detail in section 3.2.  This matter is discussed  a) orientation 1  b) orientation 2  Getter  Paralel plates  _  AJCL voltage  <C (_>**  Breakdown position:  VA  c) orientation 3  Getter-  T  Paralel plates < ^ Q^-Bub  Parallel / BB1  /  Non-Conductlng Shield  ^A—,  -  plates < ^ - C H C )  Getter  _ A C voltage V.  Breakdown position:  ^ AC. voltage VA Breakdown position:  1\  Discharge  Discharge  ^  No Discharge  Figure 4.26 The occurrences of breakdowns in the bulb relative to the applied AC field.  oo  90  4.8.3  Unipolar  Fields  and Gettered  Bulbs  Unlike the ungettered bulbs, the gettered bulbs consistentlyexhibited electrical breakdown for all of the unipolar field pulses described previously in both the getter tube and the bulb itself. dEjdt  Breakdowns predictably occurred for any field change  provided that the net change in the bulb's internal field Ex  is greater than E0 . The values of dEA/dt  ranged anywhere from  2500 V/ms (gradual field change) to 1.5*109V/ms  (very fast field  change). When a single, unipolar field pulse is applied to the same gettered bulb used above, the minimum threshold field required to break down the tube (optic fibre at position B in Figure 4.25 ) is found to be the same (14700 V/m) for orientations (a) and (b) of Figure 4.2 6 of the bulb relative to the field (i.e. the bulb is placed in the same positions depicted in Figure 4.26 (a) and (b) except instead of an AC field being applied to one of the plates, a rising voltage square pulse is used).  When the bulb is  oriented as in Figure 4.2 6 (c), breakdown could not be achieved for field values up to 80000 V/m for any of the unipolar pulses previously described (i.e. step pulses, square, spike).  The same  pulse was then re-applied at orientations (a) and (b) only this time the optic fibre was reading light pulses from the bulb (position A of Figure 4.25 ). This time the values of ET differ  91 slightly from (a) to (b) where Sr=14700 V/m and 15000 V/m respectively.  As explained in section 3.3.1 this is the result  of the Townsend parameter Y varying for both directions. Again, when the bulb was oriented as in Figure 4.2 6 (c), breakdown could not be achieved for field values up to 80000 V/m using the entire variety of unipolar pulses possible with the circuits described.  4.8.4  Wall  Charge  Freezing  Whenever gas discharges occur on the negative field slope of a square pulse (a drop in EA), the bulb must be preserving wall charges previously trapped by the external field since the requirement for breakdown is EI=\EA+EB\:=E0 (where EB is the field produced by the wall charges of the bulb only).  This wall charge  "freezing" is readily observed to occur in the gettered bulbs only if the polarity of the applied pulse is such that electrons can be frozen to the base (Figure 4.5 ) of the bulb when there is a net field component that can drive these electrons towards the base.  When the polarity of the pulse is such that electrons are  accelerated towards the getter, breakdown is no longer observed to occur on the trailing edge on the pulse indicating that electrons can not freeze to the getter side of the bulb.  This is  likely due to the fact that the normal force is eliminated with the presence of the getter hole and electrons are without a surface to freeze to.  The highly mobile electrons then serve to  92 keep the bulb's conductivity high, and the internal field ET never reaches the threshold value ET. Since the voltage requirement to achieve breakdown within the bulb was less than 2 kV, the high voltage spark gap circuit depicted in Figure 4.17 was bypassed, and replaced with the krytron switched circuit shown in Figure 4.27 on the next page (with 3:1 step-up low voltage transformers utilized).  The  krytrons, when used as high speed switches, proved to operate over a wider range of these lower voltages and the magnitude of the voltage pulse could be easily and accurately varied with adjustments of the DC high voltage source (Harrison 0-3000 V D.C. power supply).  To achieve the same voltage range, the spark gaps  would have to be manually adjusted by varying the electrode separation.  3:1  R5 TO DELAY LOW VOLTAGE vW-^3 UNIT TRANSFORMER (Ch. 1) TO UPPER PARALLEL PLATE  K  KA  HIGH D.C. VOLTAGE  4  LOW VOLTAGE A V < I TO DELAY TRANSFORMER UNIT (Ch. 2)  VW  KRYTRON HIGH D . C . ^ ^ CIRCUIT VOLTAGE ^ S ~~ (above)  1  1 O 1  1  Figure 4.27 Device used for producing low voltage three-sided unipolar voltage pulses.  h.  |l'  [up to 2 kV)  94  1Breakdowns RC=.01 86C  1 \ft-~ RC=.0001 sec  V  V  P  1  nwh»  1— t  1  1  t w =.0001 sec  Figure 4.28 Dimensions of the unipolar pulse created by the F i g u r e 4.29 Typical krytron circuit. oscilloscope trace showing the breakdown time delay relative to the pulse.  With the values of R^SOMohms, R^lOOohms,  R5=50ohms,  Rz=llkohms,  R3=1000 ohms,  and the delay between the firing of the  krytrons set at 100 \xs, the field pulse produced is shown in Figure 4.2 8 (where the slope of the top of the pulse is defined by l00i?4C and the tail of the pulse is defined by R^C) . Upon viewing both the voltage pulse across the plates and the output from the photomultiplier simultaneously on a oscilloscope, one can map out the position of the second breakdown (due to wall charge freezing) as a function of the pulse voltage Vp.  By triggering the scope with the pulse and  adding together the two signals (pulse and photomultiplier) , the typical output is shown in Figure 4.2 9 . Of the two light pulses  95 shown, the first pulse remains approximately fixed at this location throughout. The second pulse, however, slides along the falling edge of the field pulse either to the right or to the left depending on how much charge is frozen from the first pulse. Figure 4.30 depicts the graph of V2 versus Vp for the pyrex Ne-Xe gettered bulb (at 5 Torr of pressure). For pulse voltages Vp less than 1200 V, increasing the magnitude of Vp (i.e. the height of the pulse) causes the second pulse to slide up the trailing edge towards the first, decreasing the value of V2 .  Conversely, decreases in Vp moves this light  pulse to the right, away from the first thus increasing the value of V2.  According to our condition for breakdown, these results  indicate that more wall charges are frozen in at higher external field values than at lower ones.  At lower field values (slightly  above ET - with Vp to the left of the minimum in Figure 4.3 0 ) the wall charges in the bulb are weakly frozen and a large drop is required to produce the second breakdown in order to compensate for the charges which have leaked away.  As the  applied field increases, the wall charge freezing strengthens as the electrons work themselves into deeper "traps" with the application of this larger accelerating field.  Less of a field  drop is now required to cause a breakdown on the falling edge. This explains the curve of Figure 4.3 0 up to the minimum (occurring at about Vp=1200V).  The right of the minimum can be  96  Differential Voltage vs. Pulse Voltage (as read off the oscilloscope)  dp LO  I  + >  •p  rH O >  -H Q  550 1 000  1100  1200 1300 1400 1500 Pulse Voltage (Vp) + - 3 %  1600  1700  Figure 4.30 Graph which shows how the wall charge freezing depends on the magnitude of the voltage pulse.  97 explained with two possible theories.  The first explanation is  that at higher values of Vp, very strong discharges occur resulting in the production of an over-abundance of plasma.  The  higher concentration of plasma then remains in the bulb for longer periods of time without recombining.  When the pulse  voltage drops to the trailing edge ( see Figure 4.28 ), this plasma can serve to short circuit the frozen wall charges off of the walls decreasing the internal field that they normally produce.  A greater drop in the applied field is then required to  cause the second breakdown on the trailing edge to occur.  The  second pulse is observed to slide down the edge to the right of Figure 4.29 . This would explain the results which occur to the right of the minimum of Figure 4.3 0 . A second explanation of the curve can be reasoned in terms of the lag times which are created by substantial wall charge freezing.  That is, when the  applied field is held long enough (or raised high enough) a formative time lag ( tlag)  that increases as Vp increases appears.  If one assumes that the initial conditions of sufficient freezing of impurities are met for values of VP>1200V, then the increasing values of V2 to the right of the minimum in Figure 4.30 could be nothing more than the appearance of  tlag.  The time lag serves to delay the occurrence of breakdown and the second pulse slides down to the right on the tail of the pulse shown in Figure 4.2 9 and increases the value of V2 .  98  4.8.5  Gettered  Bulb Time  Lags  When applying these  t1  ^  Tube breakdowns  unipolar square pulses to the  J  gettered bulb and its tube  V P . /l  (optic fibre placed at both positions in Figure 4.25 ), it  /  k.I  I ^  \  V  \  \  \ ~ \  \ ^*^*th*. \  p*  was discovered that the time lag between the start of the pulse and the occurrence of a breakdown was larger for the  Bulb breakdowns  Figure 4.31 Oscilloscope trace showing the relative time lags for gettered breakdown.  breakdown in the bulb itself than it was for the getter tube.  In  other words, the gas (Ne-Xe at 5 Torr) in the getter tube was observed to breakdown before the gas in the bulb itself. Figure 4.31 depicts these relative lag times.  The time lags of  the getter tube tx were determined with the optic fibre at position B while those for the bulb fc2 were determined with the fibre at position A (see Figure 4.25 ). The results for a few values of Vp are shown in Table 4.  For example; when the  voltage change created by the square pulse was 1750 V, the tube broke down with an average lag time of 0.046/??s while the bulb broke down with an average of 0.067ms 0.02ms.  for a difference of  Furthermore, this difference in lag times decreases as  Vp increases.  This decrease in the relative time lags (t2-tx)  99 with increasing Vp can be easily explained if we assume that it is the flux of ions/electrons accelerated from the tube towards the bulb which provides the catalyst needed to initiate the processes leading to the breakdown inside the bulb (an assumption that will be discussed in more detail in section 4.9).  Thus, an  increase in Vp not only decreases the time of formation of the tube breakdown (fcx) , but also increases the acceleration of the ion/electron flux travelling towards the bulb.  Since the time  taken for this catalyzing process decreases, the relative time lag (tg-^) between breakdown of the tube and breakdown of the bulb also decreases.  That is, the initial conditions which are  necessary for breakdown to occur inside the bulb are achieved in less time following the tube's breakdown if Vp is increased. Pulse Voltage  bulb time lag  tube time lag  relative lag  Vp (in V)  t2 (in ms)  t1  fcg-fc! (in ms)  1750  0.068 ± .008  0.046 ± .006  0.02 ± .01  1850  0.048 ± .001  0.0408 ± .0003  0.007 ± .001  1950  0.036 ± .006  0.033 ± .004  0 (within  (in ms)  error bounds) Table 4 Relative time lags of getter tube and bulb discharges with varying unipolar voltage pulses applied.  During the experiments which utilize the above unipolar  100 fields, it was discovered that the gettered bulbs operate with the most predictability provided that the field changes used to cause breakdown occur at times when there is no initial internal wall-charge field across the bulb.  The appearance of a long  time-lag becomes an unwanted and unpredictable characteristic of these gettered bulbs if they are initially exposed to a constant field (not equal to zero) for a time long enough to freeze in an internal wall charge field EB.  Increases in this initial field,  and hence increases in EB, causes both the time lag and the time lag "jitter" (standard deviation of the time lag) to increase substantially. Figure 5.2 depicts a graph of the initial field |HW^fw>nWn'»»"«'W,'YV^'y^Vi jniSl*f,lrf*i  voltage Vz versus the formative time lag tlag (defined as the time taken for  JJ  breakdown to occur after the field changes - see Figure 5.1 ). The net field pulse applied across the  Breakdown-  t=0  lag  Figure 5.1 Typical oscilloscope trace showing the time lag as defined relative to the step pulse.  plates was produced by applying a variable  DC voltage across the  top plate Vconst, and a negative step pulse of constant height Vp (produced with the circuit shown in Figure 4.2 0 ) across the bottom.  The value of Vconst-Vp will be the initial "freezing"  voltage Vx (of Figure 5.1 ) responsible for the varying time  101 lags tlag.  For all the data points VT was held longer than one  minute to ensure some amount of wall charge freezing.  We see  that very small deviations from ^ = 0 can increase the time lag substantially. A possible explanation for these time lags created by the initial external field (due to VT of Figure 5.1 ) is to assume that the application of Vx serves to alter the secondary emission coefficient y by force-freezing any impurities within the bulb to the inner surface of the bulb where secondary electron emission takes place.  As a result of blocking these  emission sites, the formative time lag for a constant field change (in magnitude) will increase with the value of Vx as we must wait for the collision processes at the walls to dislodge the frozen impurities and increase the value of y to the point where the Townsend breakdown criterion (equation (3.2)) is met. Notice that when VT is applied such that electrons can "freeze" to the base of the bulb, smaller increases in VT are needed to achieve longer time lags.  This is consistent with the above  hypothesis that electron freezing is more dominant on this side of the bulb because there is no getter hole.  Initial Voltage vs. Time-lag 0.8  0.6-  Electrons freeze to the getter side of the bulb  O  0.40) tO  0.2-  •P r-\ O  >  to  •H •P •H C  Electrons freeze to the base of the bulb (opposite to the g e t t e r ) .  -4— 4  Figure 5.2 I n i t i a l  + 6 8 10 12 14 t i m e - l a g (msec) +- 20%  16  18  " f r e e z i n g " v o l t a g e v e r s u s f o r m a t i v e time l a g .  103  4.9 Evolution  of Discharges  in an A.C.  Field  for  Gettered  Bulbs  To be able to understand the reasons why breakdown occurs only on one edge of the EA waveforms when gettered bulbs are exposed to AC fields, it is necessary to summarize the unipolar results outlined above as these play an integral role in the model outlined below.  The results are summarized in the order  that they are utilized in the AC theory. 1. Two breakdowns occur in the gettered bulbs at different times due to the lack of symmetry in the bulb's design.  It  was discovered that the getter tube breaks down before the bulb (a fraction of a millisecond) and that the bulb will not breakdown if the tube does not breakdown. 2. It is slightly easier to break down the bulb if the electrons travel from the base of the bulb towards the getter ( Vm  is smaller by a few percent) .  3. Breakdown in the getter tube occurs irrespective of the direction of an alternating electric field applied along its axis (from base to getter tube). 4. Wall charge freezing only occurs when electrons flow to the base of the bulb (electrons will not be frozen in significant amounts at the site of the getter hole).  |+  Upper Plate  +  electrons" plasma In tube  —  Lower Plate  —  b) breakdown in tube occurs first (between A  c) flux of ions near getter hole increases the  and B of AC field)  field inside the bulb.  a) bulb breakdown position with AC field applied to upper plate.  I +  Upper Plate  +  _±]  +  Upper Plate  +  + plasma In tube  —  electrons from tube remains of bulb plasma  remains of bulb plasma  plasma in bulb loosely / frozen ions Lower Plate  — Upper Plate  frozen electrons I—  Lower Plate  —  3  I+  Lower Plate  +  -H  d) bulb breaks down and fills with plasma. e) flux of slow moving ions cannot maintain the  f) when external field changes direction (from  bulb's plasma.  B to C on AC field), flux of electrons can maintain plasma in bulb.  Figure 5.3 Pictorial evolution of gettered bulb breakdown in an AC field. o  105 The processes which allow for discharges to occur on one edge only of an AC field waveform are outlined pictorially (in order of occurrence) in Figure 5.3 on the previous page.  The  first frame (a) indicates when these processes occur with respect to the applied alternating field.  As previously summarized, the  getter tube will achieve the first breakdown immediately following the switching on of the field as depicted in frame (b) . The ions and electrons created as a result of this breakdown then separate towards the cathode and anode respectively.  This flux  of charged particles rushing to the bulb effectively increases the electric field as seen by the bulb as the separation between lower parallel plate and the upper sheet of charge (now defined by the flux of ions/electrons) drops to d1'. Recall that the value of the electric field is proportional to l/d the effective electrode separation).  (where d is  Since the applied field is  connected to the upper plate (as in Figure 4.26 (a)), the first breakdown which occurs within the bulb is on the edge A-B of the field because this direction will possess the smallest value of Vgg (described in point 2 of the summary above) . Therefore, the effective electrode separation is reduced to d' as a result of a flux of ions driving towards the bulb (as in frame (c)) and breakdown is achieved in the bulb (frame (d)). With the applied field still increasing in the same direction (i.e. still along A-B of frame (a)), Ez in the tube once again reaches the threshold value E0 (for the tube) and  106 breakdown occurs for a second time.  The same process occurs as  described above with one major exception - this time the bulb contains remnants of its last discharge (plasma).  Due to the  instantaneous polarity of the applied field, it is once again the ions from the tube which spill into the bulb.  At this time the  tube (filled with plasma) effectively becomes a conductor and the field lines converge into it as shown in frames (c) through (f). The ions from the tube follow these field lines and make collisions with the particles within the bulb.  Because ions are  very massive, the kinetic energy they receive from the field will not be enough to ionize the bulb's gas allowing the conductivity of the bulb to reset to a value close to zero as its ions and electrons recombine before the next discharge.  That is, the  plasma within the bulb from its last breakdown dissipates and the breakdown processes recycle themselves (frames (b) through (d)). Following the last discharge on A-B of frame (a), the applied field then reverses itself (B-D of frame (a)). As outlined in point 3 of the summary above, breakdown will still occur within the getter tube.  Once again this sets up field lines which  converge in at the mouth of the tube.  This time the polarity of  the applied field is such that it will be a flux of electrons from the tube discharge which spill into the bulb along these field lines. Unlike the ions, these electrons are very light and are able to accelerate to very high speeds.  This property allows  them to achieve a kinetic energy which is high enough to ionize the gas inside the bulb.  This new ionization along with the  107 remnants of the bulbs previous discharge maintain the bulb's conductivity at a level (frame (f) above) which is high enough to keep the field inside the bulb close to zero without the need of a full Townsend discharge.  Since the directions of the  avalanches created by the electrons from the tube spread out towards the surface of the bulb, a single intense light pulse (due to an amalgamation of all of the individual avalanches) is not detected by the photomultiplier, and breakdowns are not viewed in the bulb for field changes in this direction (B-D of frame (a)) . Finally, the highly mobile electrons which are created in the processes described above for the field changing from B-D of frame (a), can freeze to the base of the bulb (recalling that charge freezing in the bulb only occurs in this direction as described in point 4 of the summary).  This charge freezing can  explain why we get breakdowns on the next rising edge of the applied field (from D-E of frame (a)) as they serve to satisfy our breakdown requirement that \EI\ = \EA+EB\=E0 by providing a constant (approximately) bulb field EB (see section 3.4).  108  5. Summary and  Conclusion  Knowledge of the External Electric Discharge (EED) switching phenomena can improve discharge devices such as AC plasma display panels (which consist of an array of closely packed discharge tubes) and electric field meters (which count the number of light pulses emitted in the EED bulb per second and relate this to the magnitude of the electric field) require an increased understanding of the switching characteristics of the EED.  These  characteristics need to be examined under a variety of field conditions to be able to comprehend fully why EED devices behave as they do.  Wall charge deposition, varying field frequencies  and waveform shapes, along with gas composition and content all serve to alter the voltage requirements needed to achieve breakdown. Two types of EED bulbs have been studied in detail: spherical ungettered bulbs (filled with neon to 1 Torr of pressure), and spherical gettered bulbs (filled with a Ne-Xe Penning gas mixture to 5 Torr).  The spherical ungettered bulbs  exhibit a spatially isotropic response in the presence of an electric field.  That is, a discharge will occur along the field  lines for all orientations of the bulb relative to the field with approximately the same value of V^.  The gettered bulb response  to the electric field is not isotropic (as one would expect because of its asymmetry).  In fact, two discharges are observed  to occur at separate locations of the bulb provided the applied  109 field is high enough.  The first discharge which occurs is in the  tube connecting the bulb to its getter.  As long as this tube is  oriented in the field so that there is a component of the field which is along the tube axis, it will breakdown (more testing is needed to study the relationship between V^ alignment).  and the tube/field  Immediately following the breakdown in the tube is  the breakdown of the bulb.  Unexpectedly, when placed in an AC  field, the bulb will only breakdown provided the field changes so that any electrons present are accelerated towards the tube. Breakdown will occur in the presence of single, unipolar field pulses along both directions of electron acceleration as long as the pulses are applied at intervals long enough to allow for the plasma in the bulb (from the previous discharge) to recombine. The ungettered bulbs possess a long formative time lag (up to 2 seconds) which disappears as the frequency of the applied field increases.  These lags proved to be a menacing attribute of  experiments with the unipolar field pulses used in the efforts to test for wall charge deposition.  These bulbs behave  inconsistently for applied field frequencies less than 15 Hz and must be operated with frequencies above 3 0 Hz to achieve a stable display.  The gettered bulbs also possess long time lags but only  if they are first exposed to a low strength DC field for a few seconds.  Further tests are required to determine how the time  lags depend on the duration and magnitude of these low strength DC fields.  The unipolar field pulse tests performed for this  thesis to determine wall charge deposition effects were held  110 above V^ for only 100 \is so that the time of field exposure was too short to induce these time lags.  A lag time effect may have  occurred when the applied voltage pulse height was held well above V^ for the gas. Early indications are that these time lags can be forced to occur if high voltages are held across the plates for a short time (\is) , or low voltages are held for a long time (seconds). For either type of bulb there is an electric field threshold ET below which breakdown will not occur.  This threshold (for both  bulb types) depends on the type of gas used along with the type of bulb and its dimensions.  The AC threshold field for the  ungettered bulbs was discovered to increase over time provided that the bulb remains dormant (unexposed to breakdown).  The  ungettered bulb with the higher impurity content (termed "dirty") reaches a higher value of ET over time and both the "clean" and the "dirty" bulbs can be reset to their respective time tD=0 state of dormancy with prolonged exposure to breakdowns. Consistently, the gettered bulbs (assumed to have a very low impurity content) do not exhibit this increase in ET over time. The study of the theory of electrodeless breakdown as developed by Harries and von Engel (based on Townsend's work) had been confined to explanations of why current pulses occur when the gas is exposed to an electric field.  This thesis extends the  fundamental knowledge as to when and why breakdown occurs with the following new experimental observations  Ill 1) The increase of the AC threshold field ET over the time that the ungettered bulbs remain dormant. 2) How to reset this AC threshold field back to time fc^O (before dormancy). 3) The relation between the applied field frequency and time lags (ungettered bulbs). 4) The appearance of long formative time lags due to the presence of an initial DC field (gettered bulbs). 5) The effect that over-voltages (above V^)  have on wall  charge freezing in the gettered bulbs. 6) The two locations of breakdown in gettered bulbs and how one breakdown affects the other. 7) The connection between the orientation of the gettered bulbs with respect to the applied field and the occurrence of breakdown. Despite all of the above extensions, much more work remains to be done on the study of the electrodeless breakdown phenomena. To understand how the concentration of impurities affects ungettered AC breakdown, more bulbs (with the same dimensions, pressure, and vessel material) with varying levels of impurity would be required.  This would enable the mapping of how ET  varies with impurity concentration. that ET(tD)  Early indications suggests  rises to a much higher value for bulbs of high  impurity content.  To test whether or not this increase in ET  over time is due to adsorption of the impurity onto the vessel  112 walls, these bulbs could be constructed with interior metal electrodes across one diameter.  The value of ET(tD)  could then  be tested for two orientations of the bulb with respect to the applied field.  One orientation would be with the direction of  the electric field normal to the metal electrodes so that current flows across the electrode gap.  In this position ET should not  be affected significantly by adsorption onto the electrodes since metal maintains an extremely high value of y  (secondary emission  coefficient) which would be altered very little by adsorption. The other orientation would be with the electric field parallel to the metal electrodes so that breakdown (a pulse instead of a current) will occur across the insulating vessel as in normal EED devices.  The magnitude of ET for this position should vary as  usual over dormancy times since adsorption can affect the value of y f ° r the vessel walls. All of the experiments involving time lags could be improved upon by introducing computer controlled switching of the pulses applied to the parallel plates.  A program could be written so  that identical pulses are sent to the plates to achieve breakdown at specified intervals (one minute for example) to ensure equal "freezing" conditions (for each successive breakdown).  In this  way the time lag of breakdown for any particular voltage pulse could be graphed with its appropriate mean and standard deviation (jitter).  This would smooth out some of the rougher time lag  graphs and make fitting to a mathematical model more reliable.  . 113 Finally, as mentioned above, further tests on the orientation of the gettered bulbs relative to the applied field are required.  A graph of V^ versus the angle between the getter  tube and the applied field direction would be useful for engineering concerns.  114  Appendix  Ionization  Coefficients  A.l  Electrons1  for  To be able to utilize Townsend's breakdown criterion (equation (3.2)) for starting potential computations, we need to understand how a depends on the electric field. An electron with a small initial energy is assumed to begin with a mean free path of A .  To ionize upon collision it must  gain enough energy from the field such that eEL^.eVi the charge on an electron).  (where e is  Therefore, the distance that the  electron must move in the direction of the field is given by L=V±/E. The probability (Pr)  that the distance travelled is larger  than L is given by the statistical distribution of free paths pr  =  e-L/X  The probability Pr of travelling a path length which is longer than L is also equal to the number of ionizing collisions per mean free path in the direction of the field, i.e. ctL. Therefore, Pr = e'L/A = aA  With L-VjE  and \=kx/p  we can w r i t e  *pp 155-158 of von Engel, A.,  "Ionized  Gases" (see  references)  115 a./p=Ae-B/(E'p)  Where A=l/Xx,  B=Vi/X1,  A. (1.1)  and Xx is A at p=l rorr.  Table 511 gives several typical values for the constants A and B.  The constant A is the saturation value that a/p tends  to at large E/p whereas B is proportional to the effective ionization potential for the process which includes excitation losses.  J  refer references)  to Table 7.1, p 157, in von Engel,  "Ionized  Gases"  (see  116 Gas  A  1  cm Toir  B  V  cm Torr  "2  5  130  ^2  12  342  Air  15  365  C02  20  466  H20  13  290  A  12  180  He  3  34  Hg  20  370  Table 5 Values of the starting potential coefficients, A and B, for various gases.  References l.Tesla, N., Elect.  Engr,  Lond.,  7 (1891) 549  2.Wiedemann, E. and Ebert, H., Wied. 3.Thomson, J.J., Phil.  Mag.,  Ann.,  1 (1893) 549  32 (1891) 321, 445  4.Suzuki, M. , Mikoshiba, S., Shinada, S., Kohgami, A., Curzon, F.L., Use of Ne-Xe Mixtures in a Full-Colour Townsend Discharge Panel for TV Display, Central Research Lab., Hitachi Ltd., Kokubunji, 1989. 5.Francis, G., Ionization (London), pp. 94-172.  Phenomena  in  6.Harries, W.L. and von Engel, A., Proc. 7.Harries, W.L., Proc.  IEEE.  IIA,  Gases,  Butterworths Publ.  Phys.  Soc.  B64, 915 (1951)  100, pp 132-137 (1953)  8.Harries, W.L. and von Engel, A., Proc. 490 (1954).  Roy.  Soc.  9.von Engel, A., Electric Plasmas: Their Nature and Francis Ltd. (London and New York), 1983.  (London) A-222, and Uses,  Taylor  lO.Dutton, J., in Electrical Breakdown of Gases, edited by J.M. Meek and J.D. Craggs (Wiley, New York, 1978), pp 209-318 11.Friedmann, D.E., Measurement of Low Frequency Electric Fields Using Electrodeless Breakdown of Gases, UBC masters thesis (Jan. 1983) 12.von Engel, A., Ionized  Gases,  pp 169-173, Oxford, 1955.  13.Friedmann D.E., Curzon, F.L., Electrodeless Breakdown in Elliptically Polarized Fields at 60 Hz., Can. Journal of Phys., Vol 61, No. 9, (1983) 14.Curzon, F.L., Mikoshiba, S., External Electrode Glow Discharges,  Unstable Repetition EID87, (1991)  15.Curzon, F.L., Parfeniuk, D.A., Can. J. 16.Holloway, D.G., The Physical Properties London, England (1973), Ch. 3, pp 50-66.  Phys.  61, of  Rate  of  1260 (1983).  Glass,  Wykeham,  17.Morgan, C.G., Irradiation and Time Lags, Ch. 7, pp 655-688 of "Electrical Breakdown of Gases" by J.M. Meek and J.D. Craggs, Wiley Publ. (1978)  118 18.Dutton, J., Spark Breakdown in Uniform Fields, Ch. 3, pp 276-290 of "Electrical Breakdown of Gases" by J.M. Meek and J.D. Craggs, Wiley Publ., (1978). 19.von Engel, A., Ionized  Gases,  Oxford (1955), pp. 46-65  20.Blair, D.T.A., Breakdown Voltage Characteristics, Ch. 6, in "Electrical Breakdown of Gases", edited by J.M. Meek and J.D. Craggs, Wiley Publ. (1978) 21.Rees, J.A., Fundamental Processes in the Electrical Breakdown of Gases, Ch. 1, of "Electrical Breakdown of Gases" edited by J.M. Meek and L.D. Craggs, Wiley Publ. (1978) 22 .Friedmann, D.E., Measurement of Low Frequency Fields Using Electrodeless Breakdown of Gases, p. 86, M.A.Sc thesis, U.B.C, (1983)  


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