EEE 483 High Voltage Engineering

Electrical Breakdown in Gases

A H Chowdhury, PhD
Professor, EEE, BUET
January 2020

1
 Lecture Content
▪ Classical gas laws
▪ Velocity distribution of a swarm of molecules

▪ Free path of molecules and electrons

▪ Distribution of free paths

▪ Collision-energy transfer

2
 Lecture Content
▪ Ionization and decay processes
▪ Townsend first ionization coefficient

▪ Photoionization

▪ Ionization by interaction of metastables with atoms

▪ Thermal ionization

▪ Deionization by recombination

▪ Deionization by attachment–negative ion formation

▪ Mobility of gaseous ions and deionization by diffusion

▪ Relation between diffusion and mobility

3
 Lecture Content
▪ Cathode processes – secondary effects
▪ Photoelectric emission

▪ Electron emission by positive ion and excited atom impact

▪ Thermionic emission

▪ Field emission

▪ Townsend second ionization coefficient

▪ Secondary electron emission by photon impact

4
 Lecture Content
▪ Transition from non-self-sustained discharges to breakdown – Townsend mechanism

▪ Streamer or ‘Kanal’ mechanism of spark

▪ Sparking voltage – Paschen’s law

▪ Penning effect

▪ Breakdown field strength

▪ Breakdown in non-uniform fields

▪ Effect of electron attachment on the breakdown criteria

5
 Lecture Content
▪ Partial breakdown, corona discharges
▪ Positive or anode coronas

▪ Negative or cathode corona

▪ Polarity effect – influence of space charge

▪ Surge breakdown voltage–time lag

▪ Breakdown under impulse voltages

▪ Volt–time characteristics

▪ Experimental studies of time lags

6
Classical Gas Laws

7
 Section content
▪ Classical gas laws
▪ Kinetic theory of gases
▪ Velocity distribution of a swarm of molecules
▪ Free path of molecules and electrons
▪ Distribution of free paths
▪ Collision-energy transfer

8
 Classical Gas Laws
Boyle and Mariotte’s Law:
For a given amount of enclosed gas at a constant temperature
product of pressure p and volume V is constant

For volume V0 at temperature T0:

Gay–Lussac’s law:
For a given amount of enclosed gas at a constant pressure
volumes V and V0 are related to their absolute temperatures T Universal gas constant (R)
and T0 (in K)

• When temperatures are expressed in degrees Celsius

9
 Classical Gas Laws

R = 8.314 joules/°K mol

For n number of moles, i.e. This equation describes state of an ideal

mass m of the gas divided gas, since R is assumed a constant
by its mol-mass independent of nature of gas

In terms of gas density N Putting N = NA, where

in volume V containing N1 NA = 6.02 x 1023 molecules/mole
molecules NA - Avogadro’s number

k = R/NA – universal Boltzmann’s constant (=1.3804 x 10 –23 joules/°K)

N – number of molecules in the gas
10
 Classical Gas Laws
If n gases with initial volumes V1, V2 ….. Vn are combined at same temperature and pressure, then new
volume will be given by

Law of partial pressures

p1, p2, . . . , pn - partial
pressures of gases 1, 2, . . . ,n

11
 Kinetic Theory of Gases
• Kinetic theory of gases developed by Maxwell (mid nineteenth century)

• Fundamental equation for kinetic theory of gas derived with following assumed conditions:

1. Gas consists of molecules of same mass which are assumed spheres

2. Molecules are in continuous random motion

3. Collisions are elastic – simple mechanical

4. Mean distance between molecules is much greater than their diameter

5. Forces between molecules and the walls of the container are negligible

12
 Kinetic Theory of Gases
Cubical container of side l = 1m, • A molecule of mass m moving in x-direction with velocity ux strikes
contains N1 molecules, each of
mass m and r.m.s. velocity u wall of container plane YZ and rebounds with velocity –ux

• Change in momentum,

• Particle impacts one specific side wall once every 2l/ux sec.

Number of collisions per

second with right-hand
wall, ux/2l, therefore,

• But the same molecule will experience the same change in

momentum at opposite wall, hence,
Velocity components, ux, uy, uz
where u2 = ux2 x uy2 x uz2
13
 Kinetic Theory of Gases
For three-dimensional cube with total change in
momentum per second per molecule (which is the
force), force per particle,
Kinetic energy for a particle

For N1 particles energy due to different velocities u

of particles will become mean energy, and therefore

Force leads to pressure p, taking into

account total area of cube A = 6l2

14
 Kinetic Theory of Gases

The two equations are identical

for constant temperature

Hence, mean energy per molecule:

15
 Velocity Distribution of a Swarm of Molecules
• Distribution of molecular velocities depends on both temperature and molecular weight of the gas
• Most probable velocity is neither average nor r.m.s. velocity of all molecules
• Velocity u of gas molecules or particles has a statistical distribution and follows Boltzmann–Maxwell
distribution given by,
Where, up – most probable velocity
dNu/N – relative number of particles whose instantaneous
velocities lie in the range u/up and (u + du)/up

Let,

16
 Velocity Distribution of a Swarm of Molecules
• Average velocity u is obtained from integrating ur from
0 to 

• The function is asymmetrical about the

most probable velocity up
• A greater number of particles has a
velocity higher than up
17
 Velocity Distribution of a Swarm of Molecules
r.m.s. or effective value of velocity 〉〉

Mean kinetic energy of particle relates its

effective velocity to temperature

18
 Velocity Distribution of a Swarm of Molecules
• The foregoing considerations apply only when molecules or particles remain in thermal equilibrium,
and in absence of particle acceleration by external fields, diffusion, etc.

If gas contains electrons or ions or other atoms that are at same temperature, average particle
energy of such mixture is,

where, m, me, mi – respective masses of gas

molecules, electrons, ions
ueff, ueeff , uieff – corresponding velocities

Mean molecular
velocities at 20C
and 760 Torr

19
 Free Path of Molecules and Electrons
Free path  - the distance molecules or particles travel between collisions
• Free path is a random quantity and its mean value depends upon
concentration of particles or density of the gas

Model for determining free paths

• Consider an assembly of stationary molecules of radius
r1, and a moving layer of smaller particles of radius r2
• As smaller particles move, their density will decrease
due to scattering caused by collisions with gas molecules
• Assume that moving particles and molecules behave as
solid spheres → then a collision will occur every time
centres of two particles come within a distance r1 + r2
20
 Free Path of Molecules and Electrons
• Area for collision presented by a molecule is then (r1 + r2)2 and in a unit volume it is N(r1 + r2)2
N(r1 + r2)2 – effective area for interception, where N = number of particles per unit volume of gas

n(x) – number of particles that survived

distance x, then decrease in moving
particles due to scattering in layer dx is

Assuming the number of particles

entering (at x = 0) is n0, then integrating

Consider a layer of thickness dx,

distant x from origin
21
 Free Path of Molecules and Electrons
Probability of free path of length x is equal to probability of collisions between x and x + dx

Number of particles that

survived distance x

Differentiating

Mean free path

(r1 + r2) has dimensions of area;

→ Called cross-section for interception or collision
cross-section ()

22
 Free Path of Molecules and Electrons
Q represents effective crosssection presented by molecules or particles
Let,
in unit volume of gas for all collisions for density of N molecules/volume

• Only a fraction Pi of collisions between incoming particles and gas particles leads to ionization, hence
Pi is the probability of ionization
• If only ionizing collisions are counted, molecules present an effective area of only PiQ = Qi
→ Qi – effective cross-section for ionization
→ Similarly for other processes, excitation (Qe), photoionization (Qph), attachment (Qa) etc.,
including elastic collisions can be taken into account

23
 Free Path of Molecules and Electrons
Mean free path

m1, m2 – mass of
It was assumed that the molecules To correct this the collisional cross-
each gas
of gas 2 had no thermal velocity section must be multiplied by a factor component

In a gas mixture collisional cross-section

of particles of gas-1 (m1, r1,N1) becomes Thus mean free
equal to sum of all collisional cross- path of particles
sections of other particles of types of of gas-1
gas (m2,m3, . . . , r2, r3, . . . , N2,N3, . . .)

For an atom in its For an electron in

own gas r1 = r2 = r; a gas r1 << r2 and
m1 = m2, then m1 << m2 24
Free Path of Molecules and Electrons
Hence, mean free path is directly
proportional to temperature and
inversely as gas pressure

25
Free Path of Molecules and Electrons

Mean free paths measured at 15°C and 760 Torr

26
 Distribution of Free Paths
• Mean free path for the system studied

N - gas density
r1 and r2 - radii of two types of particles

Free path is a random quantity and have a distribution about a mean value

→ Distribution function of free paths –

Distribution of free paths
(Clausius distribution)

n(x) - number of molecules reaching a distance x without collision

dn - number of molecules colliding thereafter within a distance dx
n0 - total number of molecules at x = 0 [initial condition]]
27
 Distribution of Free Paths
• Percentage of molecules that survive collisions is
only 37%

• Absorption or decay of particles along path x in

terms of collision cross-sections

where,  may include

photoabsorption, attachment, etc.

Distribution of free paths

28
 Collision-energy Transfer
Two of types collisions between gas incidence angle scattering angle
particles
i. Elastic or simple mechanical collisions
– energy exchange always kinetic

ii. Inelastic collisions – some of the

kinetic energy of colliding particles
transferred into potential energy of
Elastic collision between two particles of masses m and M
struck particle or vice versa
Before collision velocities
Examples – excitation, ionization, of large mass M → zero i.e. at rest
attachment, etc. of smaller particle m → u0
After collision velocities (along line of centres as shown)
of mass M → V
of mass m → u1 29
 Collision-energy Transfer
Kinetic collision, so (2)
equations for conservation (3)
of momentum and energy
(4)

Squaring eqns (2) and (3)

and adding and
combining with eqn (4)

Fractional energy loss by incoming particle

during a collision at an angle 
Rearranging eqn (4) and
combining with eqn (1) (5)
(1)

30
 Collision-energy Transfer
P() – probability of a collision at an angle of incidence between  and  + d

(r1 + r2)2 – total area presented for collision

• Probability of a collision taking place between  and  + d is the ratio of projected

area to the whole area

Mean fractional loss of energy per collision

(6)
allowing for collisions at all angles

Using eqns (5) and (6)

(7)

31
 Collision-energy Transfer
Mean fractional loss of energy per collision (7)
allowing for collisions at all angles

When incoming particle is an ion of the same Indicates a high rate of energy loss
mass as the struck particle i.e. m = M in each elastic collision

Average fraction of energy lost by

Incoming particle is an electron, then m << M an electron in an elastic collision
very small

Average fractional energy loss per collision for electrons colliding

• with He gas atoms = 2.7 x 10-4 Thus electrons will not readily lose energy in

• with Argon gas atoms = 2.7 x 10-5 elastic collisions whereas ions will

32
 Collision-energy Transfer
When part of kinetic energy of incoming particle is converted into potential energy of struck particle

Wp – increase in potential energy of

Applying laws of energy and
particle of mass M initially at rest
momentum conservation

Rearranging

(8)

For maximum energy transfer, considering constant

Equation shows that potential energy gained from
kinetic energy of incoming particles,
incident particle reaches a maximum value when ratio

(9) of its final to initial velocity equals ratio of its mass to

sum of masses of individual particles 33
 Collision-energy Transfer
Condition for maximum energy transfer
(9)
(considering constant kinetic energy of incoming particles)

1. When colliding particles are identical → maximum kinetic to potential energy transfer occurs when u1 = u0/2

2. When colliding particle is an electron of maximum energy transfer corresponds to u1 = (m/M)u0

mass m << M → → indicates that new velocity u1 becomes only a small fraction of
original velocity u0
3. When target particle initially at rest the maximum amount of potential energy gained
expression obtained by inserting value of
velocity u1 from eqn (9) into eqn (8)

For an electron m << M • Almost all of an electron’s KE converted into PE → thus electrons are
good ionizers of gas, while ions are not

• To cause ionization incoming electron must have a KE of at least

34
Vi – ionization potential of atom or molecule
Ionization and Decay Processes

35
 Section content
▪ Townsend first ionization coefficient

▪ Photoionization

▪ Ionization by interaction of metastables with atoms

▪ Thermal ionization

▪ Deionization by recombination

▪ Deionization by attachment–negative ion formation

▪ Mobility of gaseous ions and deionization by diffusion

▪ Relation between diffusion and mobility

36
 Ionization and Decay Processes
At normal temperature and pressure gases are excellent insulators

Conduction in air at low field ~ 10-16 to 10-17 A/cm2 This current results from cosmic radiations and
radioactive substances present in earth and atmosphere

• At higher fields charged particles may gain sufficient energy between collisions to cause ionization on
impact with neutral molecules
• Electrons on average lose little energy in elastic collisions and readily build up their KE which may be
supplied by an external source, e.g. an applied field
During inelastic collisions a large fraction of
electron KE transferred into PE, causing, e.g., Ionization by electron impact, for higher field strength,
most important process leading to breakdown of gases
ionization of struck molecule

• Effectiveness of ionization by electron impact depends upon energy that an electron can gain along
mean free path in the direction of field
37
 Ionization and Decay Processes
If is mean free path in field direction of strength Proportional to E/p since
E then average energy gained over a distance is

To cause ionization W ≥ eVi However,

i.e., impact energy (W) must be at • Electrons with energy lower than eVi may excite particles and excited
least equal to ionization energy of particles on collision with electrons of low energy may become ionized
molecule (eVi) • Not all electrons having gained energy W ≥ eVi upon collision will cause
ionization
Vi – ionization potential

• This simple model is not applicable for quantitative calculations, because ionization by collision, is a
probability phenomenon

→ it is generally expressed in terms of cross-section for ionization defined as Pi = i

where, Pi – probability of ionization on impact
 – molecular or atomic cross-sectional area for interception defined earlier 38
 Ionization and Decay Processes
Variation of ionization cross-sections for O2, N2, H2 with electron energy
• i measured using monoenergetic electron beams of
different energy
• Cross-section strongly dependent upon electron
energy
• At energies below ionization potential collision may
lead to excitation of struck atom or molecule which on
collision with another slow moving electron may
become ionized
• This process becomes significant only when densities
of electrons are high
• Very fast moving electrons may pass near an atom
without ejecting an electron from it
• For every gas there exists an optimum electron energy
range which gives a maximum ionization probability 39
 Townsend* First Ionization Coefficient

Townsend studied variation of gas current measured

• State of equilibrium – in absence of electric field rate of
between two parallel plate electrodes as a function electron and positive ion generation in an ordinary gas is
of applied voltage counterbalanced by decay processes and a state of
equilibrium exists
→ a sufficiently high field upsets state of equilibrium
Observations

• Initially current increased proportionately with applied voltage

and then remained nearly constant at i0 which corresponded
to background current (saturation current), or if cathode was
irradiated with a u.v. light, i0 gave emitted photocurrent
• at higher voltages current increased above i0 at an exponential
rate
40
*John Sealy Edward Townsend (1868-1957), Irish mathematical physicist, Professor of physics at Oxford University
 Townsend First Ionization Coefficient (1910)
• As field increases, electrons leaving cathode are
accelerated more and more between collisions until they
Increase in current beyond V2 gain enough energy to cause ionization on collision with
Townsend ascribed to ionization
of gas by electron collision gas molecules or atoms

• To explain this current increase Townsend introduced a

quantity – Townsend’s first ionization coefficient () →
defined as number of electrons produced by an electron
per unit length of path in the direction of field

41
 Townsend First Ionization Coefficient
In terms of current
where, I0 = current leaving cathode
ed – called electron avalanche
• It represents number of electrons produced by one
electron in travelling from cathode to anode
Schematic representation of electron multiplication
(a) gap arrangement, (b) electron avalanche Increase of current
(avalanche growth) as
n = number of electrons at a distance x from cathode in shown in diagram k = number of ionizing steps = x/i
field direction
n0 = number of primary electrons generated at cathode
Transition for leads to the
Increase in electrons dn in infinitely small expression ex
additional distance dx given by values of
Integration over distance d
 describes the rate of ionization by electron collision
42
from cathode to anode gives
 Townsend First Ionization Coefficient
 cannot be readily calculated i determined for monoenergetic electrons and calculation of
from measured cross section for  from value of i is only possible when the electron energy
ionization (i) distribution in the gas is known

For ‘swarm’ conditions Raether N – concentration, molecules/atoms

f(vth) – distribution of velocities of electrons
(1941) derived a relationship ue– drift velocity of electrons in field direction
between  and i

• At a constant temperature for a given gas energy distribution

Therefore,
W depends only on E/p
• Also for a given energy distribution probability of an
ionization occurring depends on gas density or pressure

43
 Townsend First Ionization Coefficient
This eqn describes a general dependence of /p upon E/p

Derivation of the expression for simple gases using Clausius distribution

Distribution of free paths • Assumed that distribution is not altered by additional velocity of
(Clausius distribution)
electrons in field direction

• All electrons which acquire energy W ≥ eVi (Vi - ionization

potential), will ionize the gas

• These electrons have travelled a distance x, fraction of electrons

with paths exceeding a given value x
Number of molecules reaching a distance
x without collision Therefore, only with a very small probability electrons can
gain high energies if they reach long distances
44
 Townsend First Ionization Coefficient
Fraction of electrons with paths exceeding a given value x

Number of successful collisions – ionization coefficient  –

i = x
related to this distribution, and directly proportional to ionizing free path
decay of collisions in the intervals between x and x + dx

• It is assumed that E = , i.e.

Hence,
velocity distribution is not altered
by additional velocity of electrons
in the field direction
• In reality there is a difference v – electron drift velocity
between  and E
45
 Townsend First Ionization Coefficient
Rate of ionization by Corrected for field
electron collision drift velocity

Collision using the eqn. i – true cross-section for ionization

cross-section N – gas density

where,

• Even with constant values of A and B, the eqn determines ionization process within certain ranges of E/p
46
• Therefore, for various gases ‘constants’ A and B have been determined experimentally
 Townsend First Ionization Coefficient
• Constants A and B derived from kinetic theory, rarely agree with experimentally determined values,
because of the assumptions made in derivations

• Assumed every electron whose energy exceeds eVi will automatically lead to ionization

• In reality probability of ionization for electrons with energy just above ionization threshold is small
and it rises slowly to a maximum value of about 0.5 at 4 to 6 times the ionization energy, beyond that
it decreases

• Assumed that mean free path independent of electron energy; not necessarily true

• A rigorous treatment would require taking account of the dependence of ionization cross-section
upon electron energy

47
 Photoionization
• Electrons of energy lower than ionization energy eVi may on collision excite gas atoms to higher states

A* – atom in an excited state

Photoionization

On recovering from excited state in some 10-7–10-10 sec,

the atom radiates a quantum of energy of photon (hv)
A – a neutral atom or molecule in the gas
which in turn may ionize another atom whose ionization
potential energy is equal to or less than photon energy hv – photon energy

For ionization to occur hv ≥ eVi or c0 –velocity of light

photon wavelength  ≤ c0h/eVi h – Planck’s constant

• Therefore, only very short wavelength light quanta can cause photoionization of gas
• Shortest wavelength radiated from a u.v. light with quartz envelope is 145 nm, which
corresponds to eVi = 8.5 eV, lower than ionization potential of most gases 48
 Photoionization
• Probability of photon ionizing a gas or molecule is maximum when (hv – eVi) small (0.1–1 eV)

• Photoionization is a secondary ionizationprocess and may be acting in Townsend breakdown

mechanism and is essential in streamer breakdown mechanism and in some corona

discharges

• If photon energy is less than eVi it may still be absorbed by atom and raise the atom to a

higher energy level → this process is known as photoexcitation

49
Ionization by Interaction of Metastables with Atoms
• In certain elements lifetime in some of the excited electronic states extends to seconds →
metastable states → Atoms in these states referred to as metastables (Am)

• Metastables have a relatively high potential energy → able to ionize neutral particles

If Vm, the energy of a metastable Am, exceeds Vi, ionization

of another atom B, then on collision ionization may result

50
Ionization by Interaction of Metastables with Atoms
• Photon released in this reaction is of too low
energy to cause ionization in pure gas, but it
may release electrons from cathode

• Ionization by metastable interactions comes into operation long after excitation

• It has been shown that these reactions are responsible for long time lags observed in some gases

• It is effective in gas mixtures

51
 Thermal Ionization
Thermal ionization – ionizing actions • If a gas is heated to sufficiently high temperature many of the
of molecular collisions, radiation and gas atoms or molecules acquire sufficiently high velocity to
electron collisions occurring in gases
at high temperature cause ionization on collision with other atoms or molecules
• Principal source of ionization in flames and high-pressure arcs

• We have to consider recombination • Under thermodynamic equilibrium conditions rate of new

between positive ions and electrons ion formation must be equal to rate of recombination

Under these assumptions, degree of ionization in terms of gas pressure and absolute temperature,
p – pressure in torr
Wi – ionization energy of the gas
k – Boltzmann’s constant
ni – number of ionized particles of total n particles

• Strong dependence of  on temperature → degree of ionization is negligible at room temperature

52
• Thermal ionization is significant for temperatures above 1000 K
 Deionization by Recombination
• Recombination takes place whenever there are positively and negatively charged particles present

Potential energy and relative kinetic

energy of recombining electron–ion is
released as quantum of radiation

Alternatively a third body C – another

heavy particle or electron – may absorb
excess energy released in recombination

• At high pressures, ion–ion Rate of recombination for equal concentrations of

recombination takes place positive ions n+ and negative ions n−
• Rate of recombination in either
case directly proportional to
concentration of both +ve ions
and -ve ions  − a constant, recombination rate coefficient
53
 Deionization by Recombination
Since n+ ≈ n−= ni and assuming at time d𝑛𝑖
t = 0: ni = ni0 and at time t: ni = ni(t) = −𝛽𝑛𝑖2
d𝑡

Integrating,

• Variation of recombination rate coefficient

Half-time duration, during 𝛽 with pressure in air is shown in figure
which concentration of • Recombination process particularly
ions has decreased to half important at high pressures for which
its original value diffusion is relatively unimportant
54
 Deionization by Attachment – Negative Ion Formation
Electron affinity

Certain atoms/molecules in their gaseous state These gases lack one Examples
or two electrons in Halogens (F, Cl, Br, I, At); 1 electron
can readily acquire a free electron to form a missing in outer shell
stable negative ion → electronegative gases their outer shell
O, S, Se; 2 elec. missing in outer shell

• For a negative ion to remain stable for some time, total Electron affinities of some elements

energy must be lower than that of an atom in ground state

• Change in energy that occurs when an electron is added to
a gaseous atom or molecule is called electron affinity of
the atom (Wa)
• This energy is released as a quantum or kinetic energy
upon attachment
55
 Deionization by Attachment – Negative Ion Formation

How
deionization by
attachment
(electronegative gases)
helps power
engineering

Land area requirements for a conventional open terminal 132 kV double switchyard
arrangement and replacement indoor GIS housing to the top right-hand corner 56
 Deionization by Attachment – Negative Ion Formation
Processes of negative ion formation

1) Radiative attachment 2) Third body collision 3) Dissociative attachment

• Simplest mechanism - excess attachment • Predominant in molecular
energy upon attachment released • Excess energy upon gases
as quantum attachment can be • Excess energy is used to
• Reversible process, i.e., captured acquired as kinetic energy separate molecule into a
electron can be released by of a third body upon neutral particle and an atomic
absorption of a photon → collision negative ion
photodetachment

57
 Deionization by Attachment – Negative Ion Formation
4) Two stages of the process:
In dissociative attachment process in intermediate stage
molecular ion is at a higher potential level and upon
collision with a different particle this excitation energy may
be lost to colliding particle

Other processes of negative ion formation include splitting

of a molecule into positive and negative ions upon impact of
an electron without attaching the electron

A charge transfer following heavy particle collision, yielding

an ion pair according to the eqn.

• All the above electron attachment processes are reversible,

58
leading to electron detachment
 Deionization by Attachment – Negative Ion Formation

• The process of electron attachment may be expressed by cross-section for negative ion formation
𝜎A in an analogous way to ionization by electron impact

Variation of attachment cross-section with electron energy Variation of electron attachment cross-section
in SF6 - 1. Radiative attachment 2. Dissociative attachment with electron energy in CO2 (both peaks 0-)
59
 Deionization by Attachment – Negative Ion Formation
Current due to increased ionization because of
collision [electron multiplication in a gas]

Cumulatively the process of electron attachment describing removal of electrons by attachment

from ionized gas may be expressed by a relation analogous to the above expression

𝜂 – attachment coefficient as number of Loss of electron current in a distance dx due to

attachment
attachments produced in a path of a single
electron travelling a distance of 1 cm in the
direction of field [analogous to first For a gap of length d with electron current I0
Townsend ionization coefficient 𝛼 ] starting at cathode

60
 Deionization by Attachment – Negative Ion Formation
Considering simultaneous operation of electron multiplication by electron, collision and electron loss
by attachment and neglecting other processes…

• number of electrons produced by collision in distance dx → x - distance from cathode

• number of electrons lost in dx by attachment →

• number of electrons still free →

Integrating from x = 0 to x with n0

• number of electrons at any point in gap
electrons starting from cathode

Steady state current under such conditions will have two

components resulting from
(1) flow of electrons, (2) flow of negative ions 61
 Deionization by Attachment – Negative Ion Formation

In absence of attachment (𝜂 = 0) the

expression reduces to the form

It is convenient to represent the observed ionization coefficient

by a single coefficient called effective ionization coefficient

• Electron attachment reduces electron amplification in a gas → gases with a high attachment
coefficient such as sulphur hexafluoride or freon have much higher dielectric strength than air or
nitrogen

• These gases widely used as insulating medium in compact h.v. apparatus including totally enclosed
substations and h.v. cables
62
 Mobility of Gaseous Ions and Deionization by Diffusion

• An electric field forces charged particles in a gas to drift

• Drift velocity varies directly with field and inversely with density of gas through which particle the
moves

Mobility (K) is defined as drift velocity u – average drift velocity in field direction
E – electric field strength
component in field direction of unit strength

• Mobility is mainly a characteristic of gas through which the ion moves

→ is independent of E/p over a wide range of E/p so long as velocity gained by ion from field
considerably less than average thermal velocity of gas through which the ion moves

63
 Mobility of Gaseous Ions and Deionization by Diffusion

Mobility of ions in a gas under an influence of electric field (in region of low values of E/p)
• Assuming ions are in thermal equilibrium with gas molecules, their drift velocity small compared to
thermal velocity
ionic mean free path
If time interval between two successive
mean thermal velocity of ion
collisions (𝜏), is independent of E, then

During time 𝜏 ion is accelerated by field E with acceleration, a = eE/m m – ionic mass; e – charge of ion

Drift velocity Mobility

In time 𝜏 ion moves
a distance, s

In deriving this expression we assumed mean free paths unaffected by drift motion,
i.e., all ions moving with same random velocity and have same mean free
64 path
 Mobility of Gaseous Ions and Deionization by Diffusion
To take statistical distribution of mean free paths into account

Then, distance between two collisions

Assume ions moving with an average velocity
c in zig-zag projections of lengths which are a – acceleration caused by field E
distributed about mean free path x – total distance travelled between collisions

Average value of s̅
(obtained by averaging x2 over the
distribution of free paths)

If mean free time, Drift velocity Mobility

then,
• When distribution of free paths taken
into account mobility is increased by a
65
factor of 2
 Mobility of Gaseous Ions and Deionization by Diffusion

Diffusion
• In electrical discharges ions move from regions of • Diffusion causes deionizing effect in regions

higher concentration to regions of lower concentration of higher concentrations and an ionizing

effect in regions of lower concentrations
→ equilibrium is achieved by diffusion

• Flow of particles along ion concentration gradient constitutes a drift velocity similar to that of
charged particles in an electric field

• Both diffusion and mobility result in mass motion

▪ This mass motion is described by drift velocity caused in one case by net effect of unbalanced collision
forces (ion concentration gradient) and in other case by the electric field

66
 Mobility of Gaseous Ions and Deionization by Diffusion

• Consider a container with gas in which concentration varies in x-direction

 Take a layer of unit area (A) and  Number of particles crossing area
thickness dx placed perpendicularly (A) proportional to ion concentration
to direction x gradient dn/dx

Flow of particles or flux in x-direction

• ‘– ve’ sign indicates that n increases and rate of flow ()
must decrease in direction of flow
• D – diffusion coefficient, a constant

67
Relation Between Diffusion and Mobility

68
Cathode Processes – Secondary Effects

69
 Section Content
▪ Photoelectric emission

▪ Electron emission by positive ion and excited atom impact

▪ Thermionic emission

▪ Field emission

▪ Townsend second ionization coefficient

▪ Secondary electron emission by photon impact

70
 Cathode Processes – Secondary Effects
• Cathode play very important role in gas discharges → it supplies electrons for initiation, for
sustaining and for completion of a discharge

• Under normal conditions electrons are prevented from leaving solid electrode by electrostatic forces
between electrons and ions in lattice

• Energy required to remove an electron from a Fermi level → work function Wa characteristic of a
given material
• Numbers of ways to supply required energy to release electrons

• Photoelectric emission
• Electron emission by positive ion and excited atom impact
• Thermionic emission
71
• Field emission
 Photoelectric Emission
• Photons incident upon cathode surface whose energy exceeds work function (h𝜈 > Wa) may eject
electrons from surface
▪ For most metals critical frequency 𝜈0 lies in u.v. range

When photon energy > work function, excess

energy may be transferred to electron kinetic
m – electron mass ue – electron velocity
energy according to Einstein relation h𝜈0 – critical energy required to remove electron = work function, Wa

Work function for typical elements

• Work function is sensitive to

contamination which is
indicated by spread in
measured values 72
Electron Emission by Positive Ion and Excited Atom Impact

• Electrons may be emitted from metal surfaces by bombardment of positive ions or metastable atoms
▪ To cause a secondary emission of an electron impinging ion must release two electrons

Minimum energy required for positive ion electron emission WK + Wp = 2Wa

(since ion is neutralized by one electron and other electron WK and Wp – kinetic and potential
ejected) energies of incident ion, respectively

• Electron emission by positive ions is principal secondary process in Townsend spark discharge
mechanism
• Neutral excited (metastable) atoms or molecules incident upon electrode surface are also capable of
ejecting electrons from surface

73
 Thermionic Emission
• Conduction electrons in metals at room temperature do not have
sufficient thermal energy to leave surface
much lower than Wa for typical elements
If we consider electrons as a gas
at room temperature, then their
average thermal energy

• At high metal temperature (1500–2500 K), electrons receives energy from violent thermal lattice
vibrations sufficient to cross surface barrier and leave metal
Where,
Emission current related to temperature e, m – electronic charge and mass
of emitter by Richardsonrelation for h –Planck’s constant
thermionically emitted saturation k – Boltzmann’s constant
current density T – absolute temperature
Wa –surface work function 74
 Thermionic Emission
Shows that saturation current density increases with decreasing
work function and increasing temperature

• Although electrons may possess required escape energy, some of them may be reflected back into
solid from surface atoms or surface contaminants such as adsorbed gases

→ Instead of A use Aeff = A(1 – R) [R – reflection coefficient]

• Presence of a strong electric field reduce work function as the Schottky effect and thermionic
emission will be enhanced

75
 Field Emission
• Electrons may be drawn out of a metal surface by very high electrostatic fields

• Strong electric field at surface of a metal may modify potential barrier at metal surface to such an extent that

electrons in upper level close to Fermi level will have a definite probability of passing through the barrier

→ The effect is called ‘tunnel effect’

• Fields required to produce emission currents of a few microamperes are of order of 107–108 V/cm

▪ Such fields are observed at fine wires, sharp points and submicroscopic irregularities with an average

applied voltage quite low (2–5 kV)

▪ These fields are much higher than breakdown stress even in compressed gases
76
 Townsend Second Ionization Coefficient 𝛾
Current due to • A graph of log I against gap length (d) should yield a straight line
increased ionization of slope 𝛼 if pressure p and E is kept constant
because of collision
[electron multiplication in a gas] • When measuring current in parallel plate gaps Townsend
observed that at higher voltages current increased more rapidly
Why this departure from linearity?

• Townsend postulated that a second mechanism is affecting

current
→ liberation of electrons from cathode by positive ion
bombardment
• Other processes responsible for upcurving of (log I vs. d) graph
▪ secondary electron emission at cathode by photon impact
▪ photoionization in the gas itself
Plot of log I vs. electrode gap at a constant pressure 77
 Townsend Second Ionization Coefficient 𝛾
Consider a self-sustained Let,
n – number of electrons reaching anode per second
discharge where electrons are
n0 – number of electrons emitted from cathode by (say) u.v. illumination
produced at cathode by positive n+ – number of electrons released from cathode by positive ion bombardment
ion bombardment 𝛾 – number of electrons released from cathode per incident positive ion

Townsend’s suggestion Ions rapidly lose energy in Positive ions are unable to gain sufficient
for secondary ionization elastic collisions → Avg. fractional energy from field to cause ionization on
in gas by positive ion energy loss by incoming particle collision with neutral atoms or molecules
impact does not work during a collision at an angle 
78
Secondary Electron Emission by Photon Impact
Secondary emission from photon impact at cathode may be expressed by

𝜃 – number of photons produced by an electron in advancing 1 cm in direction of field

𝜇 – average absorption coefficient for photons in gas
g – a geometrical factor representing fraction of photons that reach cathode
𝜂 – fraction of photons producing electrons at cathode capable of leaving the surface

• Both positive ions and photons may be active simultaneously in producing electrons at cathode +
metastable atoms may contribute to secondary emission at cathode

• More than one mechanism may operate

in producing secondary ionization in
𝛾 may represent one or more of
discharge gap → it is customary to
several possible mechanisms
express secondary ionization by a single
(𝛾 = 𝛾I + 𝛾ph + …..)
coefficient 𝛾 and represent current by 79
Transition from non-self-sustained
discharges to breakdown

80
 Townsend Spark Mechanism
As voltage between electrodes in a gas with small or negligible electron
attachment increases, electrode current at anode increases given by Eq.

Considering • ed – electron avalanche; represents number of electrons

that, produced by one electron in travelling from cathode to anode Also, E = V/d
• At a constant temperature for a given gas energy distribution Therefore,
depends only on E/p
• For a given energy distribution probability of an ionization
occurring depends on gas pressure

Hence, we can write • This relationship holds until at some point

there is a sudden transition from dark
current I0 to a self-sustaining discharge

81
 Townsend Spark Mechanism

• When self-sustaining discharge starts, the current (I) becomes

indeterminate and denominator in the Eq. vanishes, i.e.

Taking electron attachment Townsend breakdown criterion

(𝜂) into account

And, since,
= effective ionization coefficient
see slide #55

• Electron current at anode equals current in external circuit

→ Theoretically value of current becomes infinitely large, but in practice it is limited by external
circuitry and, to a small extent, by voltage drop within arc
82
 Townsend Spark Mechanism
Townsend breakdown criterion
At this condition ion pairs produced in gap by passage of one
electron avalanche is sufficiently large that resulting positive
ions bombarding the cathode are able to release one secondary
This defines conditions for onset of spark
→ also called Townsend criterion for electron and so cause a repetition of avalanche process
spark formation

• Secondary electron may also come from photoemission process

• In either case electron avalanche will have a successor → discharge is then self-sustaining, continue

in absence of source producing I0, so that the above criterion can be said to define sparking threshold

According to Townsend spark mechanism → gap current grows as a result of ionization by

electron impact in gas and electron emission at cathode by positive ion impact 83
 Townsend Spark Mechanism
Townsend breakdown
criterion

• For ionization produced by

successive avalanches is cumulative

• Spark discharge grows more rapidly the more

exceeds unity

• For current I is not self-sustained

84
 Townsend Spark Mechanism
• Townsend mechanism of breakdown gas at atmospheric pressure have certain drawbacks

1. According to Townsend theory, current growth occurs as a result of ionization process only

→ Studies show that breakdown voltages depend on gas pressure and geometry of gap

2. According to Townsend theory, formative time lag of spark should be at best equal to electron
transit time ti
→ But, in air at pressures around atmospheric experimentally determined time lags have been
found to be much shorter than ti
• [ Townsend theory holds in uniform field under very low pressure ]
• [ In nonuniform field under higher pressure faster breakdown takes place ] → Streamer
theory of breakdown can explain this
• Raether and, Meek and Loeb independently proposed Streamer theory
85
 Townsend Mechanism
Alternative expression for the Townsend breakdown criterion

〈 Since, and 〉
see slide #82

K can often be treated as being constant, due to following phenomena

• Electron emission processes characterized by 𝛾 greatly affected by cathode surface and gas pressure
• Value of 𝛾 very small (10–2 to 10–3) → 1/𝛾 quite high

• So, does not change too much for a Townsend discharge of the order of 8–10

• As 𝛼 is often very strongly dependent upon gas pressure p or field strength E, exact value of K is of
minor importance and may be treated as a constant for many conditions of p and E
86
Streamer or ‘Kanal’ Mechanism of Spark

87
 Streamer or ‘Kanal’ Mechanism of Spark
ed – electron avalanche; represents number of electrons
produced by one electron in travelling from cathode to anode
• Growth of charge carriers in an avalanche in a uniform field (E0 = V0/d) is described by e𝛼d
– This is valid only as long as electrical field of space charges of electrons and ions can be neglected
compared to external field E0
• Raether studied effect of space charge of an avalanche on its own growth; he observed that
→ when ion concentration > 106 but <108 growth of an avalanche weakened
→ when ion concentration > 108 avalanche current was followed by a steep rise in current and
breakdown of gap followed
• Both – underexponential growth at the lower concentration and rapid growth at high concentration –
have been attributed to modification of originally uniform field E0 by space charge field
88
 Streamer or ‘Kanal’ Mechanism of Spark
• Space charge at head of avalanche assumed
concentrated within a spherical volume, with negative
charge ahead because of higher electron mobility

Field distortion in a gap

caused by space charge
of an electron
avalanche

C A

(a) Free electron and collision ionization; (b) Initial electron

avalanche; (c) Second electron avalanche; (d) Negative
streamer 89
Streamer or ‘Kanal’ Mechanism of Spark
① Field is enhanced in front of head of avalanche with field lines
from anode terminating at head
② Further back in avalanche, field between electrons and ions left
behind reduced applied field E0
③ Still further back field between cathode and positive ions
enhanced again
①

③ • Field distortion noticeable with a carrier number n > 106

• If carrier number in avalanche reaches n  108, space charge field
becomes of same magnitude as applied field and may lead to initiation of
a streamer
②
• Space charge fields play an important role in corona and spark discharges
in nonuniform field gaps
90
 Streamer or ‘Kanal’ Mechanism of Spark
• Under certain conditions space charge developed in an avalanche is capable of transforming
avalanche into channels of ionization (streamers) that lead to rapid development of breakdown

• Transformation from avalanche to streamer generally occurs when charge within avalanche head
reaches a critical value of n0e𝛼Xc ≈ 108 or 𝛼xc ≈ 18–20

→ If xc > gap length d then initiation of streamers is unlikely

German physicist Heinz Artur Raether (1909-1986) postulated

development of two types of streamers:
(1) ‘anode directed streamer’ describing apparent growth of E0
ionization and of avalanche head Xc

xc – length of avalanche path in

(2) ‘cathode directed streamer’ describing additional discharge
field direction when it reaches
growth from avalanche tail critical size 91
 Streamer or ‘Kanal’ Mechanism of Spark
• Raether, Meek and Loeb independently developed the ‘streamer’ or ‘Kanal’ mechanism for spark
formation, in which secondary mechanism results from photoionization of gas molecules and is
independent of electrodes
• Raether and Meek proposed that when avalanche in gap reaches a certain critical size combined
space charge field and externally applied field lead to intense ionization and excitation of gas
particles in front of avalanche head
• Instantaneous recombination between positive ions and
electrons releases photons which in turn generate secondary
electrons by photoionization process → these electrons under
influence of electric field in gap develop into secondary
avalanches
• Since photons travel with velocity of light, the process leads to a Secondary avalanche formation
rapid development of conduction channel across gap by photoelectrons 92
 Streamer or ‘Kanal’ Mechanism of Spark
Raether’s developed an empirical expression Er – space charge field strength
directed radially at head of avalanche
for streamer spark criterion (1939) E – externally applied field strength

• Field strength in front of avalanche is E + Er

• In positive ion region just behind head field reduced to E – Er

• Space charge increases with avalanche length (e𝛼x)

• Condition for transition from avalanche to streamer assumes that

space charge field Er  E
Space charge field Er around
avalanche head • Breakdown criterion then becomes
93
 Streamer or ‘Kanal’ Mechanism of Spark
• Minimum breakdown value for a uniform field gap by streamer mechanism
obtained on the assumption that transition from avalanche to streamer
occurs when xc = d

→ Raether’s empirical expression for this condition

xc = d gives smallest value of α to produce streamer breakdown

94
 Streamer or ‘Kanal’ Mechanism of Spark
• In Raether’s model transition assumed to take place when radial field about positive space charge in
an electron avalanche attains a value of the order of externally applied field
• A similar criterion equation for transition from avalanche to streamer has been developed by
physicist John M. Meek (1940)
x – distance (in cm) which
• Meek showed that radial field avalanche has progressed
p – gas pressure in torr and
produced by positive ions 𝛼 –Townsend coefficient of
immediately behind head of ionization by electrons
corresponding to field E
avalanche can be calculated from

Minimum breakdown voltage assumed to

correspond to condition when avalanche has
crossed gap of length d (i.e. x = d) and Er = E
95
Sparking Voltage – Paschen’s Law

96
 Sparking Voltage – Paschen’s Law
• Paschen's law is an equation that gives the breakdown
voltage between two electrodes in a gas as a function
of pressure and gap length

• Named after Friedrich Paschen who discovered it

empirically in 1889

German physicist
Louis Carl Heinrich Friedrich Paschen
(1865 - 1947)
97
 Sparking Voltage – Paschen’s Law
An analytical expression for breakdown voltage for uniform field gaps as a function of
gap length d and gas pressure p can be derived from threshold Eq. by expressing

ionization coefficient 𝛼՟/p as a function of field strength E and gas pressure p

see slide #86

• Putting 𝛼՟/p = f(E/p) in the criterion equation or,

• For uniform field, breakdown voltage Vb = Ed

Paschen’s law, established experimentally in 1889

i.e. breakdown voltage of a uniform field gap is a

or,
unique function of product of pressure and electrode
separation for a particular gas and electrode material 98
 Sparking Voltage – Paschen’s Law
• Paschen’s law does not imply that Vb increases
Sparking voltage–pd relationship
linearly with (pd) (Paschen’s curve)

• Sparking voltage goes through a minimum value

(Vbmin) at a particular value of product (pd)min

What is the relation of Paschen’s curve with

Townsend breakdown criterion?

99
 Sparking Voltage – Paschen’s Law
Relation between Townsend criterion for spark (𝛼d = k) • Rearranging Townsend criterion
and function of 𝛼/p = f(E/p)

and noting that in uniform field applied

voltage, V = Ed gives

and

For constant values of K the Eq. gives straight

lines of slope (tan𝜃) depending upon value of
applied voltage (V), curves (2, 3, 4)

Limiting value (E/p) correspond

100
to onset of ionization
Sparking Voltage – Paschen’s Law
For constant values of K Eq. • Low values of V there is no intersection between
gives straight lines of slope
(tan𝜃) depending upon value the line (curve 4) and curve 𝛼/p = f(E/p)
of applied voltage V
→ no breakdown with small V below Paschen’s minm
irrespective of value (pd) in Eq.

• At higher V, there must exist two breakdown values

at a constant pressure, one corresponding to small
value of gap length d1 and intersection at N and the
other to longer gap d2 intersection at W
• Point S (tangent) gives lowest breakdown value or
minimum sparking voltage
101
 Sparking Voltage – Paschen’s Law
Existence of minimum value in V–d relation may be explained qualitatively by considering
efficiency of ionization of electrons traversing the gap with different electron energies

• Neglecting secondary coefficient 𝛾 for (pd) > (pd)min, electrons crossing the gap make
𝛾 – represents
secondary more frequent collisions with gas molecules than at (pd)min, but energy gained
emission at
between collisions is lower than at (pd)min
cathode due to
ion and photon → hence, probability of ionization is lower unless voltage is increased
impacts; we are
neglecting this
process here
• For (pd) < (pd)min electrons cross gap without making many collisions

• Point (pd)min corresponds to highest ionization efficiency

102
 Sparking Voltage – Paschen’s Law
Analytical expression for the minimum values of Vbmin and pdmin

Townsend criterion for

spark formation

Using above expressions and considering constant 𝛾 gives,

Rearranging, we obtain

103
 Sparking Voltage – Paschen’s Law
• To obtain minimum sparking constantsVbmin and pdmin

Therefore, and

Substituting into gives

104
 Sparking Voltage – Paschen’s Law
Townsend criterion for spark Ec - limiting value of
formation may be written as, field strength E at
which effective
ionization starts
Schumann relation showed that over a wide but p – pressure
restricted range of E/p, 𝛼/p may be expressed as, C – constant

Then,
Using above expressions
breakdown
and some manipulation
voltage,
gives,

For homogeneous field gaps at p = 1 bar, 20°C gives

Ec = 24.36 (kV/cm), K/C = 45.16 (kV)2/cm ... and, 105
 Sparking Voltage – Paschen’s Law

It is often more convenient to use gas density 𝛿 instead of gas pressure p in Paschen’s
Law since is this case effect of temperature at constant p on mean free path is taken
in account

• Number of collisions by an electron in crossing the gap proportional to product 𝛿d and 𝛾

106
 Sparking Voltage – Paschen’s Law
• Atmospheric air provides basic insulation for many practical HV installations (transmission lines,
switchyards, etc.)
• Since atmospheric conditions (temperature, pressure) vary considerably in time and locations,
breakdown characteristics of various apparatus will be affected accordingly
• Breakdown characteristics can be converted to standard atmospheric conditions (p = 760 torr = 1.01
bar and t = 20°C = 293 K)

Correction for variation in ambient conditions

made by introducing relative density defined as,

Breakdown voltage at standard conditions multiplied by

this factor gives breakdown voltage corresponding to
given ambient conditions approximately,

107
 Penning Effect
• Paschen’s law not applicable in many gaseous mixtures
→ example - neon–argon mixture [Penning mixture]
• A small admixture of argon in neon reduces breakdown
strength below that of pure argon or neon
• Because–
▪ Lowest excited state of neon is metastable; its excitation
potential (16 eV) is 0.9 eV greater than ionization
potential of argon
▪ Metastable atoms have a long life in neon gas, and on
hitting argon atoms there is a very high probability of
ionizing them Breakdown voltage curves in neon–argon
• This phenomenon is known as Penning effect mixtures between parallel plates at 2-cm
spacing at 0°C 108
 Penning Effect
• A very common Penning mixture
▪ 98–99.5% of neon with 0.5–2% of argon used in some neon lamps

▪ Penning mixture is easier to ionize than either neon or argon alone,

and lowers striking voltage at which the tube becomes conductive
and starts producing light

▪ Optimal level of argon ~0.25%, but some of it gets adsorbed onto

borosilicate glass tubes, so higher concentrations are used

• Penning mixture of neon and argon also used as a starter gas in

Frans Michel Penning (1894 – 1953)
sodium vapor lamps, where it is responsible for faint reddish Dutch experimental physicist

glow before the sodium emission begins Studied low pressure gas discharges
at Philips Laboratory in Eindhoven
109
Breakdown Field Strength

110
 Breakdown Field Strength (Eb)
• For uniform field gaps breakdown field strength (Eb) in a gas may be given by,

• For a constant pressure (p) breakdown field strength (Eb) decreases steadily with gap length (d)
• Field strength to pressure ratio (Eb/p) dependent only upon (pd)
• Breakdown field strength (Eb) for a constant gap length increases with pressure but at a rate slightly
lower than directly proportional, as pressure also affects denominator in expression

111
 Breakdown Field Strength (Eb)
Relationship between ionization Functional relationship between ionization
coefficient 𝛼՟ and field strength coefficient 𝛼 and breakdown field strength at
different values of (pd)
Decrease in pressure related breakdown field strength (Eb/p)
with increasing (pd) can be understood by considering above
relationship and applying Townsend criterion equation to
different values of (pd) [shown in Fig.]

Townsend criterion for spark

formation may also be written as,

• Assuming first that this equation is satisfied for a small (pd)2 at

(Eb/p)2 when (𝛼/p) reaches point A and then increasing to (pd)1,
criterion equation will now be satisfied at a lower value of 𝛼՟/p(B)
giving a reduced breakdown strength (Eb/p)1 112
 Breakdown Field Strength (Eb)

Fig. (b)

By repeating the procedure described in slide 98 for other values of (pd) we obtain a functional
relationship between breakdown strength (Eb/p) and (pd) as shown in Fig. (b) 113
Breakdown in Non-uniform Fields

114
 Breakdown in Non-uniform Fields
• In non-uniform fields, e.g. in point-plane, sphere-plane gaps or coaxial cylinders, field strength and
hence effective ionization coefficient 𝛼՟ vary across gap length

• Electron multiplication is governed by integral of 𝛼՟ over path ∫𝛼՟ dx

Townsend criterion • Integration must be taken along line of highest

for spark at low field strength
pressures, • Expression valid also for higher pressures if field is
d – gap length only slightly non-uniform

• In strongly divergent fields there will be at first a region of high values of E/p over which 𝛼/p > 0
• When field falls below a given strength Ec integral ∫𝛼՟ dx ceases to exist [no more ionization]
• Townsend mechanism then loses its validity when Breakdown (or inception of discharge) is
criterion relies solely on 𝛾 effect, especially for low still possible if photoionization processes
field strength at cathode is taken into account 115
Breakdown in Non-uniform Fields

Electric field distribution in a

non-uniform field gap
(positive point-plane)

116
 Breakdown in Non-uniform Fields
Criterion condition for breakdown for general case can be
represented by modifying expression for breakdown criterion for
streamer mechanism to take into account non-uniform distribution of
𝛼՟, i.e.,
Ncr – critical electron concentration in an avalanche giving rise to
initiation of a streamer (approx. 108)
xc – path of avalanche to reach this size d – gap length

• This Eq. is applicable to calculation of breakdown or discharge inception

voltage, depending on whether direct breakdown occurs or only corona

117
 Breakdown in Non-uniform Fields
Coaxial cylindrical geometry in air
• An empirical relation based on many measurements of critical
field strength Ec (corona inception) for different diameters of
inner conductor (2r) and relative air density 𝛿 developed by F. W. For more on relative density 𝛿 see slide #107
Peek (1920)

118
Partial Breakdown, Corona Discharges

119
 Partial Breakdown, Corona Discharges
• Onset of measurable ionization usually leads to complete breakdown of gap in uniform field and
quasi-uniform field gaps

• In non-uniform fields various manifestations of luminous and audible discharges are observed long
before the complete breakdown occurs «corona»

▪ Cause considerable power losses in HV transmission lines

▪ Leads to deterioration of insulation by combined action of discharge ions bombarding surface

and action of chemical compounds formed by discharge

▪ Create interference in communication systems

• This phenomena has some industrial applications - high-speed printing devices, electrostatic
precipitators, paint sprayers, Geiger counters, etc.
120
 Partial Breakdown, Corona Discharges
• Positive corona - under positive voltage, a corona appears as a uniform bluish-white sheath over

entire surface of wire

• Negative corona - 0n negative wires corona appears as reddish glowing spots distributed along wire

• Impulse corona that occurs under pulsed voltage of short duration, where no space charge is

permitted to drift and accumulate

• Static field corona that occurs under long lasting (d.c.) voltages

121
 Positive or Anode Coronas
• Convenient electrode configurations for study of corona
i. Hemispherically capped rod-plane [by varying radius of electrode tip,
different degrees of field non-uniformity can be readily achieved]
ii. Point-plane gaps [suitable for obtaining a high localized stress and for
localization of dense space charge]
• Because of transient development of ionization, growth of discharge is
difficult to monitor precisely; two techniques can do it
▪ ‘Lichtenberg* figures’ techniques
▪ High-speed photographic techniques

* Georg Christoph Lichtenberg (1742 – 1799), German physicist; discovered

branching discharge patterns on dielectrics (Lichtenberg figures); a crater on
the Moon is named Lichtenberg in his honour 122
 Positive or Anode Coronas
• Fig. (a) – when a positive voltage pulse is applied
to a point electrode, first detectable ionization is
of a filamentary branch nature - this discharge is
called a streamer and is analogous to case of
uniform field gaps at higher (pd) values
• Figs. (b) & (c) – as impulse voltage level is
increased, streamers grow both in length and
their number of branches
Formation of streamers under impulse voltage
– progressive growth with increasing pulse duration –
• Velocity of streamers decreases rapidly as they
positive rod-plane gap penetrate low field region
[branches never cross each other]
123
 Positive or Anode Coronas
When voltage applied for an infinitely long time (e.g. under d.c. or 50 Hz)
→ ionization products have sufficient time to accumulate in space → cause distortion in original field

rod-plane gap
with rod tip of 1
cm radius; in
atmospheric air

Onsets of various discharge modes

with corresponding discharge
characteristic as gap length is
increased

124
 Positive or Anode Coronas
• Gap length < ~2 cm – as voltage gradually raised no
appreciable ionization detected up to breakdown
• As gap increased, field distribution becomes more
inhomogeneous
▪ On increasing voltage at first a transient slightly
branched filamentary discharge appears →
streamers
• Under steady state streamer develops with varying
frequencies
▪ Streamer currents are proportional to their length
▪ These are called onset streamers or burst pulses
125
 Positive or Anode Coronas
• Voltage increased further – streamers become more
frequent, until transient activity stops, discharge
becomes self-sustained and steady glow appears close
to anode
• Glow gives rise to continuous but fluctuating current
• Further increase in voltage – increases luminosity of
glow both in area and in intensity
• Glow corona develops only in presence of negative ions
• Voltage increased still further – new and more vigorous
streamers appear → ultimately lead to complete
breakdown of gap
126
 Positive or Anode Coronas
• Curve 1 – at smaller spacing when voltage still reasonably
uniform, streamer is capable of penetrating weaker field,
reaching cathode and initiating breakdown in same manner
as in uniform field gaps
• Curve 2 – spacing > 10 cm, streamers appear that do not
cross gap
• Curve 3 represents transition from streamers to steady glow
corona without sparking
• At larger spacings considerable spread in voltage at which
breakdown streamers develop before complete breakdown
of gap
• Dashed area – region of uncertain transitions; portion 1 indicates
onset of streamers followed immediately by transition to spark
127
 Positive or Anode Coronas
• Gap increased to a point where glow is established and then
reduced keeping voltage constant → glow discharge will
stabilize gap against breakdown at a voltage that otherwise
would have broken down
• If voltage is then raised, a spark is induced by glow corona
(curve 4); if it is lowered, streamer breakdown is induced
• By decreasing gap further and increasing voltage at various
points glow-corona sparking voltage characteristic can be
projected backwards (shown by curve 4)
• Thus if a steady corona glow is established, sparking voltage
is raised and lower breakdown by streamer is suppressed 128
 Negative or Cathode Coronas
• As negative voltage applied to point in point-plane
geometry in electronegative gases is increased, different
phases of negative corona appear …
1. Auto-stabilization
2. Regular pulses [Trichel pulse regime]
3. Continuous current discharge
• Trichel pulses are very regular in oxygen and air, but
irregular in SF6
• Trichel pulses only occur in electronegative gases

129
 Negative or Cathode Coronas
• Lowest curve shows onset voltage for Trichel pulses
not greatly affected by gap length
• Voltage rise does not change mode of pulses over a
wide voltage range
• At a much higher voltages steady glow discharge
observed
• Transition from Trichel pulses to glow discharge not
sharply defined, hence shown as a transition region
• Further increase in voltage – glow discharge persists
until breakdown occurs
130
Negative or Cathode Coronas
Relationship between pulse frequency and gap voltage for
different gap lengths and a cathode point of 0.75 mm radius in
atmospheric air
• Pulse frequency increases with voltage and depends upon
radius of cathode, gap length and pressure
• A decrease in pressure decreases frequency of Trichel pulses
• Breakdown under negative polarity occurs at considerably
higher voltage than under positive voltage, except at low
pressures; therefore, under alternating power frequency
voltage breakdown of non-uniform field gap invariably takes
place during positive half-cycle of voltage wave

131
Negative or Cathode Coronas
Characteristic points on a typical Trichel pulse
1 - beginning of pulse
2 - half-pulse rising
3 – maximum of pulse
4 - half-pulse decreasing
5 - end of pulse
6 - 4 microseconds after pulse
• Very short rise times (as short as 1.3 ns) and
very short durations (tens of ns) separated by
longer inter-pulse periods (tens of μs)

132
Polarity Effect – Influence of Space Charge

133
 Polarity Effect – Influence of Space Charge
• In non-uniform field gaps in air appearance of the first
streamer may lead to breakdown or it may lead to
establishment of a steady state corona discharge which
stabilizes the gap against breakdown
• Accordingly we may have a corona stabilized or direct
breakdown, it depends on
▪ Degree of field non-uniformity
▪ Gas pressure
▪ Voltage polarity
▪ Nature of the gas
→ For example, in air corona stabilized breakdown will extend to
higher pressures than in SF6 due to relatively immobile SF6 ions 134
 Polarity Effect – Influence of Space Charge
Comparison of positive and negative point-plane gap
breakdown characteristics measured in air as a function
of gas pressure
• At very small spacing breakdown characteristics for
two polarities nearly coincide and no corona stabilized
region is observed
• As spacing is increased, positive characteristics display
distinct high corona breakdown up to a pressure of
approximately 7 bar, followed by a sudden drop in
— positive point
- - negative point breakdown strengths
• Under negative polarity corona stabilized region
extends to much higher pressures
Point-plane breakdown and corona inception
135
characteristics in air (point radius r = 1 mm)
 Polarity Effect – Influence of Space Charge
Coaxial cylindrical arrangement – • Cylinder dimensions for such a
a frequently used non-uniform field system can be optimally designed for
geometry in construction of HV apparatus maximum corona-free breakdown

Consider a system of two coaxial cylinders with inner radii ri , outer radii ro
• Field strength in • Since breakdown or corona • Maximum breakdown
interelectrode space at onset will follow when voltage voltage for the system
radial distance r stress at smaller wire reaches given by
breakdown stress Eb we can
write the equation as

V – applied voltage
Eb – breakdown (or corona
inception) field strength of system 136
 Polarity Effect – Influence of Space Charge
Field strength (Eb) depends upon
gas density (𝛿) and radius of Dotted curve indicates
corona onset voltage
inner conductor (ri) and solid curve
breakdown voltage
• For not too small radii ri and/or strongly attaching gases
(with a steep increase of 𝛼՟/p = f(E/p)), Eb ≅ constant
• Keeping ro constant optimal design for the system is
given by

Relationship between breakdown voltage and

or, and inner radius in a coaxial cylinder system
(for a fixed radius of outer cylinder)
137
 Polarity Effect – Influence of Space Charge
Direct breakdown voltage characteristics for nitrogen • At low pressures breakdown voltage is usually lower
at low pressures between a wire and coaxial cylinder
when smaller electrode is negative
Curve 1 - positive wire, curve 2 - negative wire
→ due to higher field at cathode so that 𝛾 is greater,
therefore a lower value needed for to satisfy
sparking criterion equation
• At higher pressures order of characteristics reversed
• Large polarity effect at higher pressure can be
qualitatively explained by considering role of space
charge of pre-breakdown current

138
 Polarity Effect – Influence of Space Charge
• High mobility electrons drawn into anode →
Positive point 〉〉 an ionization by electron collision
takes place in high field region close to the point positive space charge left behind → cause a
reduction in field strength close to anode and
increase field further away from it

• High field region is in time moving further into

gap extending region for ionization → field
strength at tip of space charge may be high
enough for initiation of a cathode-directed
streamer which subsequently may lead to

(a) Space charge build-up in positive point-plane gap complete breakdown

139
(b) Field distortion caused by positive space charge
 Polarity Effect – Influence of Space Charge
Negative point 〉〉 electrons repelled into low • Vicinity of point 〉〉 field is enhanced, but
field region and in case of attaching gases get
ionization region drastically reduced → it
attached to gas molecules and tend to hold back
terminates ionization
positive space charge which remains in space
between negative charge and the point • Once ionization ceases, applied field sweeps
away negative and positive ion space charge
from vicinity of point and cycle starts again after
clearing time for space charge
• To overcome retarding action of ions a higher
voltage is required → hence negative
breakdown voltage higher than positive
breakdown voltage in gaps with marked
asymmetrical fields
(a) Space charge build-up in negative point-plane gap
140
(b) Field distortion by space charge
 Polarity Effect – Influence of Space Charge
• Mathematically at any given time voltage across the gap is given by
the field integral

• Integration of space charge distorted field in Figs (a) and (b),

(a) Positive point-plane gap respectively, shows immediately that

(b) Negative point-plane gap 141

 Surge Breakdown Voltage – Time Lag
• For initiation of breakdown an electron must be available to start avalanche

• With slowly rising voltages (d.c. and a.c.) there are usually sufficient initiatory electrons created by
cosmic rays and naturally occurring radioactive sources

• Under surge voltages and pulses of short duration → gap may not break down as peak voltage
reaches lowest breakdown value (Vs) unless presence of initiatory electrons is ensured by using
artificial irradiation

• Vs leads to breakdown of gap after a long time of application

• With weak irradiation peak value may have to be greatly increased so that voltage remains above
d.c. value Vs for long intervals of time

142
 Surge Breakdown Voltage – Time Lag

t – time lag; time which elapses between application

of voltage to a gap sufficient to cause breakdown and
the breakdown Vs

ts – statistical time lag; time which elapses during

the voltage application until a primary electron
appears to initiate the discharge

Breakdown on a step-function voltage pulse tf – formative time lag; time required for breakdown
Vp – peak value of a step voltage applied at time t = 0 to a to develop once initiated
gap that breaks down under Vs after a long time
Vs – minimum static breakdown voltage 143
 Surge Breakdown Voltage – Time Lag
statistical time lag
• ts depends upon amount of pre-ionization in gap
• Pre-ionization depends upon size of gap and the radiation
producing primary electrons
▪ Artificially irradiating gaps by UV light, radioactive materials, and
illumination by auxiliary sparks reduce ts
▪ Application of an overvoltage (Vp – Vs) to gap also reduce ts

formative time lag

• tf depends upon mechanism of spark growth in question
• When secondary electrons arise entirely from electron emission at cathode by positive ions, transit
time from anode to cathode will be dominant factor determining tf
• tf increases with gap length and field nonuniformity, but decreases with applied overvoltage
144
 Breakdown Under Impulse Voltages
• For a given impulse voltage waveshape
overvoltage duration will increase with
voltage amplitude (Vp)
• Time lags (ts, tf) are statistical in nature
→ when a given number of impulses of an
amplitude Vp > Vs are applied to a gap only
• Impulse voltage of peak value > Vs applied to a
a certain percentage will lead to
gap 〉〉 there is probability but no certainty that
breakdown
breakdown will occur
→ there is a breakdown probability P for
• For breakdown it is essential that spark develops
during interval of overvoltage [V(t) – Vs] duration, each given applied maximum impulse
i.e. [t < (t2 – t1)] voltage Vp as a function of Vp 145
 Volt-Time Characteristics
• When an impulse voltage of sufficiently high value
is applied to a gap, breakdown will result on each
voltage application

• Time required for spark development depend

upon rate of rise of voltage and field geometry

→ For a given gap geometry volt–time characteristic

is obtained by applying a number of impulses of
increasing amplitude and noting the time lag Impulse ‘volt–time’ characteristics

• Volt–time characteristic is an important practical property of any insulating device or structure

• It provides basis for establishing impulse strength of insulation as well as for design of protection level against
146
overvoltages
 Volt-Time Characteristics
• In uniform and quasi-uniform field gaps volt-time
characteristic usually sharply defined 〉〉 rises steeply
with increasing rate of rise of applied voltage

• In non-uniform field gaps 〉〉 due to larger scatter in

results, data fall into a dispersion band [shown in
figure]

• Time to breakdown is less sensitive to rate of voltage

rise 〉〉 hence quasi-uniform field gaps (sphere–sphere)
Schematic diagram of volt–time
often used as protective devices against overvoltages characteristics for uniform and non-
uniform field gaps

147