ELECTRODE EROSION MEASLHEMENTS IN A

HIGH ENERGY SPARK GAP

by

MTHOOT LYLE DONALDSON B.S.E.E.

A THESIS

IN

ELECTRICAL ENGINEERING

Submitted to the Graduate Facultv

of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of

MASTER OF SCIENCE
IN
ELECTRICAL ENGINEERING

Ap?^oved

Accepted

August, 1982
 /I 0»^ ^ . ACKNOWLEDGEMENTS

I would like to thank Dr. Magne Kristiansen, Dr. Marion Hagler,

and Dr. Lynn Hatfield for serving on my committee and for their

encouraging assistance and patient guidance during the course of this

work. Their enthusiastic approach to the research has made this work

both a reward and a challenge. I am also quite thankful for the

surface analysis work done by Dr. Allen Bowling and Sam Prien. I

thank George Jackson for his help with the processing and interpreta-

tion of the surface analysis, and Richard Ness for his contribution

to the work done in acquiring and interpreting the self-breakdown

voltage distributions. Special thanks goes to Ken Rathbun who was

responsible for much of the critical design and construction of the

test facility. I would like to thank many others who helped with the

construction and the data processing including: Brian Maas, Kim

Zinsmeyer, Bob Conover, Eric Kristiansen, David Riethmeyer, Aamer

Shaukat, and Zahid Maniya. I would also like to thank Don and Cheri

Johnson, Pat Darden, Kristi Koch, Mario Perez, Jeannette Davis, and

Sharon Lipscomb for their help on this project and the completion of

this manuscript. The financial support by the Air Force Office of

Scientific Research and Texas Tech University is appreciated.

Finally, I would like to thank my parents, Mr. and Mrs. W. Lyle

Donaldson, for their loving support and constant exhortation, and the

many brothers and sisters in Christ who provided encouragement and

prayer.

11
 TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

LIST OF TABLES iv

LIST OF FIGURES v

I. INTRODUCTION 1

II. EXPERIMENTAL DESCRIPTION 4

Test Circuit 4
Spark Gap 4
Operating Conditions 11
Materials Tested 13

III. EXPERIMENTAL RESULTS 15

Erosion Characteristics 15
Material 15
Polarity 17
Gas 20
Surface Conditions 20
Brass 21
Cathode 24
Anode 24
Insulator 29
Self-Breakdown Voltage Distribution 32
Material 32
Shot Number 37
Gas ^0

IV. CONCLUSION 42

LIST OF REFERENCES 44

APPENDIX MANUFACTURE AITO PREPARATION OF ELECTRODES 47

111
 LIST OF TABLES

Table Page

I Factors Affecting Electrode Erosion 2

II Electrode Material Properties 14

III Electrode Erosion Rates 16

IV Summary of Comparable Erosion Results 18

IV
 LIST OF FIGURES

Figure Page

1. Test Circuit for Erosion Studies 5

2. Experimental Test Facility for Erosion Studies . . . . 6
3. High Power Load 7
4. Data Acquisition System 8
5. Test Circuit Waveforms 9
6. Spark Gap for Erosion Studies 10
7. Spark Gap Electrodes 12
8. Cathode Erosion Rates for Different Electrode .... 19
Materials
9. Surface of Brass Electrodes 22
10. AES Surface Analysis of Brass Electrodes 23
11. Graphite and Copper-Graphite Cathode 25
Surfaces in Air
12. Stainless Steel and Copper-Tungsten Cathode 26
Surfaces in Air
13. Graphite and Copper-Graphite Anode 27
Surfaces in Air
14. Stainless Steel and Copper-Tungsten Anode 28
Surfaces in Air
15. Insulator Inserts Exposed to Graphite and 30
Copper-Graphite Electrodes in Air and Nitrogen
16. Insulator Inserts Exposed to Stainless Steel 31
and Copper-Tungsten Electrodes in Air and Nitrogen

V
Figure Page
17. SEM of Stainless Steel Electrode Surface 33
18. 50 ym Stainless Steel Particle on 34
Lucite Insulator
19. The Standard Deviation of the Self-Breakdown 35
Voltage VS. Shot Number for Different Electrode
Materials in Air
20. The Standard Deviation of the Self-Breakdown 36
Voltage vs. Shot Number for Different Electrode
Materials in Nitrogen
21. Optical Microscope Pictures of Typical Surface . . . . 38
Features for Copper-Tungsten and Stainless Steel
22. Self-Breakdown Voltage Distribution for Stainless . . 39
Steel Cathode with Graphite Anode and Graphite
Cathode with Stainless Steel Anode
23. Macrosopic Growth on Graphite Anode in Nitrogen ... 41
24. Electrode Assembly 48

VI
 CHAPTER I

INTRODUCTION

High energy spark gaps capable of lasting for 10^ shots are seen

as one of the critical components in pulsed power systems used for

nuclear isotope separation, electromagnetic pulse simulation, thermo-

nuclear fusion, and other applications. The performance of a pressur-

ized spark gap as a high energy, rep-rated, switching device is charac-

terized by the following parameters: hold-off voltage, recovery time,

delay time, and jitter. The switch lifetime is determined by the

degradation of these parameters resulting from electrode erosion, gas

decomposition and disassociation, and insulator damage occurring as

2
energy is dissipated in the switch.

Currently, electrode erosion is the main factor limiting switch

lifetime. For example, if one chooses an optimistic value of 1 ycm^/C

for the erosion rate, then a system transferring 0.1 C/shot would have
3 8

eroded 10 cm of material in 10 shots. As a result, many experiments

have been performed to measure erosion rates under a wide range of

experimental conditions. (Excellent summaries of the majority of these

1 3

experiments are given by Kristiansen and Hagler, and Holmes. )

However, due to the large number of parameters affecting electrode

erosion, a quite limited knowledge of the individual factors considered

in Table I has been obtained. In addition, many theoretical models

have been developed to explain the erosion process, and a few give good

agreement with experimental data under limited conditions. However, to

 Table I. Factors Affecting Electrode Erosion

Classification Factor

Electrical Current /idt

r±^ dt

di/dt

Pulse type
oscillatory
unipolar

repetition rate

polarity

Material Electrode type

thermophysical properties
melting temperature
specific heat
thermal conductivity
resistivity

fabrication technique

temperature

Insulator type

Gas type

pressure

flow rate

temperature

Miscellaneous Chemistry

Geometry

Trigger Method

The one that "does you in'

a large extent, many of the fundamental processes occurring at the

electrode surfaces are not understood and much work is still needed

to explain the erosion phenomena adequately.

The purpose of this study was to measure the erosion rate of

different electrode materials as a function of current in order to

generate a data base from which theoretical models describing the

complex erosion process could be developed and verified. In addition,

the self-breakdown voltage distributions and the electrode and insula-

tor surfaces were examined in an effort to define further the erosion

characteristics and to reduce the material parameter space used in

future studies.

Chapter II describes the experimental apparatus and the conditions

under which the experiment was performed. Chapter III presents the

experimental results and a brief comparison with the findings of other

investigators. A summary of the important conclusions is given in

Chapter IV along with possible areas to be considered for future study.

 CHAPTER II

EXPERIMENTAL DESCRIPTION

Test Circuit

Numerous investigators have measured erosion rates for high

4-9
current (10-800 kA) oscillatory discharges and a few measured

erosion rates for high current (>10 kA) unipolar discharges with brass

and copper electrodes. * Because of the nature of many switching

applications, a test circuit capable of delivering a unipolar pulse

was chosen for this study. The equivalent circuit, illustrated in

Fig. 1, consists of a six-section, type E, pulse forming network (PFN) 12

which is resistively charged to the self-breakdown voltage of the

spark gap by a 30 kV, 1 A, constant voltage power supply. The entire

test facility is shown in Fig. 2. When the spark gap breaks down,

the PFN is discharged into a matched, 0.6 Q., high power load which is

illustrated in Fig. 3. The discharge current and the charging voltage

are monitored during the experiment with the use of the data acquisition

system illustrated in Fig. 4. The computer software was developed

during the course of the experiment so all of the voltage distribution

data collected in this study were actually recorded with a storage

oscilloscope or an X-Y plotter. Sample oscilloscope traces of the

discharge current and the self-breakdown voltage distribution are

shown in Fig. 5.

Spark Gap

The spark gap shovm in Fig. 6 was designed to facilitate frequent

> <
2^ en
(U
O lO •H
ro CVJ 3
if It 4J
X X CO
< <
21 2
> M
CO
O
)>)

£ o
U
o
a
a>
•H
o 3
<o a
ll X u
•H

CM CO
rd
u 11
(U
H
O -J
 a) Top View

b) Side View

Fig. 2. Experimental Test Facility for Erosion Studies

 High Voltage
Connection {X12]

Water Cooling
Tube

Air intake

Water Flow

Air Row-

a) Top View

iU, '\ Wrj.llr.TXr ^]ri.l\ ''

n-^
Aluminum Plate
(Hifli Veltaq*)
^Mm^m^Fwi
^P "Plexiglass Hign
Voltage Enclosure
Carborundum Resistor-
(.2AHqltPo««r,Lo«
I III
i-^^t- •Phenolic Tubing
Muctanc* X 27)

G-IO Support Rods

Aluminum Fins —

'x-Vlf f tv "s
: Oround Return
Coax Coble<ai9)
Aluminum Plate
(GrMMd) mi u_'.j ' i i j — n j — u j j o ' TDT
Air Exhaust

b) S i d e View

Fig. 3. High Power Load

 8

GO

00 o
o
c
o

I-
cr
o
o
6
[^
N
^m

Oi i
6
O JP
<
b ^^^m MM^i^
<u
E >»
•u
^ GL CO
o O.
O CO
^
>
a 1 i c
o

CO
•H
3
c"
o
<3

03

GO
•H

•8
St.
3 9 o

r^ •» — T-
o > o

9>
>
^ UJ
^1^
 >
-a
>

shot number

a) Self-Breakdown Voltage Distribution

for 200 Shots

•nijHijiraiiHi
_>

sauui
<
lOl ki««"r^

KD/i.aec/d'n/
b) Current Pulse

F i g . 5. Test C i r c u i t Waveforms
 10

HV CONNECTION
FROy PFN
AIR INLET

ELECTRODE
SEPERATION
ADJUSTMENT
LUCITE

INSULATOR
BRASS INSERTS
ELECTRODE
HOLDER
AIR FLOW
H PORTS

ELECTRODE
TIPS ^ NYLON
HOUSING

-^—HV CONNECTION
TO LOAD

AIR OUTLET- -^anj=J

Fig. 6. Spark Gap for Erosion Studies

 11

electrode and insulator replacement and to allow for accurate control

of the electrode alignment and gap spacing. The electrodes shown in

Fig. 7a are composed of three parts: the brass support (which also

serves as a channel for air flow), the brass adapter, and the elec-

trode tips. The hemispherically shaped electrode tips shown in

Fig. 7b are 2.5 cm in diameter and are made from the various materials

studied. The manufacture and preparation of the electrode surfaces

is outlined in the Appendix. The lucite inserts provide protection

for the main gap housing and also provide a surface which gives a

permanent history of the discharge debris which is deposited on the

walls.

Operating Conditions

The ranges of the experimental operating conditions are

summarized below:

Voltage: < 30 kV

Current: < 25 kA

Total Capacitance: 21 yF

Charge/shot: < .6 C

Energy/shot: < 9 kJ

Pulse width: 25 ys

Rep-rate: < 5 pps

Q3s: A.ir or Nitrogen

Pressure: 1 Atm (absolute)

Flow rate: 1 Gap volume every 5 sec.

Gap spacing: < .75 cm

 11

electrode and insulator replacement and to allow for accurate control

of the electrode alignment and gap spacing. The electrodes shown in

Fig. 7a are composed of three parts: the brass support (which also

serves as a channel for air flow), the brass adapter, and the elec-

trode tips. The hemispherically shaped electrode tips shown in

Fig. 7b are 2.5 cm in diameter and are made from the various materials

studied. The manufacture and preparation of the electrode surfaces

is outlined in the Appendix. The lucite inserts provide protection

for the main gap housing and also provide a surface which gives a

permanent history of the discharge debris which is deposited on the

walls.

Operating Conditions

The ranges of the experimental operating conditions are

summarized below:

Voltage: < 30 kV

Current: < 25 kA

Total Capacitance: 21 yF

Charge/shot: < -6 0

Energy/shot: < 9 kJ

Pulse width: 25 ys

Rep-rate: < 5 pps

Gas: Air or Nitrogen

Pressure: 1 Atm (absolute)

Flow rate: 1 Gap volume every 5 sec.

Gap spacing: < .75 cm

 12

^m

a) Electrode Assembly

<^. .' I

p f> B
• 8 t:..J

b) Electrode Tips

Fig. 7. Spark Gap Electrodes

 13

Materials Tested

The electrode materials tested were: brass (SAE 660), stainless

steel (304), copper-tungsten (K-33), graphite (ACF-IOQ), and

14
copper-graphite (DFP-IC). This selection of materials allowed for:

1) a comparison with existing data for brass and stainless

,,,,l/,5,10,15

2) utilization of materials which have given good spark gap

perrormance, and

3) the testing of one new material, namely copper-graphite.

Some major properties of the materials tested are given in Table II,
 14

CO
0^ o
k O o CN

3
CO

OJ
4->
•H

a
CO

00
I
o a\
cu 0^
CO CM CO
CO in CM 0)
e
I o J2

o
U-4
CO •H
CO a CJ
0) s-/ 0)
•H
•u a
;^ CO
<u
D-
o /~\
!-i
P.1 on p>.
o CO
00
• o
iH
a.
s
o «» 00 CM
CO -v^ 1-i
•H >
V-i N so
-^
•H
(U
•M a
CO 3
S T3
d
(U o
-o a
4C *
o O
u
4-1 D./-^
O
CO o O
00 o o o
00
o
o CO
a o> CO
o CM o CM S
(U H eOu o 1—t -vT
CO
w 3 3 U x:
u H

CO 5^ •H
B^S (TV 5^2 5>S CO
H S^S CO
r^ ^s 1—1
d
d CO CM CO
o rO U Q
•H PLI CJ c 3 3
•M c
N
s CL3
•H *\ A o
CO S^ ^ &<« «
o CO ^S a\ s^s &«5 o
00 r*. v£> C3> o
§• CX3
o 3 C <u •H
CO 1*4 a 3
u Z 4J
CO
U
<U

t
•u
c OJ
(U 4J
0) 4J •H 00
CO 4= d
w 00 Cu
d CO
CD 3 U
CO 0) o 0)
CO <U 1 4J 1
•H fH V^ •H M
U CO C a; x: (U
(U CO •H d. a o.
•M CO CO Ou CO a.
CO U o VJ 0 E
u
S 03 zn u o u
 CHAPTER III

EXPERIMENTAL RESULTS

Erosion Characteristics

The change in mass of the spark gap electrodes after 50,000 shots

was measured with an analytical balance with a precision of ± 5 mg.

The individual test conditions and resulting erosion rates are given in

Table III. Although many authors report erosion rates in yg/C, the

actual factor determining lifetime is the volume eroded, hence the

units, ycm^/C. The results for brass are discussed later because of

the failure of the electrodes due to gross material extraction.

Material

A ranking of the volume erosion rate for each material investiga-

ted, from smallest to largest, is:

Cathode: Copper-tungsten, stainless steel, copper-graphite,

graphite

Anode: Copper-graphite, copper-tungsten, stainless steel,

graphite.

As expected, the copper-tungsten composite gave the lowest volume

erosion rate. Somewhat suprising, however, was the excellent perfor-

mance of the stainless steel and the poor performances of the graphite

materials as cathodes. From the results obtained for stainless steel

in a pulsed discharge it is seen that the high erosion rate reported

by Gruber and Suess for an oscillatory discharge was primarily due to

15
 16

Table III. Electrode Erosion Rates

Electrode Gas V Q CE AE

Stainless steel Air 10.3 0.21 1.8 1.2

Stainless steel Air 10.6 0.22 1.5 1.0

Stainless steel Air 18.0 0.37 1.6 1.5
9 3
Stainless steel**' ^2 7.8 0.16 0.7 -0.0

Copper-tungsten Air 9.5 0.20 1.2 0.4

Copper-tungsten Air 11.5 0.24 1.2 0.3

Copper-tungs ten Air 18.0 0.37 1.2 0.5

3
Copper-tungsten 14.8 0.31 0.4 0.4
^2
Copper-graphite Air 8.3 0.17 8.5 0.4

Copper-graphite Air 16.2 0.34 8.6 -K).0

Copper-graphite Air 11.4 0.24 7.2 0.0

3
Copper-graphite 14.8 0.31 13.5 0.8
^2
Graphite Air 9.2 0.19 24.1 3.5

Graphite Air 10.6 0,22 24.6 3.6

Graphite Air 16.0 0.33 23.5 5.0

3
Graphite 12.9 0.27 15.7 0.0
^2

V: Average Voltage, kV; Q: Charge/shot, coulombs: CE Cathode erosion,

ycmVcoul; AE: Anode Erosion, ycmVcoxil; 1 - 32,000 shots, 2 - 22,000
shots, 3 - Experiment performed at approximately 85% of maximum power,
+ indicates that an increase in mass was measured.
 17

its poor performance as an anode material. Previous studies which

indicated that graphite was highly resistant to erosion were done at

a much slower repetition rate (.03 pps) and, thus, gave a significantly

lower erosion rate (< 1 ycmVc). More recent results by Bickford

at 1000 pps gave an erosion rate of 41 ycmVc which is close to the

value of 25 ycm /C measured in this experiment. A summary of the

erosion rates found by other investigators is given in Table IV. If

one takes into account the lower values of current used in this study,

then, in general, the results obtained in this experiment are in good

agreement with the measurements of other investigators.

Polarity

Unlike previous experiments, where oscillatory current conditions

masked any polarity effect, a distinct difference in the cathode and

anode erosion rate and, most likely, the erosion mechanisms themselves

were observed for a unipolar pulse. The ratio of cathode to anode

erosion varied from 1.5 in stainless steel to 16 in copper-graphite.

Carder reported ratios of 2.5 to 5 for brass under similar conditions.

The experiments which gave cathode to anode erosion ratios less than
15 18
one ' were done at much higher pulse repetition rates (10-1000 pps).
18
In addition, the results obtained by Petr were done for relatively

small anode diameters and gap spacings (£2.5 mm).

Cathode erosion rates are plotted in Fig. 8 and show a linear

dependence upon the quantity Q = /idt over the entire range of currents.

This indicates that the main source of energy producing molten material
 18

>^ >.
5-1 5-1 5-1
O CO o o o O
4-> d . 4-1 M CO u U y^ U 4-1
CO P.< CO CO a . CO CO CO CO CO
o rH rH iH a iH r-{ Cu rH
«J-I rH CO r-t o M iH
o a rH
•H O -H •H a •H •H
> o • o •H rH a •H rH O O
CO CO ^—' CO d ^ CO d ^ CO CO
3 o o o D O
o

•u

o CM
o
O O O CM o
U ^ rH «^ O I o
3 V O I
O
<3-

S B
•u u
CO CO ^—N
CO /-~v S y->.
•u rH 4J
m s
4J
rH 4-1 e •"w/ CO CM e
4J CO
c
4-1
3 CO • CO CO
CO rH r-^ rH
(U CM s
3 ? CM
S^' •H s—/
ai CO r-{ CM U 1^ vO
d CO CN d) o •H CM •H
o a 2 K CJ < Z < CO
•H
CO
O
U
Cd
d ^ d 1 — ;
d
CU -fl <u (U 0; cu
iH 4J 4-1
4-1 (U
4J
o;
4J
ja di CO CO
CO ac C/3 00 cn CO
)-l d d
CO 3 CO —> CO 3
a (U H CO <u CO H
6 CO 4J 1 (U u 1 1
o •H •H M rH •H l-l rH U
a x: CO 0) d CO CO (U CO d CU
d. CO •H C-. CO CO 0- m •H CU
!U a d.
4-1 CO CO cu CO CO CO CO a CO CO
o CO !-i U o •u ^ 5.4 5-1 o h u 0
o fQ o V2 O PQ PQ u PQ CA) CJ
s
CO

3 (U
CO /-"^ <f
Pi u • O
C"_
rH
1 in o—1 o
00
o eO iH P-s m vO in 1 1 1 1 o
CO rg in 00
•H
CO 3.
V • o
0 ^ CM
V4
CO w

in <r CO
CO
<u CO
CU
s
CO 3
•-5 (72
o TS
4-) vO 13
CO d d CO
00 in CO CO
•H rH
4-1 •H o d !_i
u
CO d cu •1) <u CO
T3 T3 .Q
> u 5^ !U 3
0) •H 3 CO 5J CO
d PQ PC3 a O
 19

A - Graphite (ACF-iOO)
0 - Stainless steel (304)
1000 .. O - Copper-tungsten (K-33)
• - Copper-graphite (DFP-IC)

100 -
en

en
o
u
Cd

10 -

B N,

7.5 10 15 20

Q = /idt (C x 10^)
Fig.8. Cathode Erosion Rates for Different Electrode
Materials
 20

and subsequent vaporization and droplet ejection is the arc and not
2
localized i R losses in the material.

In general, anode erosion rates were somewhat scattered; however,

some agreement with an erosion rate approximately proportional to

Q * was observed for graphite, copper-tungsten, and stainless steel

in air. A similar dependence on Q * " * has been found experimentally

19—21
and derived theoretically by numerous investigators.

Some anodes actually gained mass which indicated that material

was being transferred from the cathode to the anode. This was proven

experimentally when a stainless steel cathode was found to deposit

molten material on a graphite anode. Gray and Phamey 22 proposed

a reasonable model for this mechanism that is based upon the removal

of the ion bombardment force from the molten cathode material during

the fall of the current pulse.

Gas

The erosion rate for copper-graphite increased slightly in nitrogen,

whereas the rates for all the other materials were smaller, by a

factor of 2-3, in nitrogen. However, because of other factors affecting

spark gap performance, namely the voltage distribution and the

coating of the insulator surfaces, the use of pure nitrogen is not

recommended under the conditions tested.

Surface Conditions

The surface of the electrode tips and the insulator inserts were

examined after 50,000 shots. The analysis techniques utilized were

 21

Auger electron spectroscopy (AES), scanning electron microscopy (SEM),

and optical photography and microscopy. Some of the more general

results are presented here with a more thorough discussion to be given

after further study.

Brass

The surfaces of the brass electrodes are shown in Fig. 9. Large

scale melting is evident with dendrites or metallic protrusions up to

.6 cm long existing on the surface. The self-breakdown voltage for

these electrodes dropped from 20 kV to 3 kV in approximately 2000

shots as a result of the macroscopic field enhancement. In addition,

the voltage distribution was characterized by a series of "jumps"

thought to be due to large particles being "blown" off the ends of the

protrusions. Originally it was thought that the material being

"pulled out" of the bulk electrode was lead, but the results of the

AES analysis shown in Fig. 10 indicate the surface is composed

primarily of carbon, copper, and oxygen with a notable absence of zinc

and lead. From these results and those found by Marches! and Maschio'^

it is quite obvious that brass is limited in its use at higher levels

of charge transfer.

Although the mechanism for the material extraction is not

23
completely understood, Belkin showed that the electromagnetic force

resulting from the discharge can play an important role at large

24
currents. In addition. Fitch and McCormick observed gross material

extraction from stainless steel electrodes as a result of asymmetrical

 22

a) Anode

b) Cathode

Fig. 9. Surface of Brass Electrodes

 23

a) Cathode 250 ym | 1

b) Auger Spectrum

Fig. 10. AES Surface Analysis of Brass Electrodes

 24

current connections.

Cathode

The cathodes for the remaining materials are show in Fig. 11 and

12. Considerable erosion has taken place, especially on the graphite

materials, and the stainless steel and copper-tungsten cathodes show

evidence of severe melting. Although it is not easy to distinguish

from the photographs, all the cathodes showed a distinct tendency to

form a large scale crater whose diameter increases with increasing

gap spacing and current. The idea of using a cathode cup in spark
•JO oc 26

gaps is not new, ' ' but it is interesting that the electrode

erosion results in this shape. It is thought that the location of the

current attachment at the cathode is dependent on the minimxim electrical

path length seen by the electron avalanche prior to breakdown. Thus,

the erosion pattern and the corresponding erosion rate may be highly

geometry dependent.
Anode

The anodes corresponding to the cathodes shown in Fig. 11 and

12 are shown in Fig. 13 and 14. The graphite and copper-graphite

anode erosion occurs primarily in a band, .8 cm wide, located .3 cm

from the center of the electrode. This pattern is consistent with

the results of Johnson and Pfender which showed that an annular-

shaped attachment region of high current density can exist at the

anode.27 The copper-tungsten and stainless steel anodes indicate that

 25

a) Graphite

si»

b) Copper-Graphite

Fig. 11. Graphite and Copper-Graphite Cathode Surfaces

in Air
 26

-.^^

a) Stainless Steel

b) Copper-Tungsten

Fig. 12. Stainless Steel and Copper-Tungsten Cathode

Surfaces in Air
 27

a) Graphite

b) Copper-Graphite

Fig. 13. Graphite and Copper-Graphite Anode Surfaces in Air

 28

a) Stainless Steel

b) Copper-Tungsten

Fig. 14. Stainless Steel and Copper-Tungsten Anode Surfaces

in Air
 29

melting and vaporization has taken place over the entire surface. Like

the pattern at the cathode, the diameter of the anode erosion region

increases with increasing current.

Insulator

A typical insulator insert for each of the eight possible

combinations of electrode material and gas is shown in Fig. 15 and 16.

The insulator surfaces are covered by a coating of recondensed

electrode material. The one notable exception was for graphite

electrodes in air in which no coating was found on the insulator

surface. A dramatic difference is seen in Fig. 15a in the case of a

graphite electrode run in nitrogen. The entire surface is covered

with a thick coating of fluffy black material which is thought to

consist of monoatomic layers of amorphous carbon. 28

All the insulators were covered with solid particles, 10-lOOym

in size, distributed within a 5 cm and centered on a plane passing

through the center of the gap and parallel to the electrode surfaces.

This indicates that a considerable portion of the solid or molten

29
material is ejected parallel to the electrode surfaces. Daalder has

reported similar results for vacuum arcs. McClure developed a model

which shows that the ion recoil pressure of a vacuum arc plasma is

sufficient to remove molten material from a cathode spot crater with

3 4
velocities of 2 x 10 - 2 x 10 cm/sec parallel to the electrode
30
surface. The values of velocity from McClure's model are in good
 )0

Z'
\

Air Nitrogen

a) Graphite

• : ^ ^ ^ ^ ^ ^ ^ '

Air Nitrogen

b) Copper-Graphite

Fig. 15. Insulator Inserts Exposed to Graphite and Copper-

Graphite Electrodes in Air and Nitrogen
 31

^*
SsMii&i^

V
Air Nitrogen

a) Stainless Steel

.0^ /'-«^"'(-'.«W«^,
, 0(i»

Air Nitrogen

• • - — .. w -

b) Copper-Tungsten
Fig.16. Insulator Inserts Exposed to Stainless Steel and
Copper-Tungsten Electrodes in Air and Nitrogen
 32

31
agreement with the experimental findings of Udris.

An SEM examiniation of the surface of the stainless steel

electrodes shows a remarkable similarity between the electrode surface

features shown in Fig. 17 and a 50 ym stainless steel particle

found on the insulator and shown in Fig. 18. The presence of

similar particles has been shown to have a serious effect on the

flashover potential of the insulator at high pressures for particle

sizes greater than 35 ym and densities of 20 particles/mm^. Thus,

the electrode erosion mechanism affects the switch lifetime not

only as a result of the erosion itself, but also by coating the

insulating materials with conductive particles.

Self-Breakdown Voltage Distribution

The self-breakdown voltage of the spark gap was recorded

continuously for the first 2000 shots and thereafter sampled at intervals

of 10,000 shots. A sample of approximately 400 shots was taken and

used to calculate the mean breakdown voltage (V) and the standard

deviation (a ) of the self-breakdown voltage.

Material

The standard deviations for different electrode materials are

shown in Fig. 19 and 20. In an air atmosphere the graphite electrodes

have the smallest a and, therefore, the most narrow distribution of

breakdown voltages. This combination is followed by copper-graphite,

stainless steel, and copper-tungsten, respectively. Preliminary

 33

o
M
^
cc
r-t
d
W <u
01
a
CO
cc I+-.
T3 5-1
fc-I 3
C/0
V^
a a
4-1 T3
3 O
C lu
4-1

a
^ - N
a;
w
o
4-1
CO

CO
CO
a;

CO
v ^j .«.'.v,.. , .:,.... 4J
•.'.-'-..3 '' /^ th v.- JP5 C/2

p-?
III C/3

i i
•^/-^•l.',->.;^i^
1
y^
Mr *JS«
0)
QC

u
5.1
(U
4-)
3
O

CO
00
•H

-•Ml

'•, • • * ' • . . * 1'

 34

.-^i^fb. ,^v4<fl»»*^ -V :f'

^/

e03 ^ioK^^p:f.»*:

Stainless Steel Particle

Fig. IB. 50 ym
on Lucite Insulator
 35

-IC)

10,000 20,000 30,000 40,000 50,000

Number of Shots

Fig. 19. The Standard Deviation of the Self-Breakdown Voltage

vs. Shot Number for Different Electrode Materials in
Air
 36

A - Graphite (ACF-IOQ)
2.25, O - Stainless steel (304)
^ - Copper-tungsten (K-33)
2.00- 0 - Copper-graphite (DFP-IC)

1.75

10,000 20,000 30,000 40,000 50,000

Number of Shots

Fig. 20. The Standard Deviation of the Self-Breakdown Voltage

vs. Shot Number for Different Electrode Materials in
Nitrogen
 37

examination, of the electrode surfaces with SEM and optical microscopy

indicated a direct relationship between the width of the distribution

and the field enhancement due to the different microscopic surface

conditions for each material. Typical surface features for stainless

steel and copper-tungsten are shown in Fig. 21. The copper-tungsten

surface is characterized by a rough layer of what appears to be re-

solidified beads of electrode material approximately 10-100 ym in size.

The stainless steel electrode is considerably smoother and has a uni-

form pattern of resolidification approximately 100 ym in size, covering

most of the surface.

Many of the models which describe the effect of electrode rough-

ness on breakdown voltage state that the cathode surface conditions

33-35
are extremely important. Therefore, an experiment was performed

to verify this dependence for two materials which were known to have

contrasting surface structures after a few discharges. The self-

breakdown voltage for the first 400 shots was recorded for a stainless

steel cathode used with a graphite anode and also for a graphite

cathode used with a stainless steel anode. The voltage distributions,

shown in Fig. 22, indicate that the stainless steel cathode yields a

voltage distribution roughly twice as wide as the graphite cathode.

Thus, as expected, the cathode surface conditions resulting from the

processes producing electrode erosion can greatly affect the self-

breakdown voltage distribution of a spark gap.

Shot Number

In both air and nitrogen the mean breakdown voltage decreases

5-20% during the first 15,000 shots and then increases at different
 38

a) Copper-Tungsten Cathode, 100 ym I 1

b) Stainless Steel Cathode, 100 ym V—I

Fig. 21. Optical Microscope Pictures of Typical Surface

Features for Copper-Tungsten and Stainless iteol
 39

22.5 -I
A - Stainless steel cathode
20.0 . ^ - Graphite cathode

17.5 -

15.0 -

CO
4-1
o 12.5 -
.d
<4-l
o 10.0 -
5-1

3 7.5 -

CO
4J
o 5.0 -
H
14-4
O
2.5 -

0.0 A
0 2 4 6 a 10 12 14 16 18 20 22 24

Voltage (kV)

Fig. 22. Self-Breakdown Voltage Distribution for

Stainless Steel Cathode with Graphite Anode
and Graphite Cathode with Stainless Steel Anode
 40

rates, depending on the electrode material. This increase in V is a

result of the electrode erosion producing an increase in the gap length.

Also, in Fig. 19 it is seen that, except for stainless steel, a remains

fairly constant after 20,000 shots. This suggest a conditioning process

during the first 10-20,000 shots where the tip of the electrode is

worn away and the electrode surface conditions required to initiate a

discharge become more uniform. It is thought that the conditioning

process for stainless steel occurs on a larger time scale, and thus,

was not completed after 50,000 shots.

Gas

Data from the same experiments in a nitrogen atmosphere, shown

in Fig. 20, indicate that the best electrode materials for a narrow

distribution are copper-graphite and graphite. Complete results for

stainless steel are not yet available. The large changes in a for

both graphite and copper-graphite are due to low voltage dropouts

where the breakdown was often as low as 50% of the mean. These

dropouts occur primarily in nitrogen, with increasing frequency after

20-30,000 shots. In the case of graphite a macroscopic growth, shown

in Fig. 23, was found on the anode, which may have caused the dropouts.

The nature of these dropouts is of extreme importance in a large scale

system with many spark gaps, since the premature firing of one gap may

be quite costly, or at the very least extremely time consuming. In

fact, the probability of a successful test may approach zero for a

system with a large number of gaps. Thus, these results, as well as

those obtained by Affinito et. al., indicate that the combination of

graphite electrodes and nitrogen gas should be avoided.

 41

a) Growth on Graphite Anode, 1mm

b) Growth on Graphite Anode (Enlarged), 1mm I 1

Fie. 23. Macroscopic Growth on Graphite Anode in Nitrogen

 CHAPTER IV

CONCLUSION

The erosion rates and self-breakdown voltage distribution were

determined for several materials utilized in high energy spark gaps.

The results from these preliminary studies have led to the following

conclusions:

1) The electrode erosion rates and mechanisms are highly

polarity dependent and, thus, results for oscillatory
and unipolar discharges can be considerably different.

2) A large amount of the erosion is in the form of solid

and molten material removed parallel to the electrode
surface.

3) Cathode erosion rates are proportional to the total

amount of charge transferred.

4) Stainless steel may be an economical replacement for

copper-tungsten composites as a cathode material for
the conditions studied.

5) Graphite composites give narrow self-breakdown voltage

distributions but have very high cathode volume erosion
rates.

6) The stability of the self-breakdown voltage is dependent upon

the electrode micro-surface structure which is a distinct
property of the electrode material.

7) Certain electrode-gas combinations lead to "prefires"

or voltage breakdowns as low as 50% of the mean. One
of the worst combinations investigated was graphite and
nitrogen.

42
 43
8) Coatings on the insulator, which may or may not be
detrimental to the spark gap operation, are dramatically
different for different combinations of gas and electrode
material.

In order to develop a more precise understanding of the

effects of electrode erosion on switch performance the following

objectives are being considered for future work:

1) Measure erosion rate as a function of pressure

-2
(10 to 4 atm), and rep-rate (1 - 1000 pps) for a few
of the more promising electrode-gas-insulator combinations

2) Characterize electrode surface features for a given

electrode material as a function of electrical parameters.

3) Develop a model which accurately accounts for the self-

breakdown voltage distribution as a function of electrode
surface conditions.

4) Describe the mechanism which produces droplet ejection

parallel to the surface and characterize its chief
features (particle size, particle flux, and particle
velocity) as a function of material and electrical
parameters.
 LIST OF REFERENCES

[IJ T.R. Burkes, et. al., "A Critical Analysis and Assessment of High
Power Switches," NSWC Dahlgren Lab, Report NP 30/78, 189-202,
(1978).

[2J L.B. Gordon, et. al., "Material Studies in a High Energy Spark
Gap," to appear in IEEE Trans, on Plasma Science, PS-11, (1982).

[3] J.M. Meek and J.D. Craggs, Electrical Breakdown of Gases,

John Wiley & Sons, (New York: 1978) 839.

[4] J.E. Gruber and R. Suess, "Investigations of the Erosion

Phenomenon in High Current, High Pressure Gas Discharges," Max
Planck Inst, fur Plasmaphysik, Garching bei Munchen, IPP 4/72,
(Dec, 1969).

[5] R.A. Burden and T.E. James, "Statistical Performance Data for a
High Current 60 kV Spark Gap Switch," Proc. 7th Symp. Fusion
Technology, Grenoble, France 24-27, (Oct. 1972).

[e] G.S. Belkin, and V. Ya. Kiselev, "Electrode Erosion in Pulsed

High-Current Discharges," Sov. Phys. Tech. Phys., 2i, 28 (1966).

[7] G. Marchesi and A, Maschio, "Influence of Electrode Materials on

Arc Voltage Waveforms in Pressurized Field Distortion Spark
Gaps," 5th Int. Conf. on Gas Discharges, (Sept. 1978).

[S] Y. Kawakita, et. al., "A 150-kV, 100-kA Spark Gap Switch for Marx
Generators," 3rd IEEE Int. Pulsed Power Conf., Albuquerque, N.M.,
444, (June 1981).

[9] D. Bold and P.M. Barnes,"MK IV Spark Gap Switch," Pulse

Technology Memo 22, UKAEA Culham Lab, Abingdon, Oxfordshire, U.K.
(12 Nov. 1973).

[XO] B. Carder, "Gas Spark Gap Electrode Heating and Erosion," Physics
International Report, P U R 12-74, (Dec. 1974).

[11] R. Basharov, et. al., "Erosion of Cathode Material in a Pulsed

Discharge Between Parallel Electrodes," Sov. Phys. Tech. Phys.,
22, 1383 (1966).

[12] G.N. Glasoe and J.V. Lebacqz, Pulse Generators, Dover Publishing
Co., (New York: 1965), 175.

[JL3] Schwarzkopf Development Corporation - Holliston, Massachusetts.

IJ4] Poco Graphite - Decatur, Texas.

44
 45

[l5] K.J- Bickford, et. al., "Spark Erosion Chacteristics of

Graphite and CO Gas," 15th Power Modulator Symposium, Baltimore,
MD, (June 1982).

[l6] D. Affinito, et. al., "Design and Structure of an Extended Life

High Current Spark Gap," IEEE Trans, on Plasma Science, PS-7, 162
(1979).

[17] G.S. Belkin, "Vaporization of Metal Electrodes by Pulsed

Currents," Sov. Phys. Tech. Phys., 21, 1256 (1969).

[is] R.A. Petr, "Erosion Phenomena of Arcing Electrodes," MS Thesis,

Texas Tech University, May 1980.

[19] H.W. Turner and C. Turner, "Choosing Contact Materials,"

Electronics and Power, j ^ , 437 (1968).

[20] E.M. Williams and R.E. Smith, "Phenomena Accompanying Transient

Low-Voltage Discharges in Liquid Dielectrics," lAEE Trans. Part
I, 1±, 165 (1955).

[21] V.E. Il'in and S.V. Lebedev, "Destruction of Electrodes by

Electric Discharges of High Current Density," Sov. Phys. Tech.
Phys., 2» 717 (1963).

[22] E.W. Gray and J.R. Phamey, "Electrode Erosion by Particle

Ejection in Low-Current Arcs," J. of Appl. Phys., 4^, 667
(1974).

[23] G.S. Belkin and V. Ya. Kiselev, "Effect of the Medium on the
Electrical Erosion Electrodes at High Currents," Sov. Phys.
Tech. Phys. 21, 24 (1978).

[24] R.A. Fitch and N.R. McCormick, "Low Inductance Switching Using
Parallel Spark-Gaps," Proc. lEE (London) 106A, 117 (1959).

[25] F.S. Goucher, et. al., "Spark Gap Switches for Radar," Bell Sys.
Tech. Journal, 25., 563 (1946).

[26] A.E. Bishop and G.D. Edwards, "Low-Inductance 100 kV Switch

(Spark Gap) for Starting, Diverting, and Clamping Capacitor
Discharges," Proc. lEE, 213, 1549 (1966).

[27] D. Johnson and E. Pfender, "Modeling and Measurement of the

Initial Anode Heat Fluxes in Pulsed High-Current Arcs,"
IEEE Trans, on Plasma Science, PS-7, 44.

[28] L-B. Gordon, "Material Studies in a High Energy Spark Gap",

Ph.D Dissertation, Texas Tech University, in preparation.

[29] J.E. Daalder, "Cathode Spots and Vacuum Arcs," Physica 104C, 91
(1981).
 46

[30] G.W. McClure, "Plasma Expansion as a Cause of Metal Displacement

in Vacuum-Arc Cathode Spots," J. Appl. Phys. 45^ 2078 (1974).

[3l] Y. Udris, "On the Emission of Cathode Material Particles in Low

Pressure Arc Discharges," Proc. of the Int. Conf. on Gas
Discharges, London, (lEE London, 1970 ) 108.

[32] B.F. Hampton and S.P. Fleming, "Impulse Flashover of Particle

Contaminated Spacers in Compressed Sulfur Hexafluoride,"
Proc. lEE, 220, 514 (1973).

[33] V.A. Aurutskii, et. al., "Effect of Electrode Roughness on the

Electrical Strength of Compressed Gases," Sov. Phys. Tech. Phys.,
21, 386 (1973).

[34] A. Pedersen, "The Effect of Surface Roughness on Breakdown in

SF^," IEEE Trans, on PAS, PAS-94, 1749 (1975).

[35] I.W. McAllister, "The Influence of Electrode Macroscopic

Curvature upon Surface Roughness Effects in Compressed SF,,"
Archiv fiir Elecktrotechnik H , 43 (1980).

[36] T. Martin (private communication).

 APPENDIX

MANUFACTURE AND PREPARATION OF ELECTRODES

The procedure for the manufacture and preparation of the

electrode drawn to scale in Fig. 24 is given below.

1) Cut rough tip approximately 1 1/2" long

2) Machine one end of tip flat

3) Bore drill guide hole with drill center tool

4) Drill 9/32" hole 3/4" deep

5) Start 5/16" tappered tap with tip still in the lathe using
drill chuck

6) Remove tip then tap 5/16 #18 threads

7) Re-tap tip using 5/6 #18 bottoming tap

8) Machine 1/6" seat on machine-faced end, with the diameter

to match specific tip holder being used. (Set of electrode
tips consist of two tips matched to two different tip
holders)

9) Attach tip to tip holder with machine screw, tighten

tip holder and tip onto shaft, wrap computer card around
shaft to prevent scratching, slide entire assembly into lathe
chuck, center and tighten

10) Machine outside of tip to diameter matching tip holder

11) Set up radius cutting tool and machine tip surface into
a smooth hemisphere (Note: a different cutting tool is
used for each electrode material to avoid contamination)

*12) Sand entire tip with #240 sandpaper

*13) Apply a fine finish with #600 sandpaper

47
 48

14) Remove tip from tip holder and place in plastic storage
box

15) Clean with methanol before installing in spark gap

Caution: do not "drag" tip holder metal down over tip. Sand
from tip center back towards tip holder. Do not
touch tip surface with hands. Wipe clean with clean
tissue.

/ 3 ^
GAS PORT WRENCH SOCKET MACH. SCREW

ELECTRODE
ELECTRODE SHAFT ^LECTRODE_^
-• TIP >
TIP
HOLDER

BEFORE AFTER
I I '-..'.•>

I '

1 / 16
1/2' -1/8
I ROU
ROUGH CUT FINISHED TIP

Fig. 24. Electrode Assembly