l

V . - an
J.§4FM’ ‘ ‘
‘5. .. ,Iy

' fix: ‘J ‘
,_ ‘p

. " 4- [f

Y. —_r
1.
f .53!»

.-‘~
. "

.0 .h'~ I I
L

.. .
'I
Ix} ,
IE.
11 » .
2| Q '. I
‘ I ...I.
" ,/:‘IT",.
I. . ' I...-
. l I"“‘ ,;"- ‘lA' II ‘ _ ,vt
I-f- . . ._ ‘, ':I:_ . ‘Ii, ‘- +‘ , 1II’WV‘L‘J.'
' ,.:.

I

, . ..‘ , ., . .

..I-“.," #: I’Y‘ l‘ ' :"q‘ l'*.f+;‘t

I. I‘ . . 4' 'y.‘\"
l .‘ . n “‘0" :Iflfi‘ '.. '

' - '3 ' will.

"'. ,1) a

. l; W .

4' I? WF‘“ “"h‘iuWJ I

“.ql...‘ ‘f' n"
F I 1.”...

‘I
0" I

-q
. ——&.-.‘. ~_-_ ‘ '_

. ’U ‘..‘
:yll I "r'ls ht} _. 2:}: 3.

.‘ II .‘
‘ ‘ ‘ ' I
b ‘ . ...u. .I‘ ' . .._’ ' . ‘ ‘ ‘
. -. I". {r .I - f“ .- .’.' _ H ‘ H
'.’0 II , .. ‘ . , 7. .l H _- . I
l' ' ' ' ' ' ”4!." 'I ‘I ‘ - ‘ -. .. ‘ ,
. . ‘ D ‘ q“
. -’. 1 I. I . ‘ .. - ‘ ‘ W
l - ,,
' ‘l .\.,.o - . :_ ‘ . ,

...'I- ':", v | ' ,Iw _ H‘ ',.I , l,\ . .’ l‘ I ‘ l' .‘ f
‘ . , _,‘ ' '..-" . | "31‘ :wl'l .' _ _‘ .. . . 3‘“ . ‘ .2. 9.... 1"“ hié -. I
' ."l f‘ 1;; ‘1'“ '_ ‘i-T.‘ l,“ 5'" I“. I - :37" " I" '.' :. ..:.' ‘ ‘ _. f ‘*K{ lybfi’m]
9. . ‘ :4“. .. ‘3 .13; '. ' _ f -:‘--_‘__ -. - 'hj;‘.“:?I-:. 4... a.” 5g] .' \
J". - . _ 1' i' "CAN ‘ 33):. ' {h ‘ ” ' "'1', .« I, .
I "

l

. ‘.\!6:;:':-' ‘1' ' . I
'Ifimo'mf “‘4 a (léklf' £3

r

- V
._ L
- “0-0 c 0500...;

‘\I u
‘— I

I"... :g‘:
. -' ‘1" ...\ I ‘11}, l:
h I ;‘.'_‘& \.:’D' ‘E'QWI: a"; 'n- 3:”?
‘llITl’M "5‘1“ 1&9‘ "pv‘ hjgk'.‘:& :‘u 'K'QF )lfiufMI)
. . .I:l1: ' I
”ll

-. "In" Inst...
r;
.

0 I

[EV.U ;$;f,4
:'¥-.: l'lP‘.\+ I"

.I I‘

I... 331‘.

. «I. l.‘ "0- If: V N >':&~(" 4.14:6...
“ ‘ - .; ~ — “m fit: .---:3 :2..v;....I .,.
l 31 .9 ._.. .’}‘ . —wm"l’3“rfiw ,7;
l '. . .. . . , 1‘» ‘, fiz:’\.. 25:5 $3.94“ ; 1‘“.
l l I l I

S

.,. ., ' ‘-l‘." .‘I'QEFKH'ILR
3' " .. MVP ‘J . l“
" ‘I 0%“..‘Wrm‘lc l 1‘
' I ‘- . .3: 93;.) :39“.
- r. v.1 .‘ "- [l
. ‘-.
l l " .l.’ '_ \w-ll"
‘ 9', " r I

 

5.; L‘v.)
9-.“ “(I ‘ 22'

1‘59??? .
2’, ,II
“a,

 

c;.

 

1'73; . "T

v

.I'_ {I ‘
: ”Ill

4...
'1

..-
..~

[I
m
A

if
i

dag}-
-33;
l-’*“‘. ’

$25
- -%;~‘ '1
M

i.

"2‘! T“
“Hg-1‘

"-‘4 AI‘ - a” .1 ; '1» 1:- -.
.
.
.-I
3

03, ~.
A

75".“0'
’ffinlxazq g;

:'~

‘
-t'

.

llllllllllH!lllHlHll‘ll‘lllllllllllllllil'lllllllmllll

3 1293 10_789 1420 ii

I,
T
s ‘t . g
a j I
.
hip—w-qu won-'0 ‘d-v‘ a
*-

v“--

[H ‘ 9,0
.'
L‘-‘-';~w"l

This is to certify that the

dissertation entitled
The Development of a Discharge Heated

Penning Ion Source for Multiply-charged
Ions in a K=SOO Superconducting Cyclotron

presented by

Timothy Allen Antaya

has been accepted towards fulfillment

 

 

of the requirements for
Ph . D. degree in BMW
1";
77 _ ’h/ ,
' ”H4;
Major professor

M.M. Gordon

Date 9-14—84

MS U is an Ajflrman’n Action/Equal! Opportunity Institution 0-12Tll

 

 

 

MSU

LIBRARIES
._c-

 

 

., --v.---( -_-t- a... fl“. V" V

rmw‘r ’.‘ W“

\” “I"
a» r
fir 1.
r-‘r

RETURNING MATERIAtgz
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

IF!
0..

_ ‘7. a,“ in?" ~95“;

Irv

 

 

THE DEVELOPMENT OF A DISCHARGE HEATED
PENNING ION SOURCE FOR MULTIPLY~CHARGED
IONS IN A K-SOO SUPERCONDUCTING'CYCLOTRON

BY

Timothy Allen Antaya

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Physics

198”

ABSTRACT
THE DEVELOPMENT OF A DISCHARGE HEATED

PENNING ION SOURCE FOR MULTIPLY-CHARGED
IONS IN A K-SOO SUPERCONDUCTING'CYCLOTRON

BY

Timothy Allen Antaya

We present the studies made during the design, testing,
operation and development of a PIG ion source for the
production of multiply-charged ions in the K500 Cyclotron.
In Chapter 1 the fundamental properties of Penning (PIG) ion
sources and the design of the K500 cyclotron ion source are
presented.

Pulsed operation of the source, including the power
supply Operation and modification, the ion source Operating

characteristics, and the effects on the production of Ne5+

and Ar6+ ions, is discussed in Chapter 2. At low duty
factors (< 30%) the source output of these ions increases
five to ten times over do Operation. The instantaneous arc
current and voltage are both several times their dc levels,
resulting in a large instantaneous arc power that drives the
increased ion production.

The use of hafnium as a cathode material is discussed
in Chapter 3, where analyses of lifetime and output effects
are made. Replacement of the normal tantalum cathodes with

hafnium resulted in a factor of four increase in source

Timothy Allen Antaya

lifetime for nitrogen discharges, due to the formation of a
HfN layer having a lower sputtering rate. For dc operation,
the ion output is reduced due to a large vapor pressure of
Hf inside the source, thereby increasing the loss of ions by
charge exchange.

An ion source activation hazard, due to the stripping
of accelerating ions, is discussed in Chapter u. We review
the model for this process and then show how the addition of
tantalum cover plates to the ion source reduces the
activation of its copper surfaces.

In Chapter 5, the modifications made to the source to
produce ions from solids via the back-bombardment sputtering
process developed at the Oak Ridge National Laboratory, as
well as the successful testing with lithium and boron ions,
are reviewed.

In Chapter 6, the results of a measurement of the
effect of varying the ion source cathode separation on ion

5+ .
ions decrease

production are shown. Outputs of Nu+ and N
as the cathode separation increases, and this is best
explained as a shift in the arc alignment due to magnetic

field curvature Off the axis of the cyclotron.

ACKNOWLEDGMENTS

The development of the ion sources required the
assistance Of individuals spanning nearly all of the areas
of expertise at the National Superconducting Cyclotron
Laboratory. My thanks go out to the laboratory as a whole
with special acknowledgments to the following individuals.

I would like to thank Dr. M. 1.. Mallory fWH~ an
introduction to ion source physics, which led directly to my
involvement in the development of the K500 ion sources.
Futhermore, Merrit's drive and enthusiasm has made our
collaborations truly memorable.

TO Dr. H. G. Blosser, director of the National
Superconducting Cyclotron Laboratory and responsible for the
exciting projects that we have had to work on, I would like
to thank for his advice and direction as we tackled them.

A special thanks to Dr. M. M. Gordon, who was my chief
academic advisor, for his constructive suggestions as to the
presentation of this work, and for his assistance in the
writing Of this dissertation.

I would like to acknowledge Drs. E. Kashy and

S. M. Austin, who served as the cyclotron research directors

11

during the first and second Operating periods respectively,
for their support for the inclusion of ion source
measurements in the operating schedule of the K500
cyclotron. Dr. Kashy was also the prime motivator for the
choice of topics included in this dissertation.

In addition to Dr. Mallory, I had other collaborators
for the work contained in this dissertation. Mr. J. Riedel
was responsible for the power supply modifications that
enabled us to run the ion sources pulsed, as well as giving
us advice on many other electronics hardware issues.
Dr. P. S. Miller and his cyclotron Operations group: D. Poe,
M. Garbicz, 0. Horner, G. Humenik, N. Friedman and
L. Wilkinson, were intimately involved in the Operation of
the ion sources, and their discussions and insights have
helped shape this work. Dr. Felix Marti's advice on
cyclotron tuning and other matters was always valued. I
would like to thank Gary Horner and Bruce Milton for the use
Of the data in Figures 2.7-2.9.

A substantial mechanical design and fabrication effort
was required in the development of the sources.
Mr. J. A. Kuchar was the chief mechanical designer,
contributing both to the initial source design, and also at
each step in the develOpment process, Often supervising
directly the fabrication and installation. It is a real
pleasure to work with him. I would like to thank

Mr. J. Kitsmiller and the rest of the machine shop crew who

iii

worked diligently to supply us with ion sources parts that
we so quickly wore out!

Good friends make all of this worthwhile, and I have
some of the best: Charles and Judith Filice, Donald and
Suzzanne May, Edward and Dianne Rister, Kenneth and Linda
Steibel.

My parents, John and Patricia Antaya, through early
encouragement of my interest in science, did start me out
well on the path that led to this degree, and so this is
their degree too. In the same way I do greatly value all
the help and encouragement given by Douglas and Joyce
Martyn. I have a very special grandmother, Agnes Kennedy,
who has always been a strong supporter.

One acknowledgment is difficult to express in words,
because nothing I would say could convey completely the
importance that Ann Louise Martyn, my wife, runs been to my
studies, and so it is to her that I dedicate this
dissertation. Thank you Ann, for giving me the support,
encouragement and love that was necessary to complete this

work.

iv

TABLE OF CONTENTS

Chapter Page

LIST OF TABLES . . . . . . . . . . viii

LIST OF FIGURES . . . . . . . . . . x

1.0 INTRODUCTION . . . . . . . . 1
1.1 Fundamental PrOperties of the

Penning Ion Source . . . . . . u

1.1.1 The Arc Discharge Mode . . . u

1.1.2 Production of
Multiply-charged Ions. . . . 11

1.2 The Design of the K500 Cyclotron
PIG Ion Source and the Operating
Environment. . . . . . . . 17

1.2.1 K500 Cyclotron Pig
Ion Source Design . . . . 18

1.2.2 K500 Ion Source
Commissioning Experience . . . 23

1.3 Publications in Connection with
this Research . . . . . . . 29

2.0 PULSED ION SOURCE OPERATION; NEON
AND ARGON BEAM PRODUCTION . . . . . 30

2.1 Review of the Process and
Anticipated Gains . . . . . . 32

2.2 Power Supply Operation for PIG
Sources and Modifications to
Implement Pulsing . . . . . . 35

2.3 Pulsed Operating Characteristics

5+ 6+

for Ne and Ar . . . . . . HO

Chapter

2.1 The Mechanism for Multiply-Charged
Ion Production in Pulsed Arc
Discharges . . . .

2.5 Source Lifetime with Pulsed
Neon Discharges .

2.6 Summary .

HAFNIUM CATHODES AND NITROGEN
SUPPORTED ARCS . . . . .

3.1 Choice of Cathode Materials

3.2 Initial Use of Pure Hafnium
Cathode- A review .

3.2.1 MSU K50 Cyclotron .

3.2.2 Experience at Other
Laboratories . .

3.3 Hafnium Cathodes in K500
Cyclotron PIG Sources

3.” Analysis of Lifetime and Beam
Current Effects .

3.4.1 Cathode Lifetime

3.A.2 Hafnium Cathodes and
Beam Current

3.5 Summary and the Impact of
Hafnium Cathodes .

ION SOURCE RADIOACTIVATION CORRECTION

U.1 Review of the Ion Source
Activation Model

u.2 Removable Median Plane
Tantalum Source Shielding

THE PRODUCTION OF IONS FROM SOLIDS

5.1 Origins of the Back-bombardment
Sputtering Technique .

5.2 Design Considerations for the
K500 Ion Sources

vi

Page

53

6O

65

68

72

7M

7D

77

77

86
86

88

95
98

101

10“

Chapter Page

5.3 Production of Boron and Lithium Ions . 118
5.U Analysis of the 7Li2+ Ion
Production Dynamics . . . . . 122
5.5 Summary and Conclusions . . . . 129
6.0 MEASUREMENT OF ION SOURCE OUTPUT

VERSUS CATHODE SEPARATION . . . . . 131

6.1 Design of the Variable Arc Length

Ion Source . . . . . . . . 137
11+ 5+
6.2 Measurement of N and N
Currents versus Cathode Separation . . 1A2
6.3 Analysis and Conclusions . . . . 1A6
7.0 FINAL PERSPECTIVE . . . . . . . 1H9
LIST OF REFERENCES . . . . . . . . . 155

vii

LIST OF TABLES

Table Page

Table 1.1. A comparison Of the two useful ion 11
production modes of Operation of the Penning Ion
Source is made.

22 5+

Table 2.1 The characteristics of Ne and 53
H0 6

Ar +beams extracted from the K500 cyclotron with
do and pulsed arc discharges in the ion source.
The standard deviation is given in parentheses.

Table 2.2. Comparison of cathode sizes and plasma 62
apertures in the Texas A&M and MSU ion sources.

Table 2.3. Cathode wear characteristics with neon 6A
feed.
Table 3.1. K500 operation with Hf cathodes. 79

Lifetimes for tantalum cathodes are given where
available.

Table 3.2. A tabulation of the main 80
characteristics of K500 ion sources Operated with
Hf cathodes.

Table 3.3. Hafnium and tantalum cathode 83

sputtering rates with argon gas feed for 10

sources, used during the develOpment of Ar6+ ions.

Table 3.“. Nitrogen 5+ beam currents with hafnium 8A

and tantalum cathodes. N2 feed is assumed in all

cases.

Table 3.5. Comparison of the range of dc are 85
parameters for sources equipped with hafnium or
tantalum cathodes,

5

when used to produce N +ions in the K500 cyclotron.

viii

Table

Table h.1. Calculation of the activation expression
of Eq.(u.1), with E(r) - Ecb' for selective beams

of the RCNP Laboratory K-120 cyclotron, to determine
whether that process could be the source of their
reported ion source activation.

Table 5.1. Beams chosen for the initial testing
Of the source modifications made to incorporate
the back-bombardment sputtering technique into
the design.

Table 5.2. The first beams in the K500 cyclotron
produced by back-bombardment sputtering of solid
feed inserts.

Table 6.1. The beams that were selected for the
variable arc length ion source experiment. The

N u+at 21.9 MeV/n and N5+ at 35 MeV/n had the same

magnet settings.

ix

Page

109

120

1N2

LIST OF FIGURES

Figure

Figure 1.1. The first harmonic Operating diagram
of the K500 cyclotron is shown. The central
magnetic field is plotted as a function Of the
cyclotron resonance frequency, with constant dee
voltage and Q/A lines indicated. The final beam
energy in MeV/n is also given.

Figure 1.2. The PIG ion source are discharge mode
is shown in schematic form.

Figure 1.3. The Operating diagram for the PIG ion
source is shown for operation on argon gas, from
reference [Be 72].

Figure 1.fl. The total ionization potentials from
neon and argon are plotted as a function Of the
number of electrons (Q) removed for the atom,
according to reference [Ca 70].

Figure 1.5. The rise time of argon 1+ through u+
ions as a function of time is shown for a pulsed
argon arc discharge from reference [Pa 72]. The
time t-O corresponds to the arc current coming on
in the source.

Figure 1.6. A sectional view through the anode of
the head Of the K500 PIG ion source is shown.
Compare this to the schematic PIG source diagram
in Figure 1.2.

Figure 1.7. The central region geometry for first
harmonic acceleration is shown. The ion source
and electrodes crossing the median plane are shown
as shaded areas. The equipotentials shown here
correspond to all three dees having the same
voltage and indicate the six acceleration gaps
(from reference [Ma 81]).

Figure 1.8. The dependence Of the arc voltage on
the arc current and gas flow is shown for the K500
cyclotron PIG ion source, for a carbon dioxide
arc. The decreasing voltage with increasing arc

Page

13

in

20

22

24

Figure Page

current or gas flow is characteristic of this
discharge mode.

Figure 1.9. The increase in the average ion 26
source lifetime with the number of runs, during
the early development of the source is shown.

Figure 2.1. In this schematic diagram is shown 38
the ion source power supply circuit. 0n the right

there is an Eimac HO0CW100D tetrode and its

filament, grid and screen supplies, for the

purpose Of regulating the current for the high

voltage supply Vs that powers the ion source.

Figure 2.2. Pulsed ion source are voltage and arc A1
current waveforms.

Figure 2.3. The dependence of 22Ne5+beam current AM
on the pulse on-time is shown for a fixed pulse

length of 3.0 msec and are currents of 6.0 and

9.5 A.

Figure 2.“. The dependence Of the 22Ne5+ beam US

current on the peak arc current is shown, for
fixed duty factors of 17 and 33%.

Figure 2.5. The discharge heated PIG source N8
Operating can be separated into regions of

stability and instability during pulsed operation.
Oscillograph (B) shows normal arc current and
voltage waveforms. Oscillograph (A) results from

a large decrease in the gas flow. This condition is
unstable- eventually the arc drops back to the

cold cathode mode and the production of
multiply-charged ions will cease. Oscillograph

(C) results from a large increase in the arc
current. Evidently this large arc current is not
neutralized by positive ions from the plasma, and
twice during each pulse the plasma column is seen

to collapse.

Figure 2.6. The gas flow rate into the ion source 50
is a sensitive parameter for the production of

multiply-charged ions. In this case the 22Ne5+
extracted beam current is shown for neon gas (70%
enriched 22-neon) source feed, and also for an arc
supported by nitrogen with a constant neon flow of
0.25 sccm.

xi

Figure

Figure 2.7. Nu+ extracted beam current is plotted
versus average are power, for fixed duty factors
of 50, 75 and 100%.

Figure 2.8. Nu+ extracted beam current is plotted

versus peak arc current for 50, 75 and 100 % duty
factors, corresponding to those same cases in
Figure 2.7.

Figure 2.9. Nu+extracted beam current is plotted

versus peak arc voltage for duty factors of 50, 75
and 100%, corresponding to those same cases in
Figure 2.7.

Figure 3.1. A schematic representation Of one end
of the PIG arc geometry, showing the anode end,
are defining aperture, and the cathode. The
effects of cathode sputtering are illustrated.

Figure 3.2. The thermionic emission currents
densities of pure hafnium and tantalum are
compared over the temperature range of
1500-3000 °C.

Figure 3.3. 55 hour lifetime hafnium cathodes for

N5+ show small molten domes at the base of the
craters and crumbled edges- the latter evidence of
a structure changing reaction of the hafnium and
nitrogen gas. New hafnium cathodes resemble other
refractory metals.

Figure 3.U. With hafnium cathodes, the N5+current
drops with increasing ion source duty factor. In

this case, the N2 flow rate was 1.5 sccm and the

arc current 0.6 A.

Figure 3.5. A history of the maximum cyclotron
extracted beam current versus source duty factor

for N5+ runs and hafnium or tantalum cathodes in
K500 cyclotron PIG sources, over the period
March 1983 to January 198A.

Figure 3.6. Hafnium sources (X) operate at higher
pressures than tantalum sources (*), for the same

gas flow, as shown or the same group of N5+ as in
Figures 3.A-3.5.

xii

Page

57

58

59

71

75

82

9O

91

93

Figure

Figure 3.7. A comparison of the vapor pressures
of hafnium and tantalum over the range of
1500-3uoo °C [R0 63].

Figure 3.8. For N5+, hafnium cathodes are found

to Operate at low average are power when compared
to tantalum cathodes. Note the lack of
correllation between beam current and power.

Figure A.1. The activation of the K500 PIG source

is shown after a 56 h run of 30MeV/n Nu+ions. The
arrows with numbers at their tails indicate
measured radiation levels in mrem/h. The ion
source back protrusion is not shadowed by the dee
tips, and becomes more activated.

Figure A.2. The maximum energy Of particles that
can hit the central region is shown for charge
state M+, and stripping to 5+ and for three
different K values as a function of atomic mass.

The coulomb barrier for copper and tantalum is also

shown.

Figure H.3. A median plane cross-sectional view
of an activated ion source with tantalum shields
in place and removed is shown. Much of the
activity disappears when the shields are removed,
thus reducing potential exposure.

Figure u.u. The stopping range in copper and
tantalum for the maximum energy particles that
can hit the central region plotted as a function
of initial charge (q). The atomic mass for the
lightest particle that could strip to the next
higher charge and K . 500 was used.

Figure 5.1. The central region of the K500
cyclotron is illustrated for reference purposes in
subsequent discussions. The shaded areas indicate
the median plane cross-section of the dee tip
electrodes and the ion source, and gap voltage
contours are shown. The electrode on the tip of
the first dee after the ion source is called

the "puller electrode" and the boxed region is the
critical first accelerating gap.

Figure 5.2. The conversion of the source for the
production of ions from solids was accomplished
through minimal changes in the design of the
standard source. A chimney with a welded tantalum
receptacle located in the back wall behind the

xiii

Page

9A

96

99

103

106

107

113

117

Figure Page

extraction slit holds the feed material insert,
and a double tube high voltage conductor, thin in
the direction of the Chimney, feeds the upper
cathode in place of a larger diameter coaxial
conductor.

Figure 5.3. In this photograph are shown both new 123
and used LiF pellets. The sputtering of the used

pellet is evident, and the curvature of this

erosion pattern agrees well with the erosion

pattern suggested by ion trajectory computations,

as will be shown in Section 5.A.

Figure 5.”. The charge state distribution for 123
nitrogen ion production from discharge-heated PIG
sources is seen by considering two sources from
the literature [Cl 71, Be 73]. and the MSU 50 MeV
cyclotron PIG source also in the cold cathode
(high voltage) pure Penning discharge mode

[An 83]. As can be seen for the three sources
Operated in the arc discharge mode, ions Of
charges 2 and 3 will appear in the output with
approximately the same intensity, with 1 and u
within an order of magitude less than that.

Figure 5.5. Orbit trajectories for N1+ions 126

staring after the peak voltage on the first dee
are still in the gap when the voltage changes sign
and so are accelerated back to the source. A
sinusoidal time dependence is assumed for the
voltage on the first dee, so that the peak voltage

is occuring when wr-270? Though these N1+ ions

were not the resonant species, the behavior that
they exhibit here can be used to sputter LiF feed
material located in the chimney back receptable,
putting lithium ions in the are for the production

7 2+

of the cyclotron resonant ion Li .

Figure 5.6. The trajectories of late starting 127

N2+ ions, in (A), and N3+ions in (B), show that
these too come back into the source, with some

striking the feed material insert. Nu+, not

shown, however is an analog for the resonant ion

7L12+, and survives the central region of the

cyclotron.

xiv

Figure

Figure 5.7. For the ion trajectories shown in
Figures 5.A-5.5, the final energy of nitrogen ions
returning to the source is plotted versus the ion
starting time. The truncation of the curves at the
earlier starting times was made on the basis Of
whether the ion came back to the source. At 360°
the dee voltage reverses sign and positive ions
are no longer pulled from the source.

Figure 6.1. A comparison of a conventional
cyclotron (ORIC), and superconducting (K500) axial
magnetic fields is made. For ORIC 80- 17.39 kG

and for the K500 B0- ”3.1 kG.

Figure 6.2. Magnetic field lines, obtained via a
relaxation calculation and shown for the region
above the media plane of the K500 cyclotron, for
currents of 662.5 A/550 A in the small and large
coils respectively.

Figure 6.3. The central region elements of the
variable arc length ion source are shown. See
text section 6.1 for details.

Figure 6.N. The dependence of ion source
extracted current on an increase in cathode

separation is shown for Nu+ and N5+ions in the
same magnetic field. The initial cathode

separation was 3.5 inches.

Figure 6.5. A comparison of Nu+production is made
at two different cyclotron energies, corresponding
to different magnetic fields and dee voltages (see
Table 6.1).

Figure 6.6. Raising the cathodes axially in a
curving magnetic field causes a shift in the arc
alignment relative to a fixed anode, as in the
variable arc length ion source.

Figure 7.1. The extracted currents of carbon,
nitrogen, oxygen and neon ions from heavy-ion
cyclotrons at Texas A&M University [Sa 79],
Lawrence Berkeley Laboratory [Cl 81], Oak Ridge
National Laboratory [Hu 78] are compared to the
K500 cyclotron at MSU [An 8“]. Internal discharge
heated PIG sources supplied the ions in each case.

XV

Page

128

133

1311

138

1AM

1U8

151

CHAPTER 1

INTRODUCTION

The K500 cyclotron is the first of a new generation of
high-field superconducting cyclotrons that can accelerate
heavy-ions to energies well above the coulomb barrier. For
first harmonic acceleration, where the ion orbital frequency
equals the dee voltage frequency, and the three dees are
phased 120° apart, the cyclotron operating diagram is as
shown in Figure 1.1. Light ions in general and heavy ions
with Q/A approaching 0.5 have a final energy limited by the
focusing strength Of the magnet, while heavy ions with low
Q/A have a maximum energy limited by the bending strength of
the magnet. Quite a large range of Q/A and energy values
for accelerated ions is evidently feasible. Moreover, this
does not include RF Operation on higher harmonics which will

be available in the future.

The ion source, residing at the center of the

2
cyclotron, must be able to produce ions over a broad range
of masses and charge states to fully exploit the
capabilities of the K500 cyclotron. Furthermore, the charge

state dependence Of the final energy is

sf - KQZ/A (1.1)

where K is the cyclotron K-factor, which for this cyclotron
has a maximum of K-500, and where it is assumed that the
cyclotron focusing limit is not exceeded. It is Obvious
that the final beam energy for a species with mass A can be

raised, often significantly, by increasing the charge state

Q. since the final energy_goes as Q?

 

 

When work on this dissertation began, the K500
cyclotron was nearing completion, and we were about to start
building the first ion source. At that time the Penning
Ion Gauge (PIG) ion source was considered the best type Of
source for heavy-ion cyclotrons. It uses the existing axial

magnetic field of the cyclotron to guide hot electrons

(~ 102eV) coming from an arc discharge through a chamber
called the arc chamber (or chimney) that contains a low

pressure gas (. 10-2— 10-3Torr). Collisions of the

electrons with gas atoms in this chamber result in

successive ionizations to higher and higher 0 values.

4s

46

44

42
40
30 (116)

30

36
34
32

2
F

9

so

Figure 1.1

K500 cyclotron is shown.

MSU-84-458

T/A (MeV/ 11)

4

 

 

 

 

 

 

T 1 F I I fi
40 50 60 70 80
.'5' ,,ammntun
. ,7.
O o
a"
FOCUSING
l’imn
b I. ‘
FREQUENCY
" LIMIT
_ /
F d
‘11.? , . ‘ ‘
° 10 L 101‘1110'1 111
9 10 II 12 13 I4 15 16 17 18 20 22 24 26 27.5
V (MHz)

. The first harmonic Operating diagram of the

The central magnetic field is

plotted as a function of the cyclotron resonance frequency,

with constant dee voltage and Q/A lines indicated.

The

final beam energy in MeV/n is also given.

A
By putting an aperture in this are chamber the rf voltage on
one of the dees can be used to extract positive ions for
acceleration. Such an ion source was designed for the K500

cyclotron.

This dissertation is concerned with the development Of
such ion sources in the K500 cyclotron for the production Of
heavy ions. Chapters 2-6 present the properties of these
PIG sources and the studies that were made to understand
them. Since these chapters-are generally concerned with
different properties Of these ion sources, they have been
written to stand independently and new concepts are
introduced where appropriate. The remainder Of this
introductory chapter describes the properties of PIG
sources, and then discusses the actual design of the K500

sources.

1.1 Fundamental Properties of the Penning Ion Source

1.1.1 The Arc Discharge Mode

The basic PIG ion source geometry is shown in
Figure 1.2. The steady state Operation of the source will
be described first, and later we will consider how this

Operating mode was achieved.

5

The source is operated in a vacuum chamber. The hot
cathodes produce thermionically emitted electrons that
travel along axial magnetic field lines through a hollow
grounded anode. The magnetic field may come from coils that
are part of the source design or, as in our case, from the
existing field of the cyclotron. A small gas flow to the
anode raises the pressure to a few millitorr and ionization
of the atoms there results in the formation Of a plasma.
Electrons that suffer no collisions are decelerated at the
Opposite cathode. Electrons that do suffer a collision are
effectively trapped within the anode and can make many
passes through the plasma before their energy is completely
dissipated. For this reason the effective path length of
the electrons through the plasma can be many times the
actual length of the anode, and this raises the probability
that an electron will make an ionizing collision with an ion
or a neutral gas atom. As a result, cold positive ions with
several electrons removed can be produced in the plasma.
Some of these ions bombard the cathodes and in designs such
as ours this is the source of cathode heating.
Unfortunately this also means that a source Operated in this
way has a finite lifetime set by the sputtering rate of the
cathodes. Although there is a confining electric field at
the ends of the anode acting on the electrons, this same
field is of course anti-confining for the positive ions. In

addition, the ions are highly collisional at the Operating

Figure 1.2. The PIG ion source are discharge mode is shown
in schematic form.

.m.F onsmfim

fizuocaom 20. u..:. >m: >o~o_xm..~ao._.
3265 2 29.5.55. _
>o .0. a ._.

Sou-O. R c moomhouau zo_._.o<m._.xm

n-
«23.... — :

 

 

:equm . maoz<
_
.

 

 

 

 

 

 

    

D

 

 

 

 

 

mmwucmédmi

8
pressure within the source so that a radial drift across the
nmgnetic field becomes possible. We will show later that
this lack Of ion confinement sets an upper limit on the

average charge state made in the plasma.

The discharge that we have been discussing is in fact
Just one type of arc discharge that can be produced in a
magnetic field at low pressure. Many other modes are
possible, but only two turn out to be important for the
production Of positive ions, and both are encountered in the
normal Operation of a PIG source. Externally the discharge
is characterized by the arc voltage and arc current (both
indicated in Figure 1.2). Note that the arc current has
contributions coming from electron emission in one direction
and ion bombardment in the other. The operating diagram for
this discharge is shown in Figure 1.3. The arc voltage is
plotted as a function of the arc current and two modes can
be distinguished. In the first, a very large arc voltage is
Observed at very low arc current. This is the pure Penning
discharge from which the ion source gets its name [Fe 37].
In this mode the cathodes are cold and their electron
emission is secondary emission from ion bombardment, and the
output current is primarily composed of singly-charged ions.
As the arc current increases the arc voltage rises
corresponding to a positive impedance (and also confirming

that the source is operating in this mode), and in addition,

 

 

 

 

 

MSU-84-456
8r
ARGON GAS ' 2°
7 - 18
..6 -%6
> A
:75 ~h4§
u] .2
0 - 12 a:
54 t;
a "' '00
> O.
3 -8
v s
31: 2 - 6 <1:
«4
l
-2
C) ' 1 1 1 1 1 1, 1 1 1 C)
O 2 41 6 £3 10

ARC CURRENT(AMPS)

Figure 1.3. The operating diagram for the PIG ion source is
shown for Operation on argon gas, from reference [Be 72].

10
the cathode temperature slowly rises. At some point
thermionic emission of electrons from the cathodes becomes
greater than the secondary emission. The ion production
accelerates and the flux of ions striking the cathodes then
also rises, driving the source into a second mode. In this
second mode, the cathodes run hot and therefore we call this
the hot cathode mode, the discharge heated cathode mode, or
the arc discharge mode. The pure Penning discharge can also
be driven into an arc discharge by increasing the gas flow
of a source already at high voltage; this is referred to as
"striking the arc", and is generally the main way in which
the arc discharge is initiated. The fact that the source
must be able to support both a high voltage initially and
later a high current, complicates the design. A comparison
Of the range of Operating parameters for these two
discharges is made in Table 1.1. One final Observation is
in order. The charge state distribution shifts to a higher
average charge state in the arc discharge mode, where the
large electrOn current going into the plasma significantly
raises the ionization rate over the cold cathode mode, and
therefore makes the arc discharge mode the cums that is
utilized for the production of multiply-charged ions for

acceleration in the cyclotron.

11

Table 1.1.. A comparison Of the two useful operating modes
of the Penning Ion Source.

 

 

PROPERTY PENNING DISCHARGE ARC DISCHARGE
CATHODE TEMP. << 2000°K 2000°K
ARC VOLTAGE > 1 kV S 1 kV

ARC CURRENT S 0.1 A > 1 A

GAS FLOW < 1 SCCM 2-A SCCM
SOURCE LIFETIME > 20 hours afew hours
CATHODE SPUTTERING RATE low high
ELECTRON EMISSION secondary thermionic
MECHANISM

 

1.1.2 Production of Multiply-charged Ions

 

The production of multiply-charged ions in a PIG source
is a complex balance of ion formation and ion loss
processes. Since the primary mechanism is electron impact
ionization, the rate of ion formation will depend on the
electron density (supplied by the arc discharge) and the
magnitude of the cross-sections for electron impact
ionization (there are several ionization paths). At the
same time the ions in an arc discharge are subject to
extensive losses -- principally due either to recombination
via charge exchange with neutrals or leakage to the walls of

the arc chamber via Bohm diffusion.

12

It is unfortunately true that after more than 20 years
«of Operation Of this kind of positive ion source, many key
features of this balance are still known only in a general
way. This is in part due to a strong cOupling of Operating
parameters. An obvious example is the ion source are
voltage, of importance since it establishes the initial
electron energy. The are voltage is only indirectly
adjustable through changes in the arc current and/or the gas
flow rate, but changes in these parameters will alter the
characteristics of the arc. The lack of knowledge of
fundamental processes can also be ascribed to a lack of good
measurements of fundamental parameters. The list of
unknowns is long and includes the electron velocity
distribution, the ion confinement times, the neutral
pressure in the arc, and the cross-sections for the
ionization to very high charge states. With this caveat, we

positive ions in PIG sources.

The general consensus is that in the arc discharge mode
of operation, positive ions are produced through successive
collisions with electrons, such that the ionization process
proceeds stepwise to higher and higher charge states from
lower charge states [Fu 72]. Here we review some of the

basic data supporting that consensus.

1800

1600

1?: 3
o o
o o

S
C)

IONIZATION POTENTIAL (W)
8
C)

TOTAL
0)
8

§

200

Figure 1.A.

 

13

 

MSU-84-455

_ C
P 0 “Ne
- l. 4CUAF I.
r C
,.
b o
p // [I
r . /
. / .
t y. //
- /.
- C

/ I
p " l/
t a/ /"

C I
"41‘"
2 4 6 8 10 12
Q

The total ionization potentials

from neon and

argon are plotted as a function Of the number of electrons

(Q) removed for the atom,

according to reference [Ca 70].

l”

MSU-84-460
5000

4000

 

 

IJ+ 3000

2000
(epA)

1000

 

 

 

1 (fLSeC)

Figure 1.5. The rise time Of argon 1+ through 4+ ions as a
function Of time is shown for a pulsed argon arc discharge
from reference [Pa 72]. The time t-O corresponds to the arc
current coming on in the source.

15

3+

In general, we are interested in obtaining an ion A
from a neutral atom A with as large a j as possible. In
practice, the effective maximum usable j is determined by
the resultant intensity that can be accelerated and
extracted from the cyclotron. That is, high j ions must
appear in the source output in sufficient quantities to
produce useful beam currents. Starting from a neutral atom,

one can consider many different reaction processes to obtain

ion AJ+. Two extreme processes are summarized in Eq. (1.2).

A + e ----> A+ + 2e.
4- 2+
A + e ----> A + 2e.
A(J-1)++ e ----> A‘j|+ + 2e. (1.2a)
3+
A + e ----> A + (j+1)e (1.2b)

Eq. (1.2a) represents the successive impact model, while
Eq. (1.2b) shows in a single step a total Of j electrons
being removed. In general, the cross-sections for
ionization decline as the level of ionization increases

[W1 78].

16

There are fundamental differences in the signatures Of
these two ionization processes. First, with increasing
charge, the cross-sections for multiple electron removal
decline much more rapidly than for single electron removal
[W1 78]. Part Of this difference can be ascribed to the
increasing energy required to produce higher charges.
Figure 1.“ shows the total ionization potential for the
removal Of Q electrons from neon or argon [Ca 70]. One sees
that the energy required to remove the next electron is a
very sharp function of the number of electrons removed. For
example, the removal of the eighth electron from neon

requires N00 eV, which is about A0 1 of the total energy

4.
required to make a Ne8 ion.

It is also obvious that if single electron collisions
dominated the production of multiply-charged ions, then
high-charge-state ions would appear in the source output as
soon as ionizing electrons are present. The available data
indicate on the contrary that the charge states appear
sequentially, with lower charge states appearing before
higher charge states [Pa 72]. This is shown in Figure 1.5,
where argon 1+ to N+ ions are seen to rise sequentially in
time. Note also that the lower charge states actually
decline with time, indicating their probable conversion to
higher charge states. This is a very graphic demonstration

Of the successive impact ionization model.

17
There is however, a problem connected with successive
impact ionization being the dominant ion production process.

If the lifetime of an average ion in the plasma is less than
+
the time required to produce ions AJ , the output percentage

Of AJ+ may be tOO low to be useful. The lifetime is
dominated by charge exchange with neutral gas atoms, and it
is then a balance Of ionization and recombination that sets

the average charge state achieved in an arc discharge.

1.2 The Design of the K500 Cyclotron PIG Ion Source

 

and the Operating Environment

 

The final ion source design, the properties of which
are the subject Of this thesis, was established through a
sequence Of stages. The basic design was carried over from
the discharge-heated PIG source Operated in the MSU K50
cyclotron in the mid-1970's. Testing of that same source in
the K500 cyclotron magnet with dc ion extraction occurred in
1976 and 1977. The dc testing established that a PIG source
could indeed operate in a high magnetic field with the same
extracted currents Of multiply-charged ions as PIG sources
operated at lower field levels have in the past.

There followed a period of inactivity in which the
laboratory effort centered on completing the K500 cyclotron.

During that time modifications were made to the source

18
design to accomodate it within the K500 central region.
Fabrication Of the first ion source specifically for the
K500 cyclotron was completed in October 1981, and we went
back to the then de-commissioned K50 cyclotron magnet to
test this ion source, in parallel with the first operation
of the K500 rf voltage system. In late 1981 these tasks
were successfully completed and internal beams were

Obtained.

In 1982 this first ion source was operated exclusively,
while the mirror image source for the other center plug was
being fabricated. During this time the design underwent
several modifications to improve Operation, and these were
incorporated in all succeeding sources. The second ion
source was commissioned in January 1983, and in 1983 we
employed both for the production Of cyclotron beams. In
December 1983, four additional ion sources of this design

went into Operation.

1.2.1 K500 Cyclotron PIG Ion Source Design

 

A schematic diagram of the ion source "head" is shown
in Figure 1.6. This part of the ion source sits on the end
of a six foot long shaft that is inserted axially through an
air lock in the 7-inch diameter cyclotron center plug. The

part Of the source shown in Figure 1.6 is approximately

19

Figure 1.6. A sectional view through the anode of the head
of the K500 PIG ion source is shown. Compare this to the
schematic PIG source diagram in Figure 1.2.

20

.o._ ocsmfim

mmoqoz oz< moorbuo
Azgozm .55
ume cum“. m<o com uo<._.1_o> 10.:
meP >oom

mmomzn. ozu MQOZd.

 

 

 

 

  

 

 

 

 

 

 

....1|.|I...............1...1m..o. u Baal-It... Ill all? Vill/I/l/l/I/I/I/l’ll VIZ. .15.». ilv
2 . m
. . . sessile T 7 u. a
.5! 3.11.“ .1. 1 1. /<\\\\\> 2 §\\\.“
8.52 zomom §
23325 § ism;

uzfia 24.82
3&8 Z

 

msunmuxami

21
7 inches long. or course the axial magnetic field of the
K500 cyclotron is quite large, so we have built the source
without iron to allow insertion and removal during Operation
of the cyclotron magnet. The design of the first harmonic
central region of the cyclotron, shown in Figure 1.7. fixes
precisely the location of the ion source extraction slit and
the shape of the ion source anode [Ma 81]. The ion source
extraction slit resides 0.32 inches off the central axis of
the cyclotron, with a 1 cm gap to the puller electrode on
the end of the first dee. The ion source anode terminates
the rf electric field in the center in a precise shape
intended to boost the beam energy gain on the first turn to

clear the ion source itself and the dee tip electrodes.

The ion source design emphasizes rapid recycling after
the cathodes have worn out. Surfaces exposed to the arc
will eventually wear out and hence can be removed for
replacement. The core of the anode is a water-cooled copper
part that is not intended to be replaced during ion source
recycling. Replaceable copper end pieces define a .203 inch
diameter anode aperture and seat the chimney. The tantalum
chimney has a .250 inch square internal bore with a machined
exit slit. We make chimneys out of 0.050 inch thick
tantalum sheet stock, folded into a rectangular shape,
TIC-fusing the resulting seam along one edge. When the exit

slit has eroded away (typically after about 50 hours of

22
h=|, Q/A=O.286, E '-' 30MeV/n
BO=4.O4T, uRF = 17.73 MHz, vDEE =70kV
To: "30°

 

l.___l

1 inch

Figure 1.7. The central region geometry for first harmonic
acceleration is shown. The ion source and electrodes
crossing the median plane are shown as shaded areas. The
equipotentials shown here correspond to all three dees
having the same voltage and indicate the six acceleration
gaps (from reference [Ma 81]).

23
of cyclotron Operation) from the extraction Of ions, we can
just insert a new chimney. The 3/8 inch diameter cathodes
are pressed into tantalum holders that slide over a
water-cooled center conductor, we will refer to as the high
voltage rod, that provides the cathode voltage bias for

source Operation.

The are discharge is struck in this source when the
voltage is raised to 1.5-3.5 kV and about 5 cc/min Of gas is
bled into the chimney. We then reduce the gas flow to
1-3 cc/min for normal Operation and raise the arc current to
1.5-A.0 A” thiltiply-charged ions will then appear in the
source output. Figure 1.8 shows the inverse dependence of
the arc voltage on gas flow and arc current, a
characteristic signature of the arc discharge mode of

Operation (compare with Figure 1.3).

1.2.2. K500 Ion Source Commissioning Experience

 

The source design used for this thesis was the result
of an extended period Of development in the K500 cyclotron.
That development work will be briefly reviewed here. Several
physical properties of the source Operation are

demonstrated.

The ion source designed for the K500 cyclotron was

first tested in the de-commissioned K50 cyclotron. The

2N

 

  
  
  

 

 

 

70° 1 1 1 T 1
C02 GAS FLOWS
‘ "' o 1.0 cc/min ‘
A 2.0
600'- . 3.0 +-
Von: " ‘1
(Volts)
500- .-
4001- .-
300 1 1 1 1 1
1.0 2.0 3.0

Iorc (Amps)

Figure 1.8. The dependence Of the arc voltage on the arc
current and gas flow is shown for the K500 cyclotron PIG ion
source, for a carbon dioxide arc. The decreasing voltage
with increasing arc current or gas flow is characteristic of
this discharge mode.

25
major difficulty encountered with the source there was the
failure of the vacuum voltage feedthrough; the power and
water cooling are brought in to the vacuum through an
alumina insulator at the other end Of the ion source, so
that the inside of the ion source is under vacuum, but the
feedthrough insulator itself is 5.5 feet from the main
source of pumping-- the cyclotron beam chamber. We found
that the pressure inside the source body was high enough to
initiate a glow discharge inside the ion source near this
insulator and the power dissipated by the glow caused the
insulator to crack. This problem was eliminated by drilling
holes in the side Of the source shaft to vent it to the
center plug interior, and then to pump on that space to

improve the vacuum.

Several improvements in the way we prepared and
operated the source during 1982 resulted in a significant
improvement in source lifetime. The histogram in Figure 1.9
shows the steady improvement in source lifetime with the
total number Of source runs during that period. The data
include both source and cyclotron terminated runs, with the
intrinsic source lifetime higher than the values in this
histogram by about 25%. The first runs in the K500
cyclotron could be characterized as being dominated by
failure in the source head Of the center conductor, its
support insulators, and parts of the anode that absorb are

power. Generally these problems were the result of the

26

 

 

 

 

   

 

  

 

 

 

 

           

 

MSUX-83-156
8
,‘7
U)
%6
S”:
12.5 ——
u:
24
1.:
1335-6
-: 1 I III
2" llllllll 7
'lllllllllllllllllll”
IO 20 30 4O 50 60 7O 80 90100110
NUMBER OF SOURCES
Figure 1.9. The increase in the average ion source lifetime

with the number of runs, during the early development of the
source is shown.

27
conflict between the requirement that the ion source have a
small cross-section for the clearance Of ions on the tight
first orbit in the cyclotron, and the requirement that
voltage, current and water cooling be passed through the
small median plane cross-section to feed the cathode on the
other side--which is necessary in an axial ion source
design. This is Of course a problem characteristic of both
the highly optimized central region design and the large
magnetic field at the center, which makes the first orbit
diameter small. Many small modifications to the high
voltage rod and insulators pushed the lifetime tO about four
hours average after 30 sources. The source design in

Figure 1.6 includes those modifications.

In the next phase we were running mainly carbon beams
in the cyclotron, and the source lifetime was a poor

2.5-3 in. Source Operation with CO or CO degrades the

2
resistance of the boron nitride insulators, through the
build-up of a conducting carbon layer on the surface, and
after a short time the source shorts out. After about 55
sources we started tracking the center conductor
resistance-to-ground during cathode changes and replacing
the insulators when the voltage holding threshold went below
1000 V. The source lifetime improved, and then more typical
failures resulting from deep craters in the cathodes or are

shorting from cathode material build-up in the apertures,

28

the normal failure modes for PIG sources, began to determine

the average lifetime.

One final change was made in the Operation to boost the
lifetime above 6 h average. At about 80 sources we started

supporting the arc on N2 during carbon runs, and bleeding in

only enough CO to produce adequate beam intensity. This

reduced the rate Of carbon deposition on the insulators.

The dip in the histogram at 90 sources was due to
start-up work after a 1.5 month shut-down, and that at 100
sources was due to the commissioning of the second ion
source. After that we could get 8-10 h lifetime for most
light ions up to nitrogen, and charge states up to fl+. This
was the status of the ion source development when the work

that is discussed in the dissertation was begun.

29

1.3 Publications in Connection with this Research

The following publications have appeared in the

literature in connection with this research:

A. K500 Cyclotron Central Region Design

1) E. Liukkonen, J. Bishop, S. Motzny and T. Antaya,
Design of the Central Regions for the MSU 500 MeV
Superconducting Cyclotron, IEEE Trans. Nucl. Sci., £2226
(1979) 2107.

2) F. Marti, M.M. Gordon, M.B. Chen, C. Salgado, T. Antaya
and E. Liukkonen,

Design Calculations for the Central Region (H' the NSCL
500 MeV Superconducting Cyclotron, Ninth International
Conference on Cyclotrons and Their Applications, Caen,
France (1981).

B. Cyclotron Central Region Activation

1) M.L. Mallory, T. Antaya, F. Marti, and P. Miller,
Radioactivation of a Heavy-Ion Cyclotron Central Region,
Nucl. Instr. and Meth, 222 (198A) A31.

C. K500 Ion Source Development

1) T.A. Antaya, M.L. Mallory, and J.A. Kuchar,
Operating Experience with an Ion Source in a Superconducting
Cyclotron, IEEE Trans. Nucl. Sci., N§13Q (1983) 2716.

2) T.A. Antaya, J.A. Kuchar, M.L. Mallory, P.S. Miller and
J. Riedel,

The Development of Heavy Ion PIG Sources for the NSCL
K-SOO Superconducting Cyclotron, Tenth International
Conference on Cyclotrons and Their Applications, East
Lansing, Michigan (198A).

3) T.A. Antaya and P.S. Miller,

Extended Lifetime and Other Beneficial Properties of Hafnium
Cathodes in Heavy-Ion PIG Sources, Tenth International
Conference on Cyclotrons and Their Applications, East
Lansing, Michigan (198A).

CHAPTER 2

PULSED ION SOURCE OPERATION; NEON AND ARGON BEAM PRODUCTION

During the initial operation of the K500 cyclotron,

extracted beams of more than a few nanoamperes of Ne5+ and

Ar6+ could not be achieved. A survey of other laboratories
indicated that in many instances beams Of these ions were
produced via a pulsed arc discharge. The general view was
that peak arc currents much higher than could be maintained
dc due to power dissipation limitations were possible, and
this led to an increase in the production of
multiply-charged ions. Modification of the K500 ion source
power supply for this purpose was possible, and in a short
time we were able to make the source run pulsed. The
develOpment of ion beams via pulsed Operation and a study of
the properties of the pulsed PIG arc discharge are the
subjects of this chapter. After modifying the power supply,
many beams were attempted with the ion source running

pulsed, with the most important application ultimately being

30

31

5

N *wlth hafnium cathodes, the discussion of which will be

5 6+

deferred to the next chapter of this thesis. Ne +and Ar
beams on the other hand have proved to be the most difficult

cases, and will be discussed herein.

This chapter will start with a review Of the use of
pulsed PIG discharges at other laboratories for the
production of multiply-charged ions. £Since the
implementation of pulsing from an engineering standpoint is
a question of changing the operation of the ion source power
supply and not the ion source itself, the power supply
modifications, made by the laboratory electronics group,
will be discussed. Next, the more important pulsed PIG
source Operating characteristics that became evident during
the development of K500 cyclotron beams will be introduced

via selective examples and the performance of the pulsed

source for Ne5+ and Ar6*will be compared to do operation.
An empirical model for the mechanism by which the higher ion
currents of multiply-charged ions are obtained will be
introduced and discussed. Finally, attempts to alter the
low source lifetime that seems to come with pulsed

discharges will be reviewed.

32

2.1 Review of the Process and Anticipated Gains

The most extensive study of pulsed PIG discharge
characteristics was performed in the 1960's by Pigarov and
Morozov [Pi 61a, Pi 61b]. Ions of N, 0, Ne, Ar, He and H
were studied in a test stand with do ion extraction and a
magnet to analyze the charge-to-mass ratio. In this work it
was shown that ion current does increase with the arc
current in a pulsed discharge. The highest charge ions of a
particular species generally required more arc current, and
the lowest charge ions actually declined with increasing arc
current. Since the arc went out between pulses, the time
dependence<flithe ion formation could be observed, and it
was found that there was a regular sequence of evolution in
which the lower charge ions always appeared before the
highest charge ions. This latter result served for many
years as the primary physical evidence for successive impact
ionization collisions with electrons being the main
mechanism for the production of multiply~charged ions in PIG

SOUPCGS .

Pulsed arc discharges were used extensively in Europe
in the 1970's for the production of multiply-charged ions.
This generally followed tnna development of the
indirectly-heated cathode PIG sources at Karchatov [Ma 76],
and the subsequent application of that source design at many

other laboratories including Bonn, Ganil and GSI-Darmstadt

33
[Ag 79, Sc 76, Ba 79]. However, in this type of PIG source
one of the cathodes is heated by an auxiliary electron
emitting filament and has some different Operating
principles; for example, the auxiliary electrode is the main
source of cathode heating and not the discharge, so that
there is no requirement that sufficient gas flow be
maintained to stabilize the arc. Therefore detailed
consideration of the pulsed operation Of these sources was

not made.

Pulsed discharge-heated PIG sources were developed at
two American heavy-ion laboratories, LBL and Texas A&M,
during the 1970's. The K500 ion source is of this type,
wherein the cathode heating is coming from the discharge
itself, and so these PIG sources were considered in some
detail. At LBL, the ion source is generally run do, but
when required, pulsing at 20-20,000 Hz and a 20-50% duty

factor is used to increase intensity [GO 79]. Ne5+ and

Ar6+ are obtained from the 88-inch cyclotron there with
extracted currents of “-5 euA [Cl 81]. At Texas A&M,
pulsing is used for the highest charge state of each species
accelerated [Sa 79]. A pulse width of 0.3 to 7 ms, and a
maximum arc current of 18 A are available. It was shown
that the extracted current Of multiply-charged ions
increased with pulsed Operation at arc currents of greater

than 10 A. The intensity seemed to have an Optimum for a

3A
pulse on-time of approximately 1 ms and an Off-time of 2 ms,
giving a duty factor in the range of 33%. The currents of
low charge ions were higher with a dc discharge. One
important observation was that the gas flow required to
stabilize the discharge was much lower for pulsed Operation:

0.2-0.5 sccm for Ne or Ar feed over approximately 2 sccm do.

In all Of these cases, it appeared that going to pulsed
operation allowed a much higher arc current in the ion
source, and that this is important for the production of
multiply-charged ions, even though this higher current can
only be maintained for a shorter time interval. Coupled
with this higher are current is a reduced requirement for
support gas to stabilize the arc, apparently because the
high arc current is taking over that function. A lower gas
flow would reduce the recombination rate inside the source
and the transmission through the cyclotron for an additional

improvement in beam current.

At another major heavy-ion laboratory in the United
States, ORNL, multiply-charged ions were being produced with
intensities comparable to LBL and Texas A&M, but without the
use of pulsing [Hu 75]. Instead, there it was found that
the cyclotron vacuum had indeed a significant impact on the
beam attenuation during acceleration due to charge exchange
with residual gas atoms, and that large increases in beam

intensity could be obtained by improving the cyclotron beam

35
chamber pumping . Since pulsing evidently allows arc
stability at reduced gas flow rates, the gas load that the
cyclotron sees from the ion source decreases, and so in that

sense these two different experiences are in agreement.

The plan for the implementation of pulsing at MSU
included the modification Of the power supply are current
regulation, and boosting the peak supply output current,with

initial testing for the production of Ne5+and Ar6+beams at

25 MeV/n and 10 Mev/n, respectively. The dependence of beam
current on arc current, duty factor and gas flow were to be

studied.

2.2 Power Supply Operation for PIG Souppes and Modifications

 

 

to Implement Pulsing

 

A schematic diagram of the ion source power supply
circuit is shown in Figure 2.1. The negative impedance of
the mode two arc we employ for ion production must be
stabilized by the inclusion of a positive ballast resistance
in series with the ion source and the dc power supply. The
Eimac NOOCW1OOD power tetrode can be viewed as such a
ballast, with a variable resistance ranging from a few

hundred ohms to infinity at cut-off.

36

The operation of this circuit will now be described.
The current flowing through the ion source to ground must
equal the current collected on the plate of the tetrode.
This current is itself prOportional tO the grid bias
voltage, since the tube is operated at a fixed screen
voltage of +1 kV. At this screen voltage, a grid bias of
-300 V should in principle cut-Off the plate current
completely, and then rasing the grid voltage from that value
allows current emitted from the heated filament to reach the
plate. Hence grid bias adjustment is the mechanism for
controlling the current flow in the ion source circuit. One
further point-- since there is a resistor between the tube
filament and ground, when the tube draws current the
filament potential rises from ground, reducing the net grid
bias slightly. This bit of negative feedback helps

stabilize the tube current.

With the above characteristics in mind, it is now
possible‘matndefly discussion how the arc discharge with
heated cathodes is obtained. Since the tube current is not
completely out off, at a grid voltage of -300 V, nearly the

entire power supply voltage (Vs) appears across the ion

source. Im.it would not be desirable for this voltage to
appear instantaneously, as a high voltage conditioning
period is often required initially for the source to hold

high voltage. Therefore a resistor (R) is added in

37

Figure 2.1. In this schematic diagram is shown the ion
source power supply circuit. On the right there is an Eimac
AOOCW1OOD tetrode and its filament, grid and screen
supplies, for the purpose of regulating the current for the
high voltage supply VS that powers the ion source.

38

 

 

covnvmunms

.P.m sesmss

 

 

 

h
em. on. m on
also. «ass
68:22 I 8%... 11 b.
Susi: >o...c
2.:
.2—1'. 111“ v
96 G824 xm>8m1
zwwmom wh<4d
Tl:-
msm emu

momDOm 20.

T.

 

.._T)m?1hMMUi

r

 

> _ .o u < _
.8888...
2850

8" 0.5

.835 .
23> :oood.

95>

39
parallel with the source to limit the turn-on voltage to
approximately 750 volts. Subsequently raising the grid bias
allows current flow, and the arc voltage rises from this
initial value. The addition of some gas to the source at
this time results in some ionization by field emitted
electrons. The cathodes are hit by positive ions and the
cathode temperatures rise. More electrons are produced via
secondary electron emission as a result of this ion
bombardment, feeding the process, and the ion population,
cathode temperature and electron emission current density
all increase until thermionic emission from the cathodes
predominates and we say that a mode 2 are has struck. The
tetrode provides arc current regulation and stabilization

once the mode 2 arc discharge has been reached.

In order to pulse the arc current, the grid bias
voltage must be pulsed. This is accomplished using a
circuit developed by J. Riedel [Ri 8A]. A negative square
wave signal of amplitude 0-10 V, variable frequency and duty
factor is supplied to the base of a bipolar junction
transitor, allowing current to flow through the transitor
and causing the grid voltage to rise above the -3OO volt
value with the same frequency response and duty factor. The

result is a pulsed output current from the tube.

For diagnostic purposes, both the instantaneous arc

voltage and arc current are made available. The arc voltage

110

is obtained via a frequency-compensated 10,000:1 voltage
divider circuit. The are current is taken Off a transductor
having a dc to 1MHz frequency response through which the
cable going to the ion source passes. A transductor output
of 1 V equals 10 A. These signals are best displayed on an,
oscillOscope. They are also connected to a multiplier whose
output is displayed on a meter reading average power input
to the ion source.

5 6+

2.3 Pulsed Operating Characterippics for Ne +and Ar

The most obvious result of pulsing the arc is an
increase in the number Of adjustable parameters, as shown in
Figure 2.2. Apart from the gas flow, one has only the arc
current to adjust with a dc arc, with the arc voltage
following changes in either. The two new parameters are the
duty factor and period, or, alternatively, the on-time and
off-time. One has also to introduce peak and average arc
current and voltage values, but of these only the peak arc
current is directly adjustable. It turns out that this
increase in parameter space allows more control over the
characteristics of the arc.

When running the source pulsed, it becomes immediately
apparent that the beam current is sensitive to the duty
factor. We are going to consider some curves showing the

22 +. .
cyclotron extracted current of NeS lons versus ion source

A1

MSU-84-389

PULSED ARC DISCHARGE
ARC RANGE

VOLTAGE VP VP: 0 - 5 kV

ARC Mum-um

p
CURRENT
1‘70 Torr‘ T=2.5-4msec
R——r——>1 -
DUTY FACTOR Ouch/r D=O.17-1.00
AVERAGE AND PEAK VALUEs=I=IP~O
i7::\/P°E>

Figure 2.2. Pulsed ion source arc voltage and arc current
waveforms.

A2
parameters. The extracted beam current is only proportional
to the ion current in the source, since the extracted
current depends on the central region acceptance Of the
cyclotron, the transmission efficiency during acceleration,
and the cyclotron extraction efficiency. As a result these
curves have only a relative value. That is, only if they
are taken under the same conditions at the same time can one
make relative comparisons. This is one of the main
limitations of studying ion sources in a cyclotron.
(However one does get an unambiguous beam current as a
result of the resonant acceleration generally, and also
ultimately the ion source must perform well in the
cyclotron, since getting it to work there eliminates
intermediate steps.) Figures 2.3, 2.A and 2.6 display

datataken during one early cyclotron run for develOping a

22 5+
Ne beam using a pulsed ion source. The A=22 isotope of
neon was obtained via 70% 22Ne enriched neon gas.
Ne5+is a representative multiply-charged hnn and we

find that the beam current is sensitive to the duty factor,
as shown in Figure 2.3. The extracted current is plotted
versus the pulse on-time for a fixed period of 3.0 msec.
With the period held constant, varying the on-time changes
the duty factor. With an arc current of 6.0A peak, the beam
maximizes at an on-time Of 1.0 msec, for a duty factor of

33%. However at an arc current of 9.5A peak, the beam

’43
current is seen to rise sharply with increasing on-time up
to 1.0 msec, where the increase was stopped to protect the
source from the average power which had risen above 1.5 kW.
The magnitude of the beam current at 9.5A arc current is
about twice that for 6.0A for the same on-time. These
characteristics generally agree with the previously reported
experience at other laboratories. In particular, the sharp
sensitivity of the beam current to the arc current at low
duty factors has been noted both at Texas A&M and the
studies of Pigorov and Morozov prevously mentioned.
Figure 2.A also shows the interrelation between the beam

current and the duty factor and arc current. Again 22Ne5+

ion production is considered, this time plotted versus arc
current for fixed duty factors of 17 and 33%. The beam
current is seen to increase up to the power supply limit for
the 17% duty factor case. In fact, the curve implies that
had more are current been available, more beam current would
have been possible since the beam current is still rising at

the power supply limit.

NA

 

   
 

 

 

 

fig T’ T r T
a) Ne5+
200 ... 9.5 Runs q
I.—
E
0:160 -
or I"
D
0120
I: 1— q
El 6.0 Flaps
‘3 80 _ _
Q
E
0 no .. 1
g [3.0 msec period]
:2 13 l, 1 1 1
Lu .0 1.0 2.0

ON-TIME (msec)

Figure 2.3. The dependence of 22Ne5+beam current on the

pulse on-time is shown for a fixed pulse length of 3.0 msec
and arc currents of 6.0 and 9.5 A.

“5

 

 

 

 

 

E“:

5

VII-10 ..

5122- _

E

95100 1-

O

z 80 _.

E c

on 6 P“

8 L10 _

:3 .

E 29 - [DUTY FACTOR] ‘

i— 0 1 1 1 1 1

X

L“ 4.0 6.0 8.0 10.0

PEQK RRC CURRENT (Qmperes)

Figure 2.“. The dependence of the 22Ne5+ beam current on

the peak arc current is shown, for fixed duty factors of 17

and 33%.

“6

The sharp rise in beam current at a low duty factor
like 17% is consistent with an additional observation-- a
pulsed arc has a surprising number of unstable modes, euui
therefore is quite sensitive to changes in operating
parameters at low duty factors. Oscillograph (B) in
Figure 2.5 shows normal pulsed arc voltage and arc current
traces at a 20% duty factor and an arc current Of 7 A. The
arc voltage shows a leading high voltage spike peak of
1.0 kV and a voltage spread of 100 V. (It should be
mentioned that this spike increases as the duty factor
decreases, eventually reaching 5 kV at something like 10%,
indicating that the arc is essentailly restriking each
cycle!) IT’we cut the gas flow to less that 1.0 sccm, we
often see an instability like that in oscillograph (A) in
Figure 2.5. It has the disconcerting characteristic that it
grows in time and will result in the arc discharge dropping
out unless Hzis corrected. After about one-fifth of the
on-time, the arc current collapses into oscillations which
the tetrode can no longer stabilize, and the arc voltage
goes back to 5 kV. The ion output for the source is
unstable when these oscillations are occurring. Fortunately
it can be corrected by increasing the gas flow and/or the
duty factor. Both of these will result in an increase in
the cathode temperature, via increased ion bombardment, and

that seems to be what is necessary to stabilize the arc.

“7

Figure 2.5. The discharge heated PIG source Operating can
be separated into regions of stability and instability
during pulsed operation. Oscillograph (B) shows normal arc
current and voltage waveforms. Oscillograph (A) results
from a large decrease in the gas flow; This condition is
unstable- eventually the arc drops back to the cold cathode
mode and the production of multiply-charged ions will cease.
Oscillograph (C) results from a large increase in the arc
current. Evidently this large arc current is not
neutralized by positive ions from the plasma, and twice
during each pulse the plasma column is seen to collapse.

“8

 

.m.m ccsmsa

.m.

 

 

\a /

2.202%:
m>_m oxm //
\\\ Sag“ "

Amwoozhho 5.0+:

mgowm mcmdroma om<

mmm -vmnnms.

 

moan—30>
omd

 

SBOOHLVO 0100

“9

Another unstable mode can be Obtained with excessive
arc current, and is shown in oscillograph (C) in Figure 2.5.
Under normal Operating conditions, the electron emission is
space charge compensated by the positive ions in the plasma.
If'the electron emission is excessive, there is a lack Of
compensation and the arc discharge collapses into the
Penning discharge with a subsequent decrease in the beam
current also. This is eliminated by simply reducing the arc
current or increasing the gas flow. These two particular
instabilities are the most commonly observed, and correspond
to definite regions of the PIC source operating diagram, as
indicated in the upper portion of Figure 2.5. The region of
low gas flow borders on the transition between the pure
Penning discharge with cold cathodes and the arc discharge,
whereas the high arc current region is on the far edge Of
the arc discharge region, often exhibiting again some
positive arc impedance. The average ion current produced by
the source maximizes somewhere between these two extremes,

and that is where we try to run the ion source.

An important feature of pulsed Operation is that the
high instantaneous arc current stabilizes the discharge at
much lower gas flows. For reference, a gas flow of less
than approximately 2 sccm, for a dc discharge, is not stable
and would result in the shift to the cold cathode Penning
discharge. This is important because, as shown in

Figure 2.6, gas flow has a significant effect on beam

 

 

 

 

T T T T
5+
No 1:] NITROGEN SUPPORT.
NO 0 0.25 SCCM
100 _ A
0 NEON ONLY
E-
3 86 .. -1
r—
2:
MA
1:
g 60 .. .
(J
2:
C!
UJ
a:
. “0 E -
P
><
UJ
20 _ ~
0 1 1 1 11
.8 1.6 2.4

TOTAL 098 FLON (soon)

Figure 2.6. The gas flow rate into the ion source is a
sensitive parameter for the production of multiply-charged

ions. In this case the 22Ne5+ extracted beam current is

shown for neon gas (70% enriched 22-neon) source feed, and
also for an arc supported by nitrogen witilaa constant neon
flow Of 0.25 sccm.

51
current. We return to the case Of the production of

22 5+
Ne ions as in Figures 2.3 and 2.“. Two cases are

illustrated. In the lower curve the source is being
Operated on neon only, and the beam current is seen to fall
70% from a gas flow of 1 sccm to 2 sccm. In the upper
curve, the arc is now primarily operated on nitrogen gas,
with neon bled in as an auxillary gas to produce the ion
beam. This curve also shows a sharp increase in beam
current at low gas flows. The lower overall level Of beam
current in the case of pure neon flow is due to poorer beam
chamber vacuum. Nitrogen is more easily pumped by the
vacuum system, and an increase in beam current due to a
reduced attenuation at the center of the cyclotron results
from replacing some of the neon with nitrogen. Since it
happens that nearly all of the neon can be so replaced, the
charge exchange in the cyclotron has clearly a large effect

on the final extracted Ne5+ current!

As previously mentioned, the N5+ beam is probably the

most important application of pulsed source Operation at

MSU, but Ne5+ and Ar6+have proved more difficult from the
standpoint of source Operation and useful beam currwnlts for
nuclear physics experiments. Table 2.1 shows the results of
Operating ion sources for the production of 25 Mev/n

22Ne5+ and 10 MeV/n quré+ in the K500 cyclotron. In both

cases there were approximately 50% more trials than

52
indicated in Table 2.1, particularly for dc operation, but
many were early in the first operating period of the
cyclotron, however, and have been obscured by non-source
difficulties. In one very tight period both beams were
Operated dc and pulsed when the transition to pulsed
Operation was made, and it is that data summarized in

Table 2.1. As before, the feed gas for the 22Ne5+ beam was

70% enriched 22Ne, with the remainder 20Ne. For both beams

in Table 2.1, the beam current obtained dc averaged less
than 10 enA. It can be seen in Table 2.1 that pulsed
operation at 15-33% duty factors does result in a large
increase in beam.current. In that respect the
implementation of pulsing can be regarded as a success,
since the extracted beam current is raised to a more useful
level. At this time there are two other features of
Table 2.1 that are important:(1) the lifetimes are quite low
for these beams, both pulsed and dc, and (2) the average arc
power during pulsed Operation is less that half the dc arc
power, and yet the beam currents are much higher! These
features of pulsed Operation will be addressed in the next

two sections of this chapter.

53

Table 2.1. The characteristics of 22Ne5+ “0

and

6

4.
Ar beams

extracted from the K500 cyclotron with dc and pulsed arc

discharges in the ion source.

given in parentheses.

The standard deviation is

 

 

Species Cath Duty Aver. Lifetime I(BSO)

Type Factor Power (h) (enA)
(kW)

22Ne5+ Ta 100 1.0(.11) 3.8(1.0) 5.2(2.3>
Ta 30 .“5(.09) 1.1(.20) 26.(19.7)
Hf 15 .20(.06) 3.3(1.1) 77.(6.“)

“01r6* Ta 100 1.0(.05) 1.8(.0“) 8.3(5.6)
Hf 33 .39(.02) 2.02(0.3) 59.(19.2)
Ta 33 * 1.8(.11) *

 

 

ETT‘ _______
Runs for lifetime only-

beam not extracted.

5“
§;i_122_522922322_§2£_52.1.£12l112223829_£22-3229222122_lfl

Pulsed Arc Discharges

 

A chief difficulty with the dc discharge is that, owing
to the negative impedance Of theiarc, the arc voltage sags
as the arc current increases. Both are important--the arc
current is important in setting the rate of ixnrization and
tune arc voltage sets the maximum ionizing electron energy.
Eventually one reaches a condition where the cathodes are
emitting a lot of electrons (high are current), but the
multiply-charged ion production falls anyway because Of a
low electron energy (low arc voltage). Multiply-charged
ions require a higher electron energy to remove the more
tightly bound inner electrons, and the sagging arc voltage
sets one important limit on the highest practically
attainable ionization state by setting the electron energy
limit. In this section it will be shown that pulsing
introduces a mechanism whereby the usual negative impedance
of the arc discharge can be "effectively" positive, and that
this occurs because the arc voltage increases as the duty
factor is reduced.

The relationship between the arc current and are
voltage can be altered in a do are by reducing the gas flow.
In fact, tflue arc voltage rises as the gas flow is reduced.
With lower gas flow the flux of ions to the cathodes
decreases euui the cathodes must cool. The are voltage must

then rise to maintain the same arc current, which is being

55
held constant by the power supply. This has the further
advantage Of reducing the ion recombination rate by reducing
the gas pressure inside the source and also the cyclotron
beam chamber. Of course, lowering the gas flow eventually'

puts the arc discharge out.

It is easy to write a set of ideal requirements for the
production of multiply-charged ions in a PIG source. We
.would like to raise the arc current, keeping the arc voltage
up at the same time, but also in such a way that allows a
low gas flow and also keeps the average arc power low to
protect the ion source against melting. figlglpg_£flg_§§g
current on a millisecond time scale as we have now done

'ggggmpli§p§§_£pi§! At a fixed arc current, decreasing the
duty factor of the pulse lowers the average arc power and
the cathodes cool, with a subsequent rise in the arc voltage
to maintain the constant arc current. Reviewing the pulsed

cases in Table 2.1 confirms that the average power

dissippated in the source has significantly decreased. For

example, the entries fom Ar6+show that in going from dc to
33% duty factor, the average power drops by a factor of 2.5,
but at the same time the beam current has increased a factor
of 8.5. This has come about in part from the need to have a
stable arc. If only the duty factor is lowered, the arc
discharge would eventually drop out because of cooling

cathodes. Therefore, the arc current must at the same time

56
be raised to increase the total ion production rate to
stabilize the arc. In this way it turns out that both the
arc current and arc voltage during the on-time have been
raised far above do levels, so that the time-averaged ion
production rate is well above the dc level in spite of the

low duty factor. This is why pulsing works.

Recently Nu+ beam current in the K500 cyclotron was
observed as a function Of are power, are current and are
voltage, at fixed duty factors of 50, 75, and 100%, as shown
in Figures 2.7- 2.9. These curves demonstrate the main
features Of the above argument. In Figure 2.7, there is a
remarkable linear dependence of beam current on average are
power, and also the three curves can be nearly partitioned
into separate regions Of low, intermediate and high average
arc power. We can say that for a constant duty factor the
beam current increases linearly with arc power, and yet the
beam current at 200 watts and 50% duty factor is more than
twice as large as the beam current at 500 watts and 100%
duty factor. So the average are power consumed by the
source is not as important as how it is obtained. If we
look at the peak arc current during the pulse, in
Figure 2.8, we see that this again separates on the basis of
duty factor, and with the beam current increasing with are
current in all three cases of duty factor. Observe that for

approximately the same peak arc current, the beam current is

57

 

 

 

 

 

Nu; I l T I l I
“500 r. -
CE
C
3 15021
“00 _. -
p.
2
E8
(53% __ 17521 4
C.)
2
E
00200 _. 110021 3
(MOO 1. (I A
0 [DUTY FACTOR]
100 200 300 “00 S00

QVERQGE RRC PONER (NQTTS)

Figure 2.7. Nu+ extracted beam current is plotted versus
average arc power, for fixed duty factors of 50, 75 and

100%.

58

 

 

 

 

11+
@590 _ N 15021 4
0
1.2—L100 _ -1
E 17521
%BOO .. l
O
E
32% - new *
;1OO . -
LLI
[DUTY FACTOR]
0 1 1 1 L 1 1
.6 1.0 1.11 1.8

PEQK QRC CURRENT (RMPS)

Figure 2.8. 1““ extracted beam current is plotted versus
peak arc current for 50, 75 and 100 % duty factors,
corresponding to those same cases in Figure 2.7.

59

 

 

 

 

 

9+

A N
CI5“301 b

5
V 1752]
1.1-100 ..
2
1.1.!
a:
%300 L
Q
E:

200 .
1301 F
5199 — 110021

[DUTY FACTOR]
0 1 1 1 1 1 1 1 1 1 ,
2'50 350 L1'50 550 650 750
PEAK ARC VOLTAGE (VOLTS)
Figure 2.9. Nu+extracted beam current is plotted versus

peak arc voltage for duty factors of 50. 75 and 100%,
corresponding to those same cases in Figure 2.7.

60

mymr for a lower duty factor. The explanation of this

MTect comes from Figure 2.9, in which the beam current is

mbtted as a function Of the peak arc voltage. The are

thage is increasing rapidly with the decreasing duty

radar, so that for the same peak arc current the product

Ix V is much larger for a lower duty factor. 80 at low

P

MMy factors both the arc current and are voltage are

allowed to be large, and that is compensating for the

rmumtion in the on-time to such an extent that the total
beam current still increases. Also in Figure 2.9, Observe

Hun at a constant duty factor the arc exhibits the usual

negative impedance, so that the basic character of the arc

discharge has not been altered. One final point about the

curves in Figures 2.7-2.9: the beam current varies more

rapidly with all parameters at low duty factor--implying

less stability and more critical tuning.

2.5 Source Lifetime with Pulsed Neon Discharges

For the production of Ne5+ions with tantalum cathodes

arm: a 33% duty factor, Table 2.1 shows that the source

lifWetime is only about 1 h. This low lifetime exacts a heavy

scnirce maintainence cost whenever this beam is attempted,

arid it would certainly be attempted more if the lifetime was

inipr~oved. We decided to try to boost the lifetime by making

crlariges in the arc geometry and would like to review that

61

work here. As is always the case with PIG source operation,

the strong coupling Of the various parameters makes it
difficult to alter the Operation of any one independently.
This is particularly true of the source lifetime. For
example, increasing the duty factor has been observed to
reduce the sputtering rate, but often the beam current goes
down as well. Increasing the gas feed would also help by
lowering the arc voltage, and therefore the energy at which
positive ions bombard the cathodes, but the gas flow is one
of the most sensitive parameters with regard to beam
current, and we find for instance, that beam current

decreases rapidly with increased neon flow (see

Figure 2.6).

One obvious approach would be to stop the formation of
the deep crater, since this is what is making the arc go

out, and a study was made in that direction, based on a

review Of a successful approach made at Texas A&M.

According to Sakurada [Sa 79]. the lifetime Of the

Texas A&M PIG was improved by reducing the diameter of one

cathode to no more than 1 mm greater than the diameter Of

the plasma chamber aperture. Evidently with this change a

deep depression does not form in this cathode, and the

electron emission is maintained until some other limiting

process comes into play. The lifetime for Ar was given as

62

8 r: and as much as 1“ h for Ne?+ With this change the

ueseftll source lifetime becomes more a question of beam

attenulation due to a build-up of sputtered cathode material

iri arc: defining apertures and the extraction slit than are

staixility, so that making both the cathodes the smaller

dituneter gave no additional benefit. The geometry of the

'TAMU PIG is very similar to ours, with the main differences

beiru; the larger overall diameter and the location of the

vacuum feedthru near the arc chamber. Table 2.2 below

summarizes the differences between the two geometries for

tantalum cathodes, assuming the lower ion source

configuration in our case, with "outer" referring to the

cathode at the end of the high voltage rod.

Table 2.2. Comparison of cathode sizes and plasma

aperatures in the Texas A&M and MSU ion sources.

TAMU MSU
Inner Cathode Diam. 13 mm 9.5 mm (.375 in.)
Outer Cathode Diam. 7.5 mm 9.5 mm
Plasma Aperature 6.5 mm 5.2 mm (.203 in.)

 

Tentantalum cathodes Of 6.“ mm (.250 in.) diameter were

fabricated in order to test whether this technique could be
usmito improve the lifetimes of our pulsed argon and neon

beams. This cathode diameter is 1.2 mm larger than the MSU

63

wasma aperture, and hence just outside the prescribed range

according to the TAMU data, but we have the additional
problem of setting manually the rotational position of the

cathode, and the increased diameter was viewed as an

amwOpriate safety factor there. In addition, since the

position of the inner cathode in our source is more stable,
it was thought that locating the smaller cathode there would
insure that the arc would not fall off the side, should the

augnment be off or the position of the high voltage rod

move. During a 22 Mev/n 22Ne5+ run, two trials were made

1ch a small inner cathode, each after a run with a normal

cathode size.

Operation was found to be the same as with normal

cathodes--that is, if you did not know that there was a

small cathode in the source, nothing in the behavior during

striking and operation would lead you to believe that there

was some difference. In the first trial we were

over-confident, reducing the neon gas flow to 1.0 cc/min and

cutting out the nitrogen support altogether, and after one

hCHJP lost the source because of low gas flow. Inspection

icevealed a well centered and shallow crater on the small

calthode, with a similar sized depression on the other

ceathode as well. The second small cathode trial lasted

1 :5C111, with the failure being the arc instability

easssociated with cathode erosion. Inspection revealed a deep,

6“
well centered depression in the small cathode with thin side
walls. Obviously the reduced diameter was still too large!
Both normal runs ended in the usual way with the arc
dropping out and attempts to restrike being unsuccessful.
Table 2.3 summarizes the results of the four runs made. In
this table'HP'is the depth of the cathode depression and

"w" is the corresponding width.

iTable 2.3. Cathode wear characteristics with neon feed.

 

 

LIFETIME INNER CATH. JOUTER CATHT“
(hr:min) d w d w
(inches) (inches)
1. Normal 1:“1 .1“7 .236 .135 .236
2. Small 0:52 .050 .168 .081 .22“
3. Normal 1:25 .152 .222 .1““ .218
“. Small 1:50 .128 .220 .122 .220

 

Without getting too carried away with the numbers, it
is interesting to note that the lifetimes of sources 1,3 and
“ are correlated to the inverse of the average crater depth,
as shown in Figure 2.8, while the widths are a reasonably
constant .22 - .2“ inches. (Of course, source 2 was
inadvertently gas starved out and therefore will not
correlate.) Since we have established that wecuuirun the

small cathodes, the expected gain in lifetime ought to be

65
there; we just picked a diameter slightly too large--
something like .220 inches would work better.
Alternatively, increasing the plasma aperture would have the
same effect. In future work both of these options should be

explored.

2.6 Summary

 

Operating the ion source with a pulsed discharge has

5+ and Ar6+beams of sufficient intensity to use

given us Ne
for nuclear physics experiments. The advantages Of pulsing
include having more source tuning parameters for improved
control, the use of higher arc current without at the same
time reducing the arc voltage, less gas flow to the source
and achieving a lower average arc power. The disadvantages
include low source lifetimes for some beams, new
instabilities and more critical tuning, especially at low
duty factors. At low duty factors, the need for more arc
current from the power supply is evident, and in that
regard the K500 supply with a 7-10 A maximum current does

not exploit the full potential of pulsed operation, and may

be a contributing factor in the low lifetimes.

In addition, it seems that not all experiments can
tolerate the additional time structure that goes with pulsed

Operation, for when the arc current goes out on each cycle,

66
the beam current shortly follows. This is particularly true
for coincidence experiments. Consider a single coincidence

experiment, where the simultaneous measurement of events E1

and E are made. The rate of accidental coincidences (Ra)

J

will depend upon individual reaction rates R1 and RJ, and

the time resolution of the measurement AT ,

13

Ra 0: R R.AT (2.1)

13 13’

but the rate of each event must be proportional to the

instantaneous beam current (I ), and so Eq. (2.1) can

beam

also be written as:

R or I AT (2.2)

a beam ij'

At the same time the rate of true coincidences (Rt) must

also be proportional to the beam current, which we express

as

R « OI (2.3)

t beam'

Hence the ratio of true coincidence events to accidentals,

Obtained by dividing Eq. (2.3) by Eq. (2.2) is

R
_f_ a __‘___ . (2.11)
Ra Ibeam

During pulsed Operation, a much higher beam current is
produced during a shorter time interval, and so the ratio of
trues to accidentals must then decrease according to
Eq. (2.“). In practice one finds that this difficulty limits
the lowest permissible duty factor in coincidence

experiments to 50-80%.

CHAPTER 3

HAFNIUM CATHODES AND NITROGEN SUPPORTED ARCS

When PIG ion sources are used to produce heavy ions,
the source lifetime becomes a major issue. The costs of ion
source maintenance can be large, and ion source recycling
can become a significant fraction Of the accelerator
operating time [Ba 72,Hu 73]. For a PIG source with
discharge-heated tantalum or tungsten cathodes, lifetimes
generally fall in the range of 5-20 hours, with the lowest
lifetimes resulting from Operating conditions aimed at
producing the highest charge states of the heavier elements
LSh 81,Go 79,Ke 72,Ba 72]. Experience here has shown that

source lifetimes fall in the range of 2-10 hours with

tantalum cathodes in the K500 internal PIG sources [An 83,
An 8“]. The lowest lifetimes have occurred during the

production of neon and argon beams, while the highest have

3+

been for C and Nu+beams.

68

69
The finite lifetime is due principally to the
sputtering of the cathodes by ions from the discharge, as
illustrated schematically in Figure 3.1. Any one of the
following consequences of the cathode sputtering can destroy

the ion output.

1. An arc short resulting from the build-up of

sputtered material in the gap.

2. Low output as a result of sputtered material
choking the discharge defining aperture or the
source extraction slit. One observes a slow beam
attenuation with time and finally the run is

terminated.

3. Arc instabilities that result from the formation of
deep craters in the cathodes which tend to
extinguish the arc-- especially during pulsed
operation at low duty factors, where the arc

literally has to restrike on each cycle.

The worst effect of the formation of a crater is a lowering
of the surface electric field of the cathode, which must be
high initially to switch the pure Penning discharge to the
high power arc required for the production of
multiply-charged ions. If the arc goes out for some reason,

it will not restrike if the crater has become deep, and the

70
source must be rebuilt. This is the primary cause of
maintainence difficulties associated with the Operation Of
PIG sources, and it is clear that any increase in source
lifetime would be desirable. Unfortunately the ion
bombardment that causes these problems is also the source of
cathode heating, which is essential for the thermionic

emission of electrons that sustains the arc.

In this chapter the initial testing and development of
hafnium cathodes in the K500 PIG sources will be presented.
The motivation for this switch from tantalum is extremely
long lifetimes that have been obtained as a result. This
study was accomplished in the context of the Operation of
the K500 during the production of beams for experiments and
also during the development Of new beams. The advantage has
been a rigorous testing of the limits of applicability Of
hafnium cathodes under the conditions of actual use, but
the major disadvantage is the lack of precise test
conditions as would have been possible in an ion source test
stand. As a result only the most evident prOperties will be

discussed.

71

 
     

CQTHODE MQTERIQL
DEPOSITS

QNODE

 

 

 

 

 

 

 

 

 

 

Figure 3.1. A.schematic representation of one end of the
PIG arc geometry, showing the anode end, are defining
aperture, and the cathode. The effects of cathode
sputtering are illustrated.

72

3.1 Choice Of Cathode Materials

 

Good cathode materials for PIG sources must possess low
sputtering rates and high thermionic emission current
densities. Refractory metals are usually the material of
choice for PIG sources, since they best meet these two
requirements. Cost, availability and workability are

also important.

Consider the sputtering rate first. In a PIG source
large fluxes of positive ions bombard hot cathodes, and we
want to know what materials will give the longest lifetimes
under these conditions. The most extensive study of the
variation in sputtering yield (ejected atomsper bombarding
ion) was made by Wehner with 30-“00 eV mercury ion
bombardment first [We 57], and then the noble gases over
the energy range 50-600 eV [La 61,Ro 62]. Fortunately the
data span the energy range of positive ions in a PIG source.
The sputtering yields were found to vary nearly linearly
with bombarding energy in this range and to increase
slightly with bombarding mass. Some Of the elements found to
have relatively low overall sputtering yields, and have also
higher densities, which determines the volumetric sputtering
rate, were Ti, Mo, Nb, Hf, Ta and W. These "pure" atomic
physics facts are in good agreement with the experience
gained during the Operation of PIG sources with various

cathode materials [Be 72]. In Wehner's studies, hafnium,

73
the subject of this chapter, had a slightly higher
sputtering yield generally than Ta or W, so only a slightly
higher sputtering rate would be expected. On that basis
hafnium cathodes would be expected to show lifetimes
somewhat shorter but comparible to tantalum and tungsten in

PIG sources.

In addition, cathodes for PIG sources must be efficient
electron emitters at high temperatures. Thermionic emission
from pure metals follows Dushman's Equation [Bl 7“]. In its
simplest form:

J a AT2e_b/T,

where J is the emission current density in Amperes/cm?

A is a quasi-constant of 120.“ Amperes/cmzdeg?
T is the absolute temperature, and
b is the absolute temperature equivalent of the

work function.

The work functions of the above metals do not vary
significantly, so that temperature is the dominant
parameter, and as a cathode material, metals with high
melting points are favored. On that basis Ta and W from the
above list, with melting points of 3269 K and 3683 K,

respectively, would appear to be the best cathode choices

7“
for high power PIG sources. In this regard hafnium has
a larger emission current versus temperature as tantalum
[Ha 67], but has a much lower melting point (2“95 K), so the
maximum J is less. This is shown graphically in Figure 3.2.
In practice, the lower melting point would translate into

a lower are power limit for an Operating ion source.

3.2 Initial Use Of Pure Hafnium Cathodes- A Review

 

On the basis Of the above considerations, as a cathode
material hafnium would be expected to perform like the other
refractory metals-~but not as well as tantalum and tungsten
at high arc powers, where the lower melting point sets a

limit.

3.2.1 MSU K50_Cyclotrop

 

As a result of an Observation that hafnium targets were
difficult to produce by sputter deposition, in 1978 tests
were made of hafnium cathodes in a heavy ion PIG source irl

the K50 cyclotron, which was still running at that time.

75

HAFNIUM VS. TANTALUM THERMIONIC EMISSION

 

 

N:
6 T
\ l I
a A
>-102
p.
E
Lu10‘
O
I. HRFNIU" TANTALUM
1.12.110“
a
CK
D
010"
Z
0
(£194 1 g 1
E 1500 1900 2300 2700
w TEMP. (00)

Figure 3.2. The thermionic emission currents densities of
pure hafnium and tantalum are compared over the temperature
range of 1500-3000 °C.

76
Only a few runs were possible, but some Operating
characteristics quite different from those of tantalum
cathodes were observed [Mi 79]. With CO feed, a 2“ h
lifetime was obtained, as compared to 2-3 h for a normal

tantalum run, albeit at the expense of lower beam current.

One experiment in March 1979 took 1 epA of 70 Mev Cu+ ions
on target for 2“ h from a single source equipped with
hafnium cathodes [Ha 8“]. For comparison, with tantalum
cathodes the extracted beam current was 15-30 epA. The
lower extracted current was thought to be due to a lower
maximum cathode operating temperature, as a result of the
lower melting point of hafnium. The order of magnitude
increase in lifetime was however a surprise, and since the
lifetime enhancement was not Observed with nitrogen cu" neon
arcs, it was attributed to the formation of a HfC layer on
the surface of the cathodes. Hollow tantalum cathodes
filled with pressed-in HfC powder were tried, and longer
lifetimes were obtained, but the arc operation was erratic

and gave unstable beam output.

Since high beam intensities are not always required, it
was speculated at that time that hafnium cathodes might find
application in the K500 cyclotron when run imi‘the injector
mode, where the charge states required for acceleration

would usually be low.

77

3.2.2 Experience At Other Laboratories

 

Other laboratories certainly tried hafnium cathodes
after 1979, but generally could not see any advantage over
tantalum. The experience at Texas A&M was typical [Na 8“].
A longer lifetime was fXHHHi, but intensities were low for
the highest charges states where difficulties were
encountered in maintaining high are power without melting

the hafnium cathodes, in agreement with our experience.

Since higher beam energies, and hence higher charge

states, is generally more important than source lifetime,

hafnium cathodes did not come into wide use.

3.3 Hafnium Cathodes in K500 Cyclotron PIG Sourcep

 

A review was made of the testing of hafnium cathodes in
the K50 cyclotron, and on that basis it was expected that
for low-charge-state-carbon beams, an increase in ion source
lifetime over tantalum could be expected without a reduction
in beam current. Ten pairs of hafnium cathodes of the same
design as the normal tantalum cathodes were machined, for
use in the K500 cyclotron in order to make a comparison of

Operating characteristics with tantalum. The very first

test was made with a Cl“ beam, and a factor of three

increase in lifetime was observed with hafnium without

78
significant beam attenuation. This success encouraged trial
of hafnium cathodes for nearly all of the K500 beams.
Table 3.1 presents data on the lifetimes of sources operated
with hafnium cathodes for ion production in the K500
cyclotron, and Table 3.2 summarizes the Operating

characteristic of hafnium sources.

The largest increase in lifetime is seen foerH and

5+

N beams. TO get a feel for the magnitude of the lifetime

effect, at the end of the first cyclotron operating cycle in

5+beams had beeb obtained using only 13

June 1983, 3“O h of N
sources, and the same result would have required about 53
tantalum sources! Figure 3.3 shows samples Of the

long-lived cathodes from a N5+ source, which exhibit many of

the characteristics of used hafnium cathodes.

Generally hafnium cathodes flake or crumble upon
removal from the cathode holders, and as a result
measurement of the amount of material sputtered during
source Operation contains a large uncertainty. In addition,

the large variation in are parameters that occurs during

79

Table 3.1. K500 Operation with Hf cathodes. Ipoetimes for

tantalum cathodes are given where available.

 

 

£9! Eflé EAEHIEE 1851529!
tmeafg ) tmaih) tmeaAh)

"He‘+ 1% He+N2 12. 20.5
61.12+ 96% 6LiF 37.3 67.6

pellet+N2
71.12+ LiF pellet 16. “2.5

+N

2
1203+ 00,00 + N2 10. 19.0
120"+ CO,CO + N2 1“. 31.6 9.6
1”113+ N2 15. 27.8
11‘11"+ N2 23. 55.0 .8
114N5+ N2 2“. 55.1 .“
16 “+
O 02 + N2 10. 13.9

20’22Ne5+ Ne 3.3 “.6 .5
quré+ Ar 2.0 2.“ .8

 

80

Table 3.2. A summation of the main characteristics of K500
ion sources operated with Hf cathodes.

1.

Extended lifetime.

 

2.

The best case is with N2 feed, with C0 + N2 almost as

good, and CO the next best. The emergence of N2 feed

as giving the longest source lifetimes was quite a
surprise. The lower lifetime with CO results from
carbon deposits that form on insulators, making the arc
unstable, and not specifically cathode wear.

Spurce striking.

 

The source strikes at a lower arc voltage- ~1000 volts
versus 2500-“000 volts for tantalum. Normally it is
not possible to restrike a tantalum source after more
than 1 hour Of Operation. Hafnium sources restrike
after many hours. This is difficult to quantify, as
adequate statistics have not been kept, but the

interval is greater than 20 hours for an N2 fed source

and greater than 10 hours for a CO fed source.
Multiple source restrikes are also possible in these
cases.

Cathode wear.

 

Very shallow craters form in the cathodes when
operated DC.

Beam attenuation;

 

Less sputtered material builds up in the anode bore,
resulting in slower beam attenuation with time.

Noble gas feed.

 

Same behavior as tantalum cathodes for Ne or Ar gas
feed- deep craters form in the cathodes and no
significant improvement in lifetime is observed.

Cathode-:egd ggs reactions.

 

Used cathodes show strong chemical changes after
operation with N CO or 0 gas feed. The cathode

2’ 2
color is glossy gold for N2 feed, burnt black for CO

feed, and dark gray for O2 feed. Often the cathode is

found to have fragmented, and for O2 feed exhibits

swelling. No reaction to noble gas feed is observed
(ofcourse!).

81
Table 3.2.(cont'd.).

7. Temperature effects.

 

There is Often evidence of partial cathode melting,
especially at the cathode center. With more than
about four amperes of arc current DC, hafnium cathodes
in K500 sources have melted.

8. Pulsed ion source Operation.

 

Roughly the same beam currents can be obtained as with
Ta cathodes, but at much lower duty factor. Depending
on the duty factor, the lifetime enhancement is still
possible however.

9. Normal failure mode.

 

For long lived hafnium cathode sources, the cathodes
crumble and the source dies in a short.

 

82

 

Figure 3.3. 55 hour lifetime hafnium cathodes for N5+ show

small molten domes at the base of the craters and crumbled
edges— the latter evidence of a structure changing reaction
Of the hafnium and nitrogen gas. New hafnium cathodes
resemble other refractory metals.

83
normal Operation eliminates the possibility of relating
cathode wear to arc parameters. Finally, there are not many
cases in which hafnium and tantalum cathodes are run under
the same conditions and so, comparing the rate of wear to
that for tantalum cathodes is not possible for all the cases

in Table 3.1.

During the development of an Ar6+ beam, however, both
‘tantalum and hafnium cathodes were used under the same
conditions. Since hafnium does not react with argon, the
cathodes remained hard and could be removed intact. The
sputtering rate was measured in this case and the results

are given in Table 3.3.

Table 3.3. Hafnium and tantalum cathode sputtering rates
with argon gas feed for 10 sources, used during the

 

 

development of Ar6+ ions.

Material Cases - M T S
(mg.) (min.) (mg./min.)

Hf 6 855 1“0 6.10

Ta “ 7“O 80 9.25

The tantalum sputtering rate is 1.5 times the rate for the
hafnium in these cases with argon feed. This is in good
agreement with the ratio Of the average lifetimes of hafnium
and tantalum sources with argon feed, as given in Table 3.1.

On that basis one can estimate that the tantalum sputtering

8“
rate with nitrogen feed is four times the sputtering rate of

hafnium.

Up to this point, we been considered only the lifetime
performance Of sources with hafnium cathodes. The source
output current is, of course, also affected. For low charge

state beams (L12+, 03+, N3+, Nu+), there is no difficulty in

obtaining cyclotron extracted currents greater than 10 pnA,
(which statisfies minimum requirements for nuclear physics

.1.
experiments) with hafnium cathodes. For N5 beams under

certain conditions, hafnium cathodes still provide adequate

current, but for Ne5+ and Ar6: tantalum cathodes must be-

5

used. N +was run pulsed or dc, with hafnium or tantalum

cathodes. These data are summarized in Table 3.“.

Table 3.“. Nitrogen 5+ beam currents with hafnium and

tantalum cathodes. N2 feed is assumed in all cases.

 

 

CATHODE DUTY EXT.CURRENT MEAN MAX.

TYPE FACTOR (enA) LIFETIME LIFETIME
(%) (h) (h)

Hf 30 120 29.9 55.1

Ta 100 100 6.“ 10.6

Ta 80 150 5.5 7.6

 

As can be seen in Table 3.“, three basic Operating

modes can be distinguished. In the first, Hf cathodes are

85
used to obtain long life, but pulsing at a low duty factor
is required to maximize the beam current. In the second
mode, Ta cathodes are used, the beam intensity is roughly
the same as in the first case, though obtained do, but the
average lifetime is “ times lower. This mode is necessary
when the experiment cannot tolerate the pulse structure,
such as in coincidence experiments. If a small amount of
pulsing can be tolerated, then a third mode with Ta cathodes
is used, and there is about a 50% increase in the extracted

beam versus do.

In connection with the data in Table 3.“, the arc

5+ production with hafnium and

parameter ranges for N
tantalum cathodes Operated DC are presented in Table 5. For
the same gas flow and are current, the arc voltage is
significantly lower for hafnium cathodes Operated do than

for tantalum cathodes under the same conditions. This will

be discussed in Section 3.“.

Table 3.5. Comparison of the range of do are parameters for
sources equipped with hafnium or tantalum cathodes, when

used to produce N5+ions in the K500 cyclotron.

 

 

CATH. ARC ARC ARC EXT.

TYPE VOLTAGE CURRENT POWER CURRENT
(Volts) (Amps) (KW) (enA)

Hf 200-300 2.0 O.“-0.6 20-“0

Ta 500-700 2.0 0.8-1.“ 75‘150

86

3.“ Analysis of Lifetime and Beam Current Effects

3.“.1 Cathode lifetimg

 

The dominant cause of these effects is the
significantly lowered sputtering rate over tantalum or pure

hafnium for N2 or CO feed and the same Operating conditions.

The key observation is that when a non-reactive gas
such as neon or argon supports the arc, hafnium cathodes
show the same deep crater formation as tantalum cathodes,
with the arc eventually going out in the usual way. With
the noble gases, what we are seeing is the sputtering of
pure hafnium, which differs from tantalum only in a small
way, so cathode lifetimes must also be similar. As seen in
Table 3.1 for Ar or Ne beams the average lifetimes for
hafnium are slightly larger than for tantalum, by factors

1.1“ and 1.33 respectively.

On the basis of the sputtering yield data for pure

__---‘-_--__--_---_——-“—-----__‘--‘-__-‘—--—--_--

3‘
D.)
’1
:3
H
C
5
3
O
(D
3
13‘
D)
D
O
(D
a
(D
:3
(T
H
D
d
3'
('D
I-‘
H
'1
(D
('7‘
H
3
(D
Z
O
C
F"
O.
0'
(D
10
IX
1'0
(0
O
c?
(D
D.

for nitrogen or carbon dioxide supported arcs. However, the
source lifetime with nitrogen feed is four times longer when
hafnium cathodes are used. For CO, the picture is not so
clear because cathode wear is generally not the cause of

source failure, as mentioned in Table 3.2. Furthermore,

87
K500 sources for carbon ions have been run with a mixture Of

CO and N2 to reduce the build-up of carbon on support

insulators. We will therefore concentrate on the nitrogen
feed case,1nuch is the main new result, knowing that the
previous K50 cyclotron testing of hafnium cathodes reported
a lifetime effect for CO feed. The present study has seen
lifetimes for CO feed longer than those previously reported,
so that the analysis here will also be applicable in

principle to that case as well.

Since the sputtering yield data for pure hafnium would
predict a sputtering rate about the same as tantalum, the
preponderance of evidence points to the reaction of the
hafnium with the arc support gas, and the subsequent
formationcn’a layer with a much lower sputtering rate at
these energies. This kind of lifetime enhancement has
in fact been Observed before, with a different set of
elements. Irlea early study, extended lifetimes were noted
in a pulsed low power PIG source, when aluminum oxide or
beryllium oxide cathodes were used in place of pure aluminum
or beryllium cathodes [Go 53]. In addition, pure aluminum
cathodes, initially exposed to pure oxygen in an arc, had
the same lifetimes as aluminum oxide cathodes. In this case
it is the nitride and carbide, rather that the oxide that
gives the longest lifetimes with hafnium cathodes. In that

regard, it is no coincidence that hafnium nitride is the

88
highest melting point nitride, and hafnium carbide is the
most refractory binary compound of all [Ha 7“]. As a
further corroboration of this effect, HfC cathodes have been
used at Karlsruhe and other laboratories since 1972, where

it was found that their performance was superior to other

3+ in a high power PIG arc

materials for production of Li
[Sc 8“]. There the high melting point of HfC is exploited,
with the lithium vapor coming from a small oven built into

the source, and so the lifetime increase with N2 or CO feed

could not be known.

3.“.2 Hafnium Cathodes and Beam Current

 

The best comparison between hafnium and tantalum

cathodes on the issue of beam current can be made for the

5+ions, and Tables 3.“ and 3.5 are relevant.

production of N
Hafnium cathodes in a dc ion source give a much lower beam
current than tantalum. The puzzle here is that if one
starts with a hafnium source pulsed at low duty factor, no
combination of are parameters can be found that will allow
this same beam current, but without some pulsing. On the
other hand, if one starts with a hafnium source running do,
and then adjusts arc parameters so as to maximize the beam,
one ends up with a low duty factor! The required duty

factor itself is not a fundamental quantity, and depends on

the balance achieved between gas flow rate and are current.

89
Nevertheless, the trend is clear-- hafnium cathodes are

5+ when Operated do. This

unsuitable for the production of N
can be illustrated in several ways. Figure 3.“ shows the
dependence of beam current on duty factor for an arc current

of 0.6 A and a N2 gas flow of 1.5 sccm. The beam current

steadily declines as the duty factor is increased, to about
half the peak value at 80 %. Furthermore, if one plots a
scattergram of optimum beam current versus duty factor for

5+beam runs in a one year period, as in Figure 3.5, no

all N
correlation between duty factor and optimum current is seen,
as the extracted current depends on source parameters and
non-source parameters as well, but there is a definite

clustering<fi1the hafnium runs at low duty factor and the

tantalum runs at nearly a dc operating level.

At first one would expect that the lower melting point
Of hafnium sets a lower thermionic emission current limit,
and that this then is the reason for the poor performance.
However, we have run tantalum cathodes at low power do,
where the emission currents are similar, and yet the beam
current is still higher with tantalum than for hafnium

cathodes.

9O

 

w: 1 T 1 T
0
°\ 35 5+
\,0 N

“.8 "" \ 4
E \
I:
:06 ‘- \ -1
a; \
E 0 \ 0
x.“ _. \ .1
g; \
=5: \
\
'_'.2 F’ —

HAFNIUM CATHODES

 

 

.0 J l l l J

40 60 80 100
DUTY FACTOR (Z)

 

Figure 3.“. With hafnium cathodes, the N5+current drops

with increasing ion source duty factor. In this case, the

N2 flow rate was 1.5 sccm and the arc current 0.6 A.

91

 

 

 

 

5

3180 _ 35N5+

H160 _.

(Time _ "
glgg _ “mm: TANTALUH
c2: 80

UJ " x X

m 60 _.

C.

E '40 1- X X)$<

g 20 1.

195 0 1 § 1xx1 1 1 1 1
ES 0 20 “0 60 80

 

DUTY FACTOR (2)

Figure 3£L A history of the maximum cyclotron extracted

beam current versus source duty factor for N5+ runs and
hafnium or tantalum cathodes in K500 cyclotron PIG sources,
over the period March 1983 to January 198“.

92

The explanation of this effect requires the
introduction of another important cathode property-- the
cathode contribution to the source gas pressure. 'This must
be low for high charge states to appear in the source output
current, because the charge exchange losses are lower
[Ma 76]. 'The arc voltage is indirectly proportional to the
internal source pressure, as can be easily seen by turning
up the gas flow rate. What we find is that for the same arc
current and gas flow, the hafnium source runs at a
significantly lower arc voltage than a tantalum source, as
was shown in Table 3.5. The difference in arc voltage at
the same are current and gas flow implies that the pressure
inside source is higher for a hafnium source. As a further
corroboration of this effect, the cyclotron beam chamber

pressure is higher when hafnium cathodes are used for N5+

ion production, as shown in Figure 3.6, where the vacuum is
plotted versus gas flow rate for the same cases as in
Figure 3.5. One sees that the beam chamber vacuum declines
as the gas flow increases, but that for the same gas flow
rate there is large difference in vacuum on the basis of
cathode type alone. We infer that the higher pressure
results from a higher vapor pressure as a function of
temperature for hafnium than for tantalum under the same
conditions. A comparison of vapor pressures of hafnium and
tantalum in the range Of 2000-2500 °C, in Figure 3.7, shows

that the hafnium vapor pressure is 5 orders of magnitude

93

 

 

 

 

t: l T T T l I x]
3 35 5+

N x
11202- "1
" x
z: xX
3 9.9.. HAFNIUM ..
L)
a: ’“x
> XX Xx
o; 6.0— 5* .J
151 Xx’E X
:1: 3.0 r-TANTALUH -+
C)
2:
55 .13 1 1 1 1 1 1 1
d1 .0 1.0 2.0 3.0 4.0

NITROGEN GAS FLOH (SCCM)

Figure 3.6. Hafnium sources (X) operate at higher pressures
than tantalum sources (*), for the same gas flow, as shown

5+

or the same group of N as in Figures 3.“-3.5.

9“

TANTALUM AND HAFNIUM VAPOR PRESSURES

 

10{

10" HAFNIUM
tie”-
0
$310“
0310's
:’109
107

lgfgr 1 1 1 1
1500 1900 2300 2700 3100

TEMP. 1°C)

Figure 3.7. A comparison of the vapor pressures of hafnium
and tantalum over the range of 1500-3“00 °C [R0 63].

TANTALUM

 

95
larger at the same temperature [R0 63]. Since one of the
main effects of lowering the duty factor is to lower the
average cathode temperature, there would also be a decrease
in the source's internal pressure as the duty factor is
lowered. The cathode temperture is not a directly knowable
quantity, but must depend directly on the average arc power
dissipated in an operating source, and one would expect that
the average are power must decrease with decreasing duty
factor-- for the hafnium vapor pressure to be the cause of
the lower currents when operated do. As Figure 3.8 shows,
this is indeed the case. Hafnium sources run at less than

half the average power level of tantalum sources for the

production of N5+ beams.

3.5 Summary and the Impact of Hafnium Cathodes

 

By replacing tantalum cathodes with hafnium, one
obtains longer lifetime in exchange for lower extracted
current. Pulsing the arc, when permitted, partially
compensates by increasing the beam current. The higher
vapor pressure of hafnium best explains this observed
difference in performance. A factor of four increase in

source lifetime is observed with N2 feed, and about half

that for CO feed. These longer lifetimes would not be
expected on the basis of sputtering yield data for pure

hafnium, and must instead be due to the formation of

96

 

 

 

 

E
C: I T l l 1 I
8 35 5+

120 1. x N ..
E
or. HAFNIUM TANTALUI‘I
% 80
o "' x 5" 1

X ¥
g so .. .
LLJ
X

CD

“0 1.4,. ..
CD
L1J - ’3, a1:
'5 20 _. x 316 x ..
CI )( fit

X X
E 0 L x1 1 1 * 1 1
if, .0 .11 .8 1.2
ARC PONER (klnl)
Figure 3.8. For N5+, hafnium cathodes are found to operate

at low average arc power when compared to tantalum cathodes.
Note the lack of correllation between beam current and
power.

97
(“unpounds of the feed gas and the hafnium. In other cases

where N2 can be used to partially support the arc, some

degree of lifetime enhancement usually results.

On an absolute scale we have had many nitrogen sources
that ran in excess of 50 hours. In addition, hafnium
sources can be restruck many times, and after many hours.
Sources are Often turned off during energy changes, and left
in the cyclotron to be restruck, and this further reduces
the number of required source changes. Under those
conditions, the ion source ceases to be an important
Operational issue. This carries over to situations in which
nitrogen can be used to partially or completely support an
arc used to produce other ions. As will be discussed in
Chapter 5, the production of ions of metals and other solids
has been added to the K500 sources via back-bombardment

6’7Li2+ ions the arc can be supported

sputtering, and for
with nitrogen, and thereby very long somwleifetimes are
obtained as with nitrogen beams. We see the lifetime effect
also with CO feed, but C0 has a deleterious effect on the HV
stand-off insulators in the source, as carbon deposits form
on the surface and shorts result. However, by mixing in

some N this process can be controlled and the lifetime

2 9

increased.

CHAPTER “

ION SOURCE RADIOACTIVATION CORRECTION

In the early K500 cyclotron runs,.we were mainly
accelerating carbon and nitrogen 3+ and “+ ions, and we
soon encountered a serious ion source activation problem
with these beams. Activities of a few hundred mrem/h were
typical, and some rates higher than 1 rem/h were observed.
In one case, shown in Figure “.1, a 56 hour long run of

30 MeV/n 1“NJ4+ resulted in an average level of over

100 mrem/h on the source, with the highest activation on

surfaces protruding the most from the cyclotron center.

At that time we had only two ion sources, and often a
hot ion source had to be recycled right away to support the
cyclotron Operation. Eventually it was determined that the
accelerating ions were stripping to the next higher charge
state through an interaction with the residual gas. The

subsequent "walk" of these off-centered trajectories through

98

99

MSU-83-283

 

(: IINCH

L I
I 1

Figure “.1. The activation of the K500 PIG source is shown

after a 56 h run Of 3OMeV/n Nu+ions. The arrows with
numbers at their tails indicate measured radiation levels in
mrem/h. The ion source back protrusion is not shadowed by
the dee tips, and becomes more activated.

100
the cyclotron central region resulted in a bombardment of
the ion source. Since the internal energy of K500 heavy ion
beams easily exceeds the coulomb barrier for copper, such
activation is possible. This process also makes the dee
tips hot, but since they are not routinely removed from the

cyclotron the hazard is not as great.

The analysis of this problem led to an analytical
formulation of tune systematics that showed a dependence on
the cyclotron K factor [Ma 8“], and thereby explained why
this is such a problem in the K500 cyclotron, where the K of
500 is about three times that of room temperature heavy-ion
cyclotrons. This chapter will start with a brief review of
that analysis. The paper also included a suggestion for a
method for reducing the hazard to personnel handling the
sources, and it is the main purpose of this chapter to
demonstrate how that works. The solution is to cover the
median plane surfaces Of the ion source with thin tantalum
shields that can be removed during handling to a safe
shielded area. These shields, now in place, are not however
thick enough to stop the highest energy K500 beams, and we
will demonstrate that as well. Finally, after analyzing the
problem here, we observed in a report from the RCNP
Laboratory at Osaka University in Japan that they had
encountered similar difficulties with the activation of
their PIG sources when run in the 230 cm. cyclotron. An

analysis was made that showed that this stripping process

101
could be the source of their ion source activation. One can
show that there will be no activation at all if the copper
anode surfaces are replaced with tantalum, since returning;

ions will not exceed the coulomb barrier on tantalum!

“.1 Review of the Ion Source Activation Model

 

The analysis of this problem has been the subject of a
paper: [Ma 8“], of which this student was a co-author. The
details of the derivation of the condition for activation
will not be repeated here. In summary, if an ion of charge

qe and mass Amo, is accelerated to an energy E(r) at a

radius r where its charge is shifted to (q+1)e, then it can

hit the central region if

(“.1)

1:!"

>LD
1+
‘_a
[v

< a
E(r) ' Emax

For activation to occur, the particle energy E(r) must be

greater than the coulomb barrier (E ), which in the lab

Cb

frame and evaluated in MeV is

E = 1.11“ ~~~~~ ) --------------- . (11.2)
Cb A2 1.“5(A:/3+A;/3)

102
In Eq.(“.2) subscript 1 refers to the projectile and

subscript 2 to the target, while Z1 and 22 are the atomic

numbers of the projectile and target, respectively. Here
1.“5 fm is used for the coulomb barrier radius [Ev 82].
With this pair of equations one can study the ion source

activation problem.

We have found several ways of demonstrating why the
level of activation is high in the K500 cyclotron, and one
of the best is shown in Figure “.2. For the case Of any
“+ ion stripping to 5+, the maximum energy for ions hitting
the central region is calculated from Eq.(“.1). This is
plotted in Figure “.2 as a function of ion mass (A) for
three different K values: 100, 300 and 500. The last case
corresponds to our cyclotron, for a central field of

B0: “8 kG, while the region near K = 100 is where the

conventional heavy-ion cyclotrons in this country Operate.
The coulomb barriers for Cu and Ta are also shown. The
region above the Cu coulomb barrier for K=1OO is small,
corresponding to the lightest ions only, whereas it is quite
large for the K=500 case. Figure “.2 can be used to
introduce the topic of the next section--since the coulomb
barrier for tantalum is about twice that for OOpper, it is
Obvious that less activation will occur if tantalum is

placed in the median plane elements of the ion source.

103

MSU—83-278

A
O

Z=4+

E/A (MeV/n1
N 01
O O

('5

 

 

 

 

 

 

 

 

 

x) 15 20 25
ATOMIC MASS

Figure “.2. The maximum energy of particles that can hit
the central region is shown for charge state “+, and
stripping to 5+ and for three different K values as a
function Of atomic mass. The coulomb barrier for copper and
tantalum is also shown.

10“

“.2 Removable Median Plane Tantalum Source Shielding

We took the simplest approach in dealing with this
problem-- the median plane surfaces were covered by
removable tantalum shields. First, the activity Of the ion
source c0pper is reduced because the tantalum stOps some
fraction of the ions, and second, the activation of tantalum
is less than an equivalent thickness of copper because the
coulomb barrier is higher. To do a complete median plane
shielding, copper must be removed from the source surface to
accommodate the shield, since this surface forms one side of
the dee voltage gap. 'Also the ion source must pass through
an alignment collar with a 0.001 inch clearance on the
source diameter as it slides into position in the cyclotron,
and no part of the median plane section of the source can
protrude beyond that diameter. The only two sources
available for the early runs were unfortunately too
radioactive to allow this machining, and so they received.
thin (0.010 in. thick), partial surface shields. In later
sources this shield was extended to cover all the exposed
copper. The effectiveness of the tantalum shielding is
shown in Figure “.3 for the case of the initial partial

“ “
shields, after the 56 h run of the 30 MeV/n 1 N +beam of

Figure “.1. As can be seen, the activity at the surface of

the source is reduced by more than a factor of two after the

105
shields are removed. Figure “.3 also shows why later
sources had to have complete shields-~the area of the
highest activation was not covered by this partial

shielding.

It is clear that some of the ions are getting through
the thin tantalum shield, and as the energy of cyclotron
beams is increased, one may have to accept increasingly
greater transmission of the bombarding particles through
the tantalum. We can use Eqs.(“.1)-(“.2) to determine how
thick the shield must be to stop the bombardment in the
tantalum over the whole range of stand-alone beams that the
K500 cyclotron can produce. The maximum activation energy
will occur for the case of the lightest mass species that
can strip from q to (q+1) electrons removed (assuming also
K=500). For these maximum bombarding energies (which were
computed from Eq.(“.1) for various masses and charge
states), mean ion depths were obtained from range tables for
copper and tantalum [Li 80]. Figure “.“ shows the results
from the calculations Of the stopping range in copper and
tantalum at the maximum activation energy. A 0.5 mm (0.020
in.) tantalum shield, only twice as thick as that currently
in use, VHMJld stop all but the lightest K500 cyclotron
beams. This would substantially reduce the hazard to
personnel handling the activated sources since nearly all of

the activity would be gone when the shield was removed.

106

MSU-84- 432

40

\

/
\ 40 \200

I

100

80

200/ \ \
300

SOURCE WITH SHIELD 200/ SHIELD ONLY

Figure 1Lg3. A median plane cross-sectional view of an
activated ion source with tantalum shields in place and
removed is shown. Much of the activity disappears when the
shields are removed, thus reducing potential exposure.

107

MSU-83-275
2.01-

 

 

123 567

1
Figure “.“A 'The stopping;range in copper and tantalum for
the maximum energy particles that can hit the central region
plotted as a function of initial charge (q). The atomic

mass for the lightest particle that could strip to the next
higher charge and K - 500 was used.

108

Finally, after we had made an analysis of this problem
at MSU, we came across a reference in section 8.3 of the
1982 RCNP Annual Report to source activation problems that
were encountered after running heavy-ion beams in the K=120
cyclotron there. It turns out that the ion source anodes
there are also made from copper. We decided to find out if
this stripping process could be the source of their
activation, and if so, how hard it would be to correct.
Table “.1 shows the results of applying the activation
calculations to three of the beams listed in that annual

11 3+
report. It shows that B

at 95 MeV and 12C3+ at 80 MeV
will both cause central region activation in that cyclotron

while 1uNlHat 135 Mev will not. Note however that no

activation would occur if the copper ion source anode were
replaced with tantalum. Note also that this situation is
much more favorable than ours, since our K values are

higher.

109

Table “.1. Calculation of the activation expression of
Eq.(“.1), with E(r) - Ecb' for selective beams of the RCNP

Laboratory K-120 cyclotron, to determine whether that
process could be the source of their reported ion source
activation.

 

BEAM ENERGY COULOMB BARRIER Ecb/Emax s 1?
On Cu On Ta On Cu On Ta
(MeV) (MeV)

1183+ 95 26.0 “8.8 0.61 1.1“

12C3+ 80 32.8 58.3 0.90 1.63

11N1+ 135 38.5 79.1 1.1“ 2.31

 

CHAPTER 5

THE PRODUCTION OF IONS FROM SOLID MATERIALS

Up to this point, consideration has been given only to
the production of ions of elements that can be fed into the
PIG source as a gas. This is in fact a rather fundamental
limitation, since the design of the source and also the ion
source power supply actually depend on the initiation of an
arc in a low pressure gas. However not all elements are
available in gaseous form, and in many other cases the
gaseous form is highly toxic or corrosive and requires
special handling techniques that preclude routine use.
Since a large and important region of the Operating diagram
of a heavy-ion accelerator is potentially excluded, some
method of making available gases or ions liberated from

solid materials, must be developed for each PIG source.

Several methods have been developed for the purpose of
converting solids to ions in PIG sources. Ovens connected

directly to the arc chamber have been used to boil solid

111
materials to produce vapor for admission into the arc
[Be 69, J0 72, Va 69]. In this case there is very little
control over the material consumption rate and ion
intensities. Another approach has been to insert electrodes
fabricated of the desired element directly into the plasma,
where some erosion via ion bombardment sputtering takes
place--to control the rate Of sputtering the electrode can
be biased [Ga 72, Ga 75, Pa 72, So 76]. These methods have
proved successful but generally require the development of
some complex auxiliary apparatus, and then one must find

space in which to locate it.

The discovery by the ORIC Cyclotron Group at Oak Ridge
National Laboratory that nonresonant ions returning to the
source have sufficient energy to cause substantial
sputtering of source structures [Hu 7“, Hu 76], is the basis
of another method. This seems ideally suited to
requiremenmscfi'a heavy-ion cyclotron ion source, in that
the mechanism is provided by the existing rf field of the
cyclotron and, in addition, changes in the source to
accomodate the method are minimal. We chose to apply this
back-bombardment sputtering technique develOped at Oak Ridge
for the production of ions from solids, and in this chapter
the application of this technique in K500 ion sources will
be explained. First, a brief review of the origins of the
process and the subsequent development as a viable method

for obtaining ions from solids will be reviewed. Next,

112
physical modifications made to the K500 ions sources to
implement this technique will be introduced and justified,
and the plan for testing those design changes will be
presented. The main result of that work is that it has now
become possible to accelerate lithium beams in the K500
cyclotron, and this Opens up the possibilities for other
species as well. An analysis of the ion dynamics at the
center of the cyclotron that results in one Of our beams,

7L12+, has been done, and will be discussed.

namely 22 Mev/n
The agreement between the calculations and the experimental

situation is quite good.

We are concerned here with the first acceleration gap,
which, for reference purposes, has been identified in

Figure 5.1.

5.1 Origins of the Back-bombardment Sputtering_:echnigue

A spurious beam with large beam current of 0.5 euA was

130 12

observed during the acceleration of a Xe +beam in the

Oak Ridge Isochronous Cyclotron. Subsequently this extra

beam was identified as being 65Cu6+, and coming most likely

from an erosion pattern in the inside of the back of the ion
source chimney [Hu 7“]. Several studies were made of the

properties of this beam. The replacement of the xenon gas

113

 

|———_l
1 inch

Figure 5.1. The central region of the K500 cyclotron is
illustrated for reference purposes in subsequent
discussions. The shaded areas indicate the median plane
cross-section of the dee tip electrodes and the ion source,
and gap voltage contours are shown. The electrode on the
tip of the first dee after the ion source is called the
"puller electrode" and the boxed region is the critical
first accelerating gap.

11“
in the source with neon cut the copper beam current a factor
of 20, indicating that the mass of the ion source feed gas
was important to the process. The replacement of the
eroding copper section of the source with nickle completely
eliminated the copper beam, confirming the copper "source".
Orbit computations showed that low charge-to-mass ions were
failing to cross the 1 cm accelerating gap between the
source and the puller electrode for nearly all starting
times during the negative voltage part of the accelerating
cycle. These ions are accelerated back into the source when
the voltage on the dee changed sign and struck the back with

energies as high as 30 keV.

Additional studies at Oak Ridge expanded the range of
elements in which ions were made this way by changing the
material making up the back of the chimney, establishing the
viability of the method in the process [Hu 76]. Further
orbit computations, coupled with estimates of the sputtering
effectiveness of different source feed gases, led to
predictions that krypton and xenon would be the most
effective for sputtering nickle. The measurement of the

58

extracted beam intensity of N15+

ions for several gases

confirmed this.

115

5.2 Design Considerations for the K500 Ion Sources

 

In preparing a source to Operate on this technique we

felt the following issues were important:

(1) Compatibility with existing source design--

allowing for easy change-over.

(2) Properly sized feed material insert. (This in part
determines the useful lifetime, but not absolutely
because the support gas species, its sputtering

yield and the slit size also play a role.)

(3) Feed material inserts in a standard size, simple

shape for all materials.

(“) Operation where possible of the ion source on
nitrogen support gas with hafnium cathodes to
obtain long source lifetimes (recognizing that
whether nitrogen can be used will depend on its

sputtering yield with particular feed materials.)

The solid to be sputtered must be placed in the chimney
back. Then the key design constraint for us is the close
proximity of the high voltage rod behind the chimney in our
existing source design. Also, the feed material must come

in contact with the chimney, which has an unknown

116
temperature, but has been observed to be incandescent on
occasion. The design that we have chosen is shown in Figure
5.2. A double-tube high voltage rod end is traded for the
standard .250 in. O.D. copper tube to Open up the space
behind the chimney. This allows a tantalum receptacle to
be fused to a standard tantalum chimney for holding the feed
material. With this design we can put 0.170 inch thick feed
material inserts in a standard source. The design
clearances between the high-voltage rod and the grounded
chimney are within the range of acceptable clearances for
holding voltage in the source, even when including typical
assembly tolerances. This would change however if the
high-voltage rod is not positioned parallel to the back of
the chimney, as the resulting voltage holding asymmetry

would certainly cause arcing.

The other important part of this development is the
selection of suitable test beams. These are summarized in

Table 5.1. The initial goal of the work was the develOpment

6’7Li2+beams, but since lithium fluoride melts at about

of
900°C, and the operating temperature Of the conductively
cooled tantalum chimney was not known (but on occasion

observed incandescent), the first beam trial would instead

11 2+ . .
be B , u51ng boron nitride as the feed material. After

that, lithium beams were attempted. Lithium floride was

available in powered form, with natural abundances of

117

   

DOUBLE TUBE

 

 

 

 

\\\\\\\K\\\\ \\\

 

 

 

 

FEED MATERIAL

CHIMNEY WITH
BACK CHAMBER

 

 

 

 

 

Figure 5.2. The conversion of the source for the production
of ions from solids was accomplished thrOugh minimal
changes in the design of the standard source. A chimney with
a welded tantalum receptacle located in the back wall behind
the extraction slit holds the feed material insert, and a
double tube high voltage conductor, thin in the direction of
the chimney, feeds the upper cathode in place of a larger
diameter coaxial conductor.

118

6 7

isotopes Li and Li of 7.5% and 92.5 % respectively, or

with the 6L1 in various enrichments. We started with

lithium floride containing lithium isotopes in natural
abundances. This was then packed in a carbon boat and
heated to the melting point, where upon cooling it
solidified. The ingots were then machined to get the

required insert shape.

The 6Li2+and 71.12+ beam in Table 5.1 happen to be by

choice analogs (same q/a) as two well developed beams of

12C“+ a 1“ “+

ions nd N , respectively. By first optimizing the

acceleration of the analogs and then making a frequency
change, the lithium beams would be found. In that regard

6 2+

the expected frequency shift is -.0““ MHz for Li and

7 .2+

-.030 MHz for L1 .

5.3 Production of Boron and Lithium Ionp

 

There was time to try only the first three beams
entered in Table 5.1. The source modifications worked and
the results of these trials are summarized in Table 5.2.

In all three cases the run was stopped to switch to a new
test of some other cyclotron component (the cyclotron was
undergoing intensive development at that tinue‘that and the

development Of solid feed was only one aspect). There were

119

Table 5.1. Beams chosen for the initial testing of the
source nuuiifications made to incorporate the
back-bombardment Sputtering technique into the design.

 

 

ION FEED SUPPORT CATHODE BEAM DEE
MATERIAL GAS MATERIAL ENERGY VOLTAGE
1182+ BN N2 Hf 38.1
6L12+ LiF N2 Hf 30.0 60.“
7Li2+ LiF N2 Hf 22.0 67.0
19F3+ LiF N2 Hf * *
27Alu+ Al Ne,Ar Ta 10.0 50.9
2”Mg!4+ Mg Ne,Ar Ta 10.0 52.9

 

*Beam parameters not selected in advance.

120

Table 5.2. The first beams in the K500 cyclotron produced
by back-bombardment sputtering of solid feed inserts.

 

 

ION EXTRACTED BEAM SOURCE
CURRENT (enA) LIFETIME (h)

1182+ >1.0 >13

6Li2+ 12.0 > 8

7L12+ “00.0 > 9

 

no significant difficulties with the design during this
testing period. Inspection after the boron run indicated
that some arcing from the high-voltage rod to the chimney
back had occurred. A greater effort was made to insure that
the high voltage rod was straight for the two lithium runs,
and no further arcing was observed. Certainly since that

time the L12+ beams have been used extensively, but this

additional experience has not been included in this
discussion. We will simply note here that with hafnium
cathodes and nitrogen gas supporting the arc, we have
obtained lifetimes in excess of 20 h and extracted beam
current averaging 200 enA. The physical evidence that the
ions are indeed being made via back-bombardment is shown in
Figure 5.3, where an eroded LiF block is shown along with a

fresh unused block.

 

Figure 5.3. In this photograph are shown both new and used
LiF pellets. The sputtering of the used pellet is evident,
and the curvature of this erosion pattern agrees well with
the erosion pattern suggested by ion trajectory
computations, as will be shown in Section 5.“.

122

5.“ Analysis of the 7Li2+Ion Production Dynamics

7Li2+ at a final beam energy of 22 MeV/n

The case Of
has been selected for consideration here in detail. We want
to show the interaction of the initial ion trajectories with

the rf accelerating field.

In this case, the arc discharge in the ion source is
being maintained by a flow of nitrogen gas at a few cubic
centimeters per second. After the LiF has been sputtered
there will be lithium and flourine ions also present, but
the extracted beam current will be dominated by nitrogen
ions. We first have to decide, on the basis of source
output current and other factors, which nitrogen ions are
likely to contribute to the sputtering process. Figure 5.“
makes a comparison of the nitrogen charge state distribution
of three typical PIG sources. The data in Figure 5.“ were
obtained with do extraction from the sources at a fixed
voltage, and then separated on the basis of M/Q by magnetic
fields. Most Of the output current is going into 2+ and 3+
ions, with a fall Off by approximately 1/5 of those levels
for 1+ and “+ ions, and 5+ and 6+ ions by more than an order

of magnitude below that.

According to Figure 5.“, something like 80-90 % of the

extracted current from a nitrogen discharge will consist

123

 

 
 
 
 
  

'00 _ I l l l 1 1 =

Z OBENNETT- GAVIN I

: XCLARK- et. 01. ‘

AMSU HIGH FIELD-7‘

" HIGH VOLTAGE MODE

- DMsu HIGH FIELD-.—

Low VOLTAGE MODE

LU IC)E? f:

E — Z

I- — I

U) —’ ‘-

m -— —

L9 — —

O:

E 1 0

o — —I

U I :1

b _

g I I

g "" —1

-— -—1

0.1 l I I L I l

 

 

 

I 2 3 4 5 6
N CHARGE STATES

Figure 5.“. The charge state distribution for nitrogen ion
production from discharge-heated PIG sources is seen by
considering two sources from the literature [Cl 71, Be 73],
and the MSU 50 MeV cyclotron PIG source also in the cold
cathode (high voltage) pure Penning discharge mode [An 83].
As can be seen for the three sources operated in the arc
dischargermxkh ions Of charges 2 and 3 will appear in the
output with approximately the same intensity, with 1 and “
within an order of magitude less than that.

12“
of the first three charge states, and we will therefore only
consider in detail the trajectories of these ions.
Figure 5.5 and Figure 5.6 show the central region

trajectories of the N1'2’3+

ions for starting times near the
peak voltage on the first dee. These we obtained via the
computer program Supercyclone, developed at MSU to compute

accelerated orbits in the superconducting cyclotrons, and

used originally during the design of the K500 cyclotron

central region [Ma 81]. Figure 5.“ shows only N1+ions but
includes more of the central region details. For reference,
the area exhibited in Figure 5.5 corresponds to the boxed
region at the center of Figure 5.1, which showed the
positions of the ion source and dee tips. Starting times
earlier than those plotted in Figures 5.5 and 5.6 correspond
to orbits that reached the puller electrode before the
voltage changed sign-~these ions do not return to the
source. For those trajectories plotted, many do return to
the source slit, and it is likely that other orbits would
also, but only trajectories of ions starting from the center
of the source extraction slit were computed. These figures
show the ions tracing out a curved trajectory inside the
source, and as a result, preferential sputtering of the
right side CH1 the feed insert would be expected. This
agrees well with the erosion pattern exhibited in
Figure 5.3. The trajectories definitely show that the

position of the feed material insert, dictated by the

125
available space, is not Optimum. The right side of the
chimney, as viewed down from the top, is an external wall of
the ion source and although the feed insert would be better
placed here, there is, unfortunately, no space for that

purpose.

On the basis of the ion trajectory calculations, one
can make an estimate of the energy that the ions will
bombard the feed insert with. Figure 5.7 shows a plot of
the final energy versus starting time for the ions exhibited
in Figures 5.5 and 5.6. As can be seen, the width in time
of the returning ions is decreasing with increasing charge,
because the greater energy gain allows more higher charge
ions to reach the puller. The final energy however is

increasing with the charge.

In order to sputter a material appreciably, the mass
and energy of the bombarding species are both important.
Since nitrogen is a relatively light sputtering agent, the
energy must then be high. While there is no available data
on the energy dependence of LiF sputtering, a review of the
situation for other materials [Ca 68] shows that energies of
the order of a few keV for nitrogen should result in
sputtering yields of 1-2 for nitrogen ions on metals. (A
sputtering yield of 1 means that 1 sputtered atom is
obtained for each bmnbarding ion.) Our energies are well

above that.

126

//-

“\\\ .

FULLER ELECTRODE SOURCE-FULLER REGION

+
ENTRANCE CHANNEL 270 N' CENTRAL RAY TRAJECTORIES
N To-270-360'1n 10‘s1ops cow

7/ W \\
116’” I)» CRoss see...

1110’ ION SOURCE
I’, CHIMNEY

 

CYCLOTRON CENTER .2

1.:

21‘. ~ :,:~”':"1"::l"
.25". .g I

 

Figure 5.5. Orbit trajectories for N1+ions staring after
the peak voltage on the first dee are still in the gap when
the voltage changes sign and so are accelerated back to the
source. A sinusoidal time dependence is assumed for the
voltage on the first dee, so that the peak voltage is

occuring when wT=2709 Though these N1+ ions were not the

resonant species, the behavior that they exhibit here can be
used to sputter LiF feed material located in the chimney
back receptable, putting lithium ions in the are for the

production of the cyclotron resonant ion 7L12+.

127

SOURCE - FULLER

‘ N” CENTRAL RAY TRAJECTORIES
Toszeo-sso' In Io'nops ccw

 

 

2W)

 

Figure 5.6.

(A), and N3+ions in (B),
with some striking the feed material insert.

the source,

The trajectories of late starting N2+

0"}...a:

SOURCE- -PULLER

N3+ CENTRAL RAY TRAJECTORIES
1.0-270 360 in 10' stops CCW

O-..

ions, in

show that these too come back into

128

 

 

 

 

 

 

 

 

3+
35 _ 2’ .
4+

30 L. i
% 25 _ .
5
)-
<9
5
5 2° — .4
‘2
<0
a-o 1*
.4
EEIS p A
lo
P

10 L- +

5 ‘

F
0 1 1
280 300 320 3‘0 360

STRRTING TIHE (RF DEGREES)

Figure 5.7. For the ion trajectories shown in Figures 5.u-

5.5, and also Nu+, the final energy of nitrogen ions

returning to the source is plotted versus the ion starting
time. The truncation of the curves at the earlier starting
times was made on the basis of whether the ion came back to
the source. At 360° the dee voltage reverses sign and
positive ions are no longer pulled from the source.

129

5.5 Summary and Conclusions

 

At the time of this writing, there have been
approximately 200 hours of cyclotron beam time devoted to
lithium beams using the design that we have been discussing,
and so we can say that this development has certainly
worked. However the erosion pattern in the feed inserts
shows, and the orbit computations confirm that the design is
not optimal-”ideally the feed insert should be moved over to
intercept more of the ion trajectories. But in this ion
source there is no space to do this, except by replacing the
side wall of the chimney with some other material of the
same thickness. During the above-mentioned Operating
experience, we have found that the position of the
high-voltage rod behind the chimney is critical, and errors
in placement have resulted in arcing and some source

failures.

We have found that there is good qualitative agreement
with orbit computations, and these imply large return
energies of more than a keV, which is quite good and will

prove helpful when more difficult ions than L12+ are

attempted. In that regard, one future direction would be to

+
attempt higher charge states, like, for instance L13 , where

it might be found that some heavier gas than nitrogen is

required to obtain adequate intensityu Pushing<NIin the

130
direction of doing some metal ions, one should try first
perhaps the aluminum and magnesium beams originally

proposed as test beams.

CHAPTER 6

MEASUREMENT OF ION SOURCE OUTPUT VERSUS CATHODE SEPARATION

The previous chapters have been concerned with ion
source studies that have been important in the develOpment
of operating heavy-ion PIG sources for the K500 cyclotron.
One question that we asked when the K500 cyclotron was under
construction was: what will be different about the Operation
of an ion source in the high magnetic field ? Part of this
was answered through the measurement of dc extracted ion
currents in a test stand operated for a time in 1978 in the
K500 magnet at high field [An 83]. That study showed that
the charge state distribution was the same as for low field
PIG sources (see for example Figure 5.“). Factors that
would be expected to change at high field, like the electron
gyro~radius and the radial magnetic pressure on the plasma,

evidently are not fundamentally limiting.

The main purpose of the magnetic field in the PIC

source is to guide cathode emitted electrons through the

131

132
arc. In that sense field curvature is more important than
magnitude: if the magnetic field does not direct ions into
the arc chamber a plasma will not form. This is because
electrons tied to field lines that cross metal surfaces will
be stopped at the surface, and will not reach the arc
chamber to contribute to the ionization process. One more
observation can be made about the field shape-- a diverging
field passing through the arc would reduce the electron
density there, and with it the ionization rate, and so a
minimum requirement would be for a uniform magnethafdeld

between the cathodes.

It is here that there is some difference between K500
axial magnetic field, and those from low field room
temperature cyclotrons. The axial magnetic field of the
K500 cyclotron, coming mainly from the superconducting
coils, does not fall off as sharply along the axis at large
distances from the median plane. Figure 6.1 shows a
comparison of measured axial field of the ORIC cyclotron for

BO= 17.39 kG and the K500 cyclotron for BO= “3.1 kG. The

ratio Bz/B is quite comparable for small 2 values, but

0
drops to a low level for ORIC at large 2 values. Observe
that at 20 inches the K500 magnetic field is ten times
larger that in ORIC! In addition, the curvature of the K500

axial field is quite small. Figure 6.2 shows field lines

133

 

T I I r I I l l I T

E] ORIC 3870 Q

   

K500 662.5 n/sso n

 

B(z)/B(0)
'o)

 

 

 

 

3 20$3kc
.Li J
.2 4
J 1.91 k6
.0 1 1 1 1 1 L 1 1 1 1
0 L1 8 12 16 20
z (INCHES)

Figure 6.1. A comparison of a conventional cyclotron (ORIC),
and superconducting (K500) axial magnetic fields is made.
For ORIC 80: 17.39 kG and for the K500 BO= “3.1 kG.

13H

MSU-84-433

z finches)

 

 

 

 

 

 

 

 

2 4 6
r finches)

Figure 6.2. Magnetic field lines, obtained via a relaxation
calculation and shown for the region above the media plane
of the K500 cyclotron, for currents of 662.5 A/SSO A in the
small and large coils reapectively.

135
in the cyclotron center for the K500 field plotted in
Figure 6.1. These were obtained via the magnetic field
calculating program POISSON, using a relaxation technique
that includes both the coils and iron geometry [Ho 79].
Figure 6.2 shows that field uniformity at the center of the
cyclotron, in the 2 inch diameter hole where the ion source

resides is quite good.

It was conjectured that this large difference in the
axial magnetic field might make it possible to build a
longer ion source in the KSOO cyclotron. This would be
advantageous for several reasons. For one thing, at Dubna it
has been shown that neutral cathode vapor in the arc reduces
the percentage of high charge state ions from the feed gas
[Ma 76]. This cathode vapor is highest near the cathodes,
decreasing in the direction of the arc chamber center.
Moving the cathodes away from the center might therefore
result in a significant gain by reducing this charge
exchange loss mechanism. Secondly, the high magnetic field
has required that the central region of the KSOO cyclotron
be compact and this restricts possible changes that can be
made to tflua ion source. Moving the cathodes away from the
median plane would allow, for instance, larger cathodes, and
might also result in a more reliable ion source, through the
use of larger insulators, water cooling passages and voltage

holding surfaces.

136
Two fundamental processes controlling the production of
multiply-charged ions would be affected by increasing the
ion source length. One can show that the probability that
an electron, traversing the arc, will make an ionizing

collision, is approximately

p = 1 - e-nOAz (6.1)

where

n is the neutral gas density,
a is the total cross-section for the reaction, and

A2 is the separation between the cathodes.

The neutral gas density in the arc is about 1012 Atoms/cm3,

and ionization cross-sections are of the order of 10-16cm2,

which makes the collision probability for a single pass
through a 10 cm are of P=0.001. This would increase to
P=0.00S for a 50 cm arc. The change is evidently in the

right direction.

The second important physical process is the
confinement time of the positive ions, which must be large
for multiply-charged ions to appear in the output. In a PIG
source unfortunately the positive ions have no real
confinement, for in the axial direction the electric fields

at the ends of the arc chamber draw positive ions to the

137
cathodes (this large ion flux to the cathodes being the
important source of cathode heating), while in the radial
direction the highly collisional ions can walk across the
field lines, where they are lost at the arc chamber
surfaces. If the axial drift time sets the effective ion
lifetime, then increasing the length of the ion source
should result in a shift in the charge state distribution

toward higher charge values.

In order to study these effects, we built an
experimental source for the K500 cyclotron that had a
variable cathode separation, and measured the production of

U+' +
N and N5 ions as a function of cathode separation. In

this chapter the design Of this source, its Operation and

measured properties will be discussed.

6.1 Design of the Variable Arc Length Ion Source

The basic design of the variable arc length ion source
is shown in Figure 6.3. This device can best be pictured as
cutting the runwnal source design on planes through the arc
chamber end plates into three separate elements for the
purpose of allowing independent motion of the cathodes

relative to the anode.

138

MSU! £3448

 

UPPER C. P.

LPPER
SOURCE

 

 

  

___._. 1 _ _ MEDIAN
PLANE

\

ANODE (OFF LOWER CENTER PLUG)

LOWER
SOURCE

LOWER C. P.

Figure 6.3. The central region elements of the variable arc
length ion source are shown. See text section 6.1 for
details.

 

139
The anode, mounted off the lower center plug, maintains
the required central region shape of the ion source, as well
as holding and shielding the plasma from the rf field. A
slight median plane asymmetry had to be introduced into this
anode in order that it fit into the center plug vacuum lock.
The chimney mounted in this anode uses same type of square

tantalum tube as for normal sources.

The other two elements of the source, which we will
call the electron guns, house the cathodes and define the
arc. An electron gun actually is a normal ion source with a
shortened chamber housing only one cathode on a shortened
high voltage rod. The inside gap between the chamber
end-plate and the cathode must hold the arc voltage during
striking and operation. The aperture in the end-plate
allows cathode emitted electrons to pass out of this chamber
along magnetic field lines, in the direction of the anode.
Each electron gun has its own gas feed, so in principle each
electron gun would strike independently and behave

electrically like a normal arc discharge.

A substantial element in the preparation of this
experiment was the completion of the upper ion source and
the mechanisms that allow its insertion through a vacuum
lock and into the cyclotron. The upper ion source is a
mirror image of the lower source about the median plane.

One feature of its installation is quite different-~with the

1N0
lower source, we could check the central region alignment
with the cyclotron cap up. This was not possible for the
upper source, and a special fixture was constructed to set
its alignment. After commissioning this source, the
operators reported no difference in cyclotron tuning for

lower or upper source operation.

A key feature of variable arc length source design is
that the arc is defined via the intersection of the
apertures of the electron gun end plates and the anode.
This requires that the magnetic field curvature be small,
and that accurate alignment of these apertures into both
angle and radius be made. The former requirement is seen in
Figure 6.2 to be reasonably satisfied-~to the extent that
the apertures are large and field curvature suggested by
Figure 6.2 is not viewed as significant. The latter
requirement was achieved by constructing fixtures to set the
positions of the parts. The lower electron gun to anode
alignment, set with these fixtures, was checked with the
lower ion source installed in the center plug while it was
out for installation of the anode. The upper source,
installed in the upper center plug, could not be checked in
this way, but since its position was set with the same
fixtures as the lower source its alignment was reasonably

assured.

1A1

The operational plan was as follows. The sources would
be struck with the cathodes all the way in. In this
configuration there are only small differences between this
device and the normal source. The minimum cathode
separation is 3.5 inches. The cyclotron could then be tuned
and the beam current maximized on the main beam probe. The
sources would then be retracted, and the beam current
recorded as a function of cathode separation. The cathode
motion is obtained from the drives that are used to insert
and retract‘Uuaion sources from the cyclotron. At first
the drives were operated manually from the cyclotron vault,
and this meant turning off the cyclotron to move the
sources. Measurement was slowed by this method, and we
could see some effect on the vacuum coming from cycling the
rf system. For later runs hydraulic check cylinders were
added to allow provision for stopping the drives remotely.
These check cylinders were affixed with helical follow
potentiometers, that gave a voltage signal proportional t4)
displacement, so that the position of the cathodes could be

determined remotely as well.

The main goal was simply to measure the source length
dependence<n?the ion source extracted current. Since we
were interestxni in the development of multiply-charged ion

sources, any such ion that could be obtained normally with a

large current would be suitable, and we chose 1uNlHto be the

1“2
primary study species. Additionally, we wanted to study
whether there was any magnetic field dependence in this
source output, and this could be accomplished by the

acceleration of 1uNu+at more than one energy. Finally one

would like to know as well the charge state dependence of

observed effects, and so a measurement of 1”N5“, at the same

. “+
magnetic field as one of the N cases, was also planned.
The characteristics of the beams selected are summarized in

Table 6.1.

Table 6.1. The beams that were selected for the variable

arc length ion source experiment. The Nu+at 21.9 MeV/n and

N5+ at 35 MeV/n had the same magnet settings.

 

ION E FREQ

B

 

final 0 dee

(MeV/n) (MHz) (kG) (kV)
1uNu+ 21.9 15.19 3“.6 52.
11‘11'4+ 30.0 17.67 “0.3 70.5
1”115" 35.0 19.00 3“.6 65.3

 

14+ 5+
6.2 Measurement of N and N Currents versus

 

Cathode Separation

 

The main results of this measurement are shown

Figure 6.“ and Figure 6.5.

in

In Figure 6.“ we see the

1“3
dependence of the ion source extracted current on the change

5+for the first and third

in cathode separation for Nu+ and N
beams.U1Table 6fiL.After an initial rise in current, the
curves show a sharp drop in beam current for a small

displacement of 0.5 inches, to 70% of the peak value for

NHI, and about 50% for N5+.

The slope of the curve then
changes to a more gradual decline in current out to 6 inches
net change in in the position of a cathode separation (for a

total separation of 15.5 inches when the initial separation

of 3.5 inches is included). In Figure 6.5 the two Nu+cases
are shown, and these results demonstrate good agreement at

two different sets of cyclotron parameters.

Several investigations were made to study the nature of
the these curves. First, moving the cathodes "out" versus
"in" did not change the curve structure. Moving the
cathodes "fast" versus "slow" also had no effect. Leaving
the cathodes for some time at a fixed position, one could
adjust the ixn1 source and thereby change the beam current,
but after moving the cathodes the shape of the curve was
found to be the same. These data were so reproducable that
it became clear that the cathode motion did ruflziaffect the

ion source dynamics.

1uu

 

I I I I I I I I I I I

vnRIRBLE RRc LENGTH ‘
ION SOURCE

H
G

- Nitrogen 1eed q

00

4:

1'0

 

REL. ENTRRCTED CURRENT
O)

 

 

.0 2.€) 14.0 6.21
CHRNGE IN CRTHODE SEPQRQTION (in.)

I L

 

Q

Figure 6.“. The dependence of ion source extracted current
on an increase in cathode separation is shown for NL|+ and
N5+ions in the same magnetic field. The initial cathode

separation was 3.5 inches.

1“5

 

 

 

 

 

E NITROGEN 11+
£121. .1
D 22 MeV/n
”.8 - o 30 MeV/n -
o -
1“." 0
006 - 00 ..
Cl:
93
is.” "" c-
. 0
_,.2 _. 0 0 -
Lu
0:

.Q) 1 1 1 1 1 1 1 1 1 1 1

O 2.0 “.O 6.0

CHQNGE IN CQTHODE SEPQRQTION (in.)

Figure 6Ji. A comparison of Nu+production is made at two
different cyclotron energies, corresponding to different
magnetic fields and dee voltages (see Table 6.1).

1“6

6.3 Analysis and Conclusions

 

The electron gun sections of the variable arc length
source produce an electron beam whose shape and direction
are defined by the magnetic field. If the center of this
electron beam shifts relative to the center of the anode
apertures, then some of the electrons will spill onto the
anode end-plates, reducing the electron current in the arc
chamber. In Chapter 1 we showed that the rate of ion
formation is directly proportional to the electron current,
so that the loss of electrons above will reduce the ion
density in the plasma, and also the extracted current. We
will now propose an untested hypothesis based on these
ideas, as the best explanation for the results shown in

Figures 6.“ and 6.5.

The initial rise in beam current at small Az is well
correlated with the disengagement of an alignment collar in
the ion source drive mechanisms as the sources retract.
After disengagement a shift of .030 inches away from the
cyclotron axis is possible, and in this case does result in
a better arc alignment than obtained through the mechanical

positioning process.

The decline of the beam current after this maximum is
related to the effect that the actual curvature of the

magnetic field in the center has on the arc alignment. In

1“?
the introduction to this chapter we argued that the magnetic
field was very uniform-- and that therefore its possible
effect on the arc alignment would be small, certainly
smaller than the measured decline in the beam with cathode
separation. The question that we would like to answer is,
if it is larger than supposed, what would be the likely
effect? The cathodes, after an initial shift, are moving
axially, but at a fixed radius, into a region where the
radial component of the magnetic field off the axis
increases as 2 increases. This is shown in Figure 6.6, and
we note also that in this cyclotron the cathodes do not lie
on the axis, where the field would have no radial component.
At higher z the field lines intercepted by the cathodes will
not pass through the anode as they go through the median
plane. Some of the electrons emitted by the cathodes at
this elevated position will be stopped on the anode end-
plates, and the percentage stopped will grow as the cathode
separation grows, and would result in a reduction in the ion
current. Since this loss process is entirely geometrical,
it agrees with the characterics observed during the variable

are source operation.

During the last source run we changed the gun
apertures to try to influence this effect, but did not have

enough time to complete the measurements. After an extended
cyclotron shutdown, our goals had shifted and unforetunately

new measurements were not made.

1“8

MSU-84-457

CYCLOTRON
A X I S

'3' (EXAGGERATED
CURVATURE)

   
 
 
 

. MEDIAN
"" - "' ---PLANE

ANODE ""

    
  

  

“\\ \\\\\‘

CATHODE

 

Figure 6.6. Raising the cathodes axially in a curving
magnetic field causes a shift in the arc alignment rwalative
to a fixed anode, as in the variable arc length ion source.

CHAPTER 7

FINAL PERSPECTIVE

The K500 Cyclotron PIG ion sources now fully exploit
most of the techniques others have found useful for the
production of multiply-charged ions. Some of those
techniques were discussed in this dissertation. For
nitrogen discharges, hafnium cathodes have produced the
longest reported lifetimes for a PIG source operated with
ions heavier than hydrogen or helium. Pulsed operation of
the sources has given us a way to raise the intensity above

do operating levels for high charge state ions, and this has

been found essential for Ne5+ or Ar6+ ion production. In

addition, pulsed operation with hafnium cathodes has made

115“

a very successful beam. Solid sputter feed was
accomplished through minimal changes in the source design,
increasing the number of species that we are able to

accelerate in the cyclotron. The variable cathode

separation source showed that the ionization rate in the

1“9

150
region behind the source extraction slit is the most
important, and that the cathodes must not cast a shadow in
that region or the output of multiply-charged ions will
decline. We have already seen our source design and
operating procedures applied in a small heavy-ion source at

Tohoku University in Japan [Ya 8“].

Yet there is one outstanding problem which we have not
solved -- the extracted beam currents from the K500 are
smaller that those reported at other heavy-ion cyclotron
laboratories for the same type of internal ion source.
Figure'LJ shows a comparison of K500 cyclotron extracted
currents of carbon, nitrogen, oxygen and neon ions with
those from the cyclotrons at Texas A&M University [Sa 79].
Lawrence Berkeley Laboratory [Cl 81], and Oak Ridge National
Laboratory [Hu 78]. As can be seen, the outputs of these
three laboratories agree reasonably closely, and yet our
performance is systematically about two orders of magnitude
lower. The cyclotron extracted current depends on source
ion production, source extraction, transmission during
acceleration and cyclotron extraction efficiency, and
differences in these factors would be expected from
laboratory to laboratory. Yet the beam intensities obtained
in those laboratories mentioned above are quite similar and
differ by about 2 orders of magnitude from the beam

intensities obtained from the K500.

151

 

 

Y Y ' T

CORPORISON 0F CYCLOTRON PERFORHRICE FOR NITROG

I
E COIPRRISOI OF CYCLOTROI PEREORRQICE FOR CARBON

1
F\.
"Ht-“'T' "R

I I I?"
.6/
.‘7
‘ I
I

m. cuRRENI IpRI
.

 

(I
_|.J—UultL—_L—‘—J—‘-M—L_LJ-LLW_‘——J-—M
m. CURRENI 1,.m

 

 

 

E 1
: \
: \
g I
I .1
‘W .. |
.. ‘. -¢- mu no -
I. E ‘2 \‘ —._ L“ n ’ 1
: .2- | c...- °“ .- * m “a 4
r- 2}, mg. m. I. -11- LIL co
' 1". «un- one.
.. \ .--a- - mm: 09
1. ' _ I. '. A 1 1
3 . O S . O " I) 3 . O S . O 7 . 0
CHRRGE STRTE CHRRGE STRTE

 

 

 

 

 

 

convansou 0! CYCLOTIOI PERFORM! FDA 0:76:11 2 cmmxsou or cycgonou PEIFOAHANCE F0. um“
K .
--——-\ / x
e .- ‘.'.' ” \\ p
!.| . . ..-"” \ 1.| r-\‘
‘\ “ ’1 ' '. \ \\\ 1i \
‘- \ . a- \
‘i V ‘ “n . E. ~‘~'.""'- . \ A
U . \ w - \“‘ \ '.O “
E . \ E ‘ “ " ""\ ‘s
310' 5&‘10‘ 1:- ‘ e- ......... ,--"'~..,.._ x\ .
g \"0 g E “l‘ \\\
:' .—.‘ I \ ‘
J U u-
I \
Ia“ : 1.4: \
- --o— 3 =
r. - 3‘00?” 3 1; 4—“. no.
'4 L 1 >- _(" UL ~
-. ..-°“¢ _ -0”.. Ouc \
~°A~~ TAMU” 1. a ~rmu e.
19" 4 1 A a 10" 1
3.0 5.0 ' a 3.0 5.0
CHARGE STATE CHARGE STATE

Figure 7.1. The extracted currents of carbon, nitrogen,
oxygen and neon ions from heavy-ion cyclotrons at Texas A&M
University [Sa 79], Lawrence Berkeley Laboratory [Cl 81],
Oak Ridge National Laboratory [Hu 78] are compared to the
K500 cyclotron at MSU [An 8“]. Internal discharge heated
PIG sources supplied the ions in each case.

152

In examining the reasons for this discrepancy, the four
factors above need to be considered. The extraction
efficiency of the K500 cyclotron is not the origin of the
intensity problem since the measured internal beam before
extraction is already low. In any case the extraction
efficiency of the K500 cyclotron is quite comparable to
those at Texas A&M, Berkeley, or Oak Ridge. In fact the
problem is already apparent at a radius of 11 inches where
the internal beam is first measured. Since we are unable to
measure the current at a smaller radius, we can only

speculate on possible difficulties in the first 11 inches.

The ion source output could be low. We have looked at
the Operating characteristics relative to sources at other
laboratories, and find that the arc geometry and arc
parameters of our source are quite comparable. For stable
dc operation we require a gas flow of 1-2 sccm for arc
currents of 2-“ A and arc powers of 1-2 kW; the gas flow
decreases and the arc current increases as we go to pulsed
operation. Furthermore, the K50 cyclotron at MSU Operated
the precursor to our K500 ion source; it had a similar
geometry and the extracted beam currents gege comparable to
the above laboratories' experience, implying that the basic
design is sound. In summary, we do not see any large
deviation in source operation from that reported at other
laboratories. We will, however, continue to seek

improvements in source operation.

153

The extraction of ions from the source is another
important area. This extraction depends on the arc
alignment with the source extraction aperture, the size of
the aperture, and the magnitude of the extraction electric
field in the first gap. The extraction aperture and
electric fields do not significantly differ from those at
the above mentioned laboratories. With regard to the arc
alignment, we recently did a series of measurements with
tilted chimneys to alter the alignment, and found no

alternate alignment more preferred than the normal case.

The final area of uncertainty is the transmission of
the beam to 11 inches where it is first observed. We
considered the possibility of large losses due to orbit
errors in the cyclotron central region, but we see no
unexpected erosion there -- evidence for large currents
dumping on the metal surfaces in the cyclotron. Moreover,
recent foil burn studies showed reasonably good axial and
radial centering of the beam beyond the first turn. The
cyclotron pressure at the center and its effect on the beam
have not been explored. Large charge exchange losses of

beam would occur if the pressure there is poor.

We have a situation where all of the component systems
perform reasonably well, and where we have not found a large
error in either the ion source design or the beam

transmission process, but even so, the external comparisons

15“
in Figure'LJ nevertheless imply that we could do better.

We can only conclude that further investigation is required.

[As

[An

[An

[Ba

[Ba

[Be

[Be

[Be

[Bl

[CA

[Ca

79]

83]

8“]

72]

79]

69]

72]

73]

7“]

68]

70]

LIST OF REFERENCES

A

M.Agenda, K.Euler, B.Franczak, F.Hinterberger,
H.D.Rosendaal, and H.V.Schmidt. IEEE Trans. Nucl.
Sci.. N§;gg (1979) 2156.

T.A.Antaya, et.al. IEEE Trans. Nucl. Sci., NS-3O
(1983) 2176.

T.A.Antaya, et.al. Proceedings, 10th International

Conference on Cyclotrons and Their Applications,
East Lansing, Michigan (198“).

B

E.Baron, et.al. eIP Conf. Proceedings, No.9.
Cyclotrons-1972, AIP (1972) 2“3.

 

 

A.Backouche, L.Bex, G.Cardin, B.Piquet and
M.Voisin. 16th European Cyclotron Progress
Meeting, SIN, Villigen Switzerland (1979).

J.R.J.Bennett. Fifth International Cyclotron
Conference, Harwell, England (1969) “99.

J.R.J.Bennett. IEEE Trans. Nucl. Sci., N§:19
(1972) “8.

J.R.J.Bennett and Basil Gavin. Particle
Accelerators, 3 (1973) 85.

 

J.S.Blakemore. Solid State Physics, Philadelphia:
W.B. Saunders Co., (197“) 188-192.
C

G.Carter and J.S.Colligon. Ion Bombardment of
Solids, New York: American Elsevier Pub. CO.

 

T.A.Carlson, et.al. Oak Ridge National Laboratory
Report ORNL-“562, unpublished (1972).

155

[C1

[C1

[Ev

[Fu

[Ga

[Ga

[Go

[Go

[Ha

[Ha

[Ha

[Ho

[Hu

[Hu

71]

81]

82]

72]

72]

76]

53]

79]

67]

7“]

8“]

79]

73]

75]

156

D.J.Clark, et.al.. IEEE Trans. Nucl. Sci., Ngele
(1971) 102.

D.J.Clark, et.al. LBL Annual Report 1980-81,
unpublished.
E

R.D.Evans. The Atomic Nucleus, Malibar,Florida:
Krieger (1982).

 

F

G.Fuchs. IEEE Trans. Nucl. Sci., N§elg (1972) 160.

G

B.Gavin. 2nd International Conference on Ion
Sources, Vienna, Austria (1972) 512.

B.Gavin. IEEE Trans. Nucl. Sci., N§:2§ (1976)
1008.

J.D.Gow and J.S.Foster Jr., Rev. Sci. Instr., 23
(1953) 606.

R.A.Gough, D.J.Clark and L.R.Glasgow. IEEE Trans.
Nucl. Sci., NS-26 (1979) 216“.

H

C.A.Hampel. Rare Metals Handbook, New York:
Reinhold Pub. Co., (1967) 205.

 

Handboek of Chemistry and Phyeiee, Cleveland: CRC
Press (197“) B-16.

 

L.Harwood, private comm.

R.F.Holsinger. Poisson Group Programs, New England
Nuclear (1979).

E.D.Hudson, et.al. Physics Division Annual
Progress Report, ORNL-“937 (1973) “O3.

E.D.Hudson, R.S.Lord, M.L.Mallory, J.E.Mann,
J.A.Martin and W.R.Smith. IEEE Trans. Nucl. Sci.,
N§z§§ (1975) 15““.

[Hu

[It

[Jo

[Ke

[La

[Li

[Ma

[Ma

[Ma

[Mi

[Pa

[Fe

78]

82]

72]

72]

67]

80]

76]
8“]

81]

79]

72]

37]

157
E.D.Hudson, ORIC Beam List, Oak Ridge National
Laboratory (1978) unpublished.

I

T.Itahashi, et.al. RCNP Annual Report (1982) 167.

J
E.J.Jones. IEEE Trans. Nucl. Sci., N2:12 (1972)
101.

K

R.A.Kenefick, et.al. AIP Conf. Proceedinge, N329,
Cyclotrons-1972, AIP (1972) 259.

 

 

L

N.Laegreid and G.K.Wehner. Journ. Appl. Physics,
32 (1961) 365.

V.Littmark and J.F.Ziegler. Range Distribution for
Energetic Ions in all Elemenfie. Elmsford, NJ:

 

 

Pergamon Press (1980).

M

B.N.Makov, IEEE Trans. Nucl. Sci., fl§l3§ (1976)
1035.

M.L.Mallory, T.Antaya, F.Marti and P.Miller. Nucl.
Instr. and Meth., 222 (198“) “31.

F.Marti, M.M.Gordon, M.B.Chen, C.Salgado, T.Antaya
and E.Liukkonen. Ninth Int. Conf. on Cyclotrons
and Their Applications, Caen, Fr (1981) “65.

 

 

P.Miller, et.al. IEEE Trans. Nucl. Sci, N§:2e
(1979) 3716.

P

A.S.Pasyuk and Yu.D.Tretyakov. 2nd International
Conference on Ion Siurces, Vienna, Austria (1972)
512.

F.M.Penning. Physica, “ (1937) 71.

[Pi

[P1

[Ri

[R0

[R0

[Sa

[Sc

[80

[So

[Sh

[Va

[Wa
[We

[WI

61a]

61b]

8“]
62]

63]

79]

76a]

76b]

8“]
81]

69]

8“]
57]
78]

158

Yu.D.Pigarov and P.M.Morozov. Soviet Physics-
Technical Physics, 9 (1961) 336.

Yu.D.Pigarov and P.M.Morozov. Soviet Physics-
Technical Physics, e (1961) 3“2.

R
J.Riedel, private communication (198“).

D.Rosenberg and G.K.Wehner. Journ. Appl. Physics,
3; (1962) 18“2.

R.W.Roberts and T.A.Vanderslice. Ultrahigh High

Vacuum and its Applications, Englewood Cliffs,
N.J.: Prentice-Hall (19637 90.

 

 

S

Y.Sakurada, et. al.. IEEE Trans. Nucl. Sci., N§e2e
(1979) 2175.

H.Schulte, W.Jacoby, B.H.Wolf. IEEE Trans. Nucl.
Sci., NS-23 (1976) 10“2.

H.Schulte, B.H.Wolf, H.Winter. IEEE Trans. Nucl.
Sci., NS-23 (1976) 1053.

H.Schweickert, private comm.

T.Shinozuka, et.al. Ninth Int. Conf. On Cyclotrone
and their eeplicatieee, Caen (1981) 273.

 

V
P.I.Vasiliev, et.al.. Nuclear Instr. and Meth., 11
(1969) 201.

W
W.Waltersheid, private comm.
G.K.Wehner. Physical Review, 198 (1957) 35.
H.Winter. "Production of Multiply Charged Ions",

Lecture Notes in Physics, 82, Berlin:
Springer-Verlag (1978) 1.

159
Y

[Ya 8“] T.Yamada, et.al.. 5th Symposium on Accelerator
Science and Technology, Ibaraki, Japan (198“).

 

nICHIcRN STATE UNIV. LIBRARIES
mll”WINNIWIIWW1|WIN)"11111111111111"
31293107891420