Robert F Welton

1
Recent Advances in the Performance and
Understanding of the SNS Ion Source

• Robert F Welton
• SNS Accelerator Physics and Technology Forum
• Dec 9, 2004

2
Introduction

• Source has been highly developed for low
  duty-factor operation (50-200 us, 1-20 Hz or duty-factor) at LBNL and works well in that role
  provided ORNL antennas are used!
• This work First experiments to determine if the
  LBNL source be operated at 6 beam duty-factor
  reliably for sustained periods, meeting the
  long-term operational requirement of the SNS by
  2008.
• Beam current 40 mA
• Pulse length 1 ms
• Repetition rate 60 Hz
• Run period 21 days with a very high reliability!
• If the source cannot meet the requirement,
  determine if the goal can be met by implementing
  relatively simple design modifications to the
  existing source.
• Once the operational goal is met, begin
  development which will support facility upgrade
  plans to 3 MW operation.
• Beam current 60 mA
• Pulse length 1.3 ms
• Repetition rate 60 Hz
• Run period 21 days with a very high reliability!

3
Outline

• The Ion Source, LEBT and Test Stand
• High Duty-Factor Ion Source Tests
• Data Analysis Cs Release and Transport
• A Direct-Transfer Cs Collar
• Outlook

4
Part I The RF-Multicusp Ion Source, LEBT and
Test Stand
Dumping magnets
Plasma
Multicusp magnets
Dumping electrode
8 Cs2CrO4 dispensers loaded into the collar
H- beam
RF antenna
Gas inlet
Cs collar
Filter magnets
5
Detailed View of Source
6
Low Energy Beam Transport (LEBT)
Lens 2
Ground
BCM
Extractor
E-dump
Lens 1
Ground
Outlet Aperture
7
Ion Source Test Stand at the SNS
Lockable fence
Main operating console EPICS controls most
Parameters are Archived!
Ion source cage
LEBT tank and vacuum pumps
Emittance scanner and Faraday cups
Residual gas spectrometer
2 MHz matching network
Optical spectrometer for plasma diagnostics
Communication racks
Enclosed high voltage platform
AC distribution
8
Operational Procedure for Runs 1-7(Based on LBNL
Guidance)
Part II Experimental Data at full duty-factor
1.2 ms H- pulses at 60 Hz!

• Each source was operated in a similar fashion
  thoroughly cleaned, inspected, mounted and leak
  checked.
• It is operated at low duty-factor (several hours while increasing collar temperature
  to 300 C.
• Source is then ramped to 7 duty-factor.
• The first cesiation is preformed collar to
  500-550C for 1/2 hour after which the collar is
  cooled to 300C and RF power is ramped to the
  operating point of 40-60 kW.
• Each run was cesiated as necessary to maintain
  the highest beam current.
• The run was terminated when the beam current fell
  below 20 mA.

9
Source condition Run 5, 6, 7-conductive on ant.
coating no punctures
Run 5, 6 and 7 History
Mounted source 2, open collar
Mounted source 3, enclosed collar
Mounted source 3, 2x Cs
Begin 7 operation
Begin 7 operation
Begin 7 operation
51 kW
Cs
Trip
Cs
Cs
50 kW
Cs
Cs
Trip
Cs
50 kW
Cs
40 kW
Cs
10
Antenna Position Studies
D12 mm
Moving the antenna to a forward position seemed
to reduce the average power efficiency from 0.57
to 0.51 mA / kW but allowed stable operation at
10 kW higher RF power resulting in 17 more beam
11
Opening the Cs collar outlet aperture
Opening the outlet area of the cesium collar
required much higher H2 flow (40 SCCM) for stable
operation. The extracted beam current was
essentially unchanged from the closed collar
configurations the beam attenuation-rate was made
considerably worse 9 mA / day versus an
average of 5 mA / day
Nominal collar
Open collar
12
Increasing the number of Cs dispensers
In run 6 we increased the number of cesium
dispensers in the source from 8 to 14. We
observed a lower peak H- yield which decayed less
rapidly than the other runs resulting in
comparable average H- current over the course of
the run 2 mA/day versus an average of 5 mA/day
13
Run Summary (6-7 duty-factor)
14
Does This Meet the SNS Requirement?

• Our best run so far employed twice the quantity
  of Cs yielding an average current of 27 mA with
  mean beam attenuation rate of 2 mA/day.
• This falls short of the SNS requirement of 40 mA
  for 21 days of continuous operation.
• In the next sections we carefully analyze the
  data to develop procedural and design
  modifications with will bring us closer to our
  goal the initial focus will be reduce beam
  attenuation since we have demonstrated we can
  very briefly meet the SNS current requirement of
  38 mA.

15
Part III Data Analysis First a little
background in H- production
How is H- produced in the source?
Cs
H, H
Surface Production ( Cs fast neutrals and
ions required)
H-
H-
e-
Volume Production (cold electrons
vibrationally excited H2 required)
H
H2(n0)
16
Data Analysis
Why Does the Beam attenuate rapidly (5 mA/day)?

• In the case of the SNS source, the amount of Cs
  introduced is insufficient to explain enhanced
  volume production if the number density of Cs in
  the volume was sufficient to significantly
  participate in volume production Cs would be
  pumped from the source 100-1000x faster than the
  observed enhancement. Therefore we consider Cs
  enhancement to be mainly a surface effect.
• The SNS source in an uncesiated state typically
  produces 10 mA of H- current. After adding Cs,
  the source usually produces 30-40 mA suggesting
  20-30 mA result from purely surface ionization.
• Volume production should track with plasma
  density which is monitored by a fiber optic
  spectrometer (300-1400 nm) hard mounted to the
  source plasma chamber.
• Since we observe essentially no decay in plasma
  density over the course of a run, provided that
  RF power is held constant, we conclude that much
  of the beam decay can be attributed to a decrease
  in surface production.

17
Why the Decrease in Surface Production?

• Efficient surface ionization requires
• An intense flux of ions and/or fast neutrals
  arriving at the surface from the plasma core
  (Kishinevskii probability increases rapidly with
  bombardment energy).
• A partial monolayer of Cs coating the ionization
  surface creating a low work-function.
• Efficient extraction of the H- ions once they are
  produced.

• Density measurements strongly suggest that the
  plasma flux arriving at the surfaces near the
  extraction region is essentially constant over
  the course of a run. These data also suggest
  that since plasma conditions are not changing
  extraction efficiency should not significantly
  change over the run.
• Therefore, the most likely cause of overall beam
  attenuation over the course of a run is failure
  to maintain low work-function ionization
  surfaces.

18
Cs Required!

• In order to maintain a low-work function
  ionization surface
• Cs must be released from the dispensers into the
  vacuum
• Cs must also be transported through the vacuum to
  the ionization surface via two possible pathways
  
• Directly from the Cs dispenser
• Indirectly through condensation on the source
  interior
• We now carefully consider each of these
  processes

19
(i) Cs Release from the Dispensers
Each source contains eight 12 mm Cs cartridges
(5.2mg of Cs per cartridge) from SAES Getters
containing a mixture of Cs2CrO4, Al and Zr
(St101). Analysis of this mixture using the
chemical thermodynamic equilibrium computer code
(HSC) shows the reactions below are dominant
4 Cs2CrO4 5 ZrY8 Cs (g) 5 ZrO22 Cr2O3
6 Cs2CrO4 10 AlY12 Cs (g) 5 Al2O33 Cr2O3
20
Thermodynamics of Cs Release
The equilibrium composition of Cs2CrO4(17), Al
(70), Zr (13) and gaseous H2 (1 mTorr)
P1 mTorr
This analysis shows that Cs is released in
elemental form and is thermodynamically probable
at temperatures above 100 C.
21
Thermodynamics of Cs Release
In the absence of the getter material st-101 the
Cs2CrO4 will not release Cs until nearly 1000C!
Thus we should be concerned about poisoning of
the st-101 material by reactive gases including
H2!
22
Gases Released While Ramping Source to High
Duty-Factor (RGA Measurements)
Pressure measured in LEBT chamber the ion source
is 100x greater
10-6
Partial Pressure (Torr)
H2O
CO2
O2
N2
0
Time (hours)
1.2 ms / 10 ? 20 Hz
1.2 ms / 20 ? 30 Hz
1.2 ms / 40 ? 50 Hz
1.2 ms / 50 ? 60 Hz
23
Overall Release Process

• ST-101 is a highly efficient low-temperature
  getter for reactive gases such as O2, CO, CO2,
  H2O and H2
• Product literature states that st-101 begins to
  sorb these gases beginning at 250 C and readily
  sorbed at 300 C.
• As opposed to all other reactive gases, H2 does
  not chemically react with the getter material but
  instead forms a solid solution which has the
  characteristic of reversibility.
• Thus care should be taken to outgas the source
  thoroughly at full duty-factor and power-level,
  typically for several hours, while maintaining
  collar temperatures below 250C.
• In run 8 and 9 this procedure was followed which
  resulted the highest average beam current as well
  as the lowest decay rate even for modest
  cesiation suggesting much more Cs was delivered
  to the source!

24
Run 8 History
Mounted source 2, shifted cold collar D2.3 mm

• Average beam current 30 mA
• Beam attenuation rate 0.7 mA / day

3.6 duty factor Tc136 C, beam off
Trip
Trip
Trip
Tc 71 C
Cs Tc420 C Begin 7
Trip
Tc 150 C
51 kW
Tc 71 C
Tc60 C
25
Run 9 History

• Average beam current 33 mA!
• Beam attenuation rate 0.4 mA/day!
• We still have not yet met the SNS requirement!

Begin 7 operation
40 kW
Water leak in antenna
26
(ii) Cs Transport Pathways
Now having understood Cs release from the
dispensers we must now understand Cs transport
through the vacuum to the ionization surface in
order design an optimal system.
Two possibilities
But first we must answer these basic questions

• Where is the principle ionization surface?
• How much Cs does the collar require?
• How much Cs is provided by the dispensers?

27
Where is The Principle Ionization Surface?

• H- transport simulations suggest H- ions can only
  transit the plasma for a 1cm before getting
  destroyed collisions with plasma particles.
  Therefore we only look at regions of the source
  near the outlet aperture. This leaves only the
  interior surface of the outlet aperture and the
  Cs collar.
• During cesiations the collar temperature is
  raised to 550C and we nominally see no evidence
  of Cs-enhancement and a large reduction in beam
  intensity.
• It is unlikely that raising the temperature of
  collar effects the outlet aperture temperature.
• During cesiations the outlet aperture is likely
  coated with Cs yet we see no enhancement.
• Therefore, the Cs collar is most likely the
  primary ionization surface.

28
How Much Cs Does the Collar Require?

• In order to maintain a constant fractional
  monolayer of Cs, the rates of supply and loss
  must be equal. Loss processes are thermal
  desorption and kinetic ejection (sputtering) by
  hot plasma particles.

Kinetic Ejection

• The Cs-Me bond strength can be estimated by
  various methods which converge at U 2 eV for ½
  monolayer coverage.
• H current densities extracted from a nearly
  identical multicusp source operating at 50 kW was
  1.5 A/cm2 or 1019 ions/cm2 s striking the walls.
• Energy spreads of protons extracted from similar
  sources (scaled to RF powers of 50 kW) have been
  found to be Maxwellian with T 20 eV.
• Folding this particle distribution into the
  Bohdansky formula for low energy sputtering using
  a surface bond energy of U 2 eV yields a loss
  rate of 7 x 1013 Cs/ cm2 s.
• At this ejection rate an optimally coated surface
  will be cleaned in 2.5 s.

Thermal Desorption

• The thermal evaporation rate at nominal operating
  temperature (300C) is governed by the Frenkel
  equation and was found to be 1x109 Cs/ cm2 s.

29
How Much Cs is Provided?
Several experiments were conducted where
dispensers were heated and the released Cs
monitored using atomic absorption spectroscopy
(852.1 nm). It was found that 1ug/min
(detection limit) of Cs was released in the
480-560 C range depending on the particular
dispenser sample. This corresponds to a flow of
7.5x1013 atoms/s
At 300 C a maximum of 1x1010 Cs/s
30
Indirect Cs Transfer Dominates
Direct Transfer
Cs dispenser
Ionization Surface

• At the Cs dispensers nominal operating
  temperature of 300 C a maximum of 1x1010 Cs/s
  are released from the dispensers. This is an
  insufficient flow to maintain 1 cm2 optimally
  covered given a loss rate of 7x1013 Cs/s.
• To rule-out the possibility of direct Cs transfer
  playing an important role, the dispensers were
  held at T70 C for 8 days while observing steady
  Cs enhanced beams of 30 mA. At 70 C 2 atoms/s
  are released!
• We therefore conclude that indirect Cs transfer
  dominates.

31
Indirect Cs Transfer
20 C
20 C

• During a cesiation, Cs entering the source will
  condense in areas of the interior of the source
  that are cool enough so the condensation rate
  exceeds the evaporation rate as determined by Cs
  vapor pressure curves.

50 C
300 C
1.5x1014 Cs/s
5x1013 Cs/s
Cs escaping through extraction opening will not
be returned.
100 C
550 C

• Cs condensing on plasma facing surfaces will be
  sputter ejected rapidly and will seek, shady
  regions.
• Most evidence of accumulated Cs was found behind
  the heat shield of the outlet aperture
• 2.7x1017 Cs atoms are delivered into source
  during nominal cesiation.

300 C
50 C
20 C
20 C
32
Indirect Cs Transfer

• After cesiation the source will evaporate Cs from
  the walls at a rate determined by surface
  temperature and the vapor pressure curve of Cs.
• The dose of 2.7x1017 Cs atoms is only
  sufficient to feed an ionization surface of unit
  area at optimal coverage for 1 h. Observed Cs
  enhancement typically lasts much longer than this
  suggesting the nominal Cs flow is less than
  optimal.

Cs vapor pressure curves can be converted to
evaporations rates by using the equations of
kinetic theory
33
Is This Physical Picture Consistent with the
Gross Features of the Data?

• Post Cesiation Decay- Once a finite amount of Cs
  has been transferred from the dispensers to the
  source it evaporates at the rates G proportional
  to the number of condensed Cs atoms (see the
  equation). The observed exponential decay is
  characteristic of the resulting reduction of Cs
  flux to the ionization surface. Since several
  regions of the source contributes, each at
  different temperatures, several component
  exponentials are typically observed.

• Plasma Trip Beam Enhancement- During a plasma
  trip the source cools since plasma heating ceases
  but the Cs collar remains hot (externally heated)
  and continues to deliver Cs to the source. Once
  operation resumes, the finite amount of condensed
  Cs depletes in the same manner described above
  yielding exponential decay curves.

34
The Problem with Indirect Cs Transfer

• During source operation the flow-rate of Cs
  evaporated from the source walls is essentially
  uncontrollable dependent on the temperature
  distribution of the source and its construction
  details.
• Because the flow rate is uncontrollable it cannot
  be matched to the Cs loss rate from the
  ionization surface to insure optimal coverage.
• Cs transfer is also inefficient, most of the Cs
  will condense in regions which are too cold to
  re-evaporate at significant rates.
• A new collar is proposed in which Cs is directly
  transferred from the dispensers to the ionization
  surface. Both the Cs dispensers and ionization
  surface can be independently controlled over a
  wide range of temperatures.

35
Part IV A Direct-Transfer Cs Collar
Air duct

• A separate ionization surface is inserted into
  the existing collar. The existing collar now
  becomes purely a Cs dispenser while the insert
  serves as the ionization surface.
• Separating these components allows their
  functions to be optimized in terms of geometry
  and temperature. The existing collar remains in
  the original location and the new component will
  be supported by a Ta strap orientated transverse
  to the beam direction.

SS air line
Al
304 Stainless Steel (50 transparent)
Ta holder

• The use of stainless steel preserves the existing
  ionization surface material, others will be
  tested in the future.

• Al is needed for heat transfer
• Distance to outlet adjustable.

36
Thermal Analysis
Air cooled surface
Heat load 65kW, 1.21 ms, 60 Hz Heat load into
collar 136,400 W/m2 Power density on
surface 17,948 W/m2
Air temp20C
Air temp500C
Max temp 30 C
Max temp 511 C
37
Cs Transfer to Ionization Surface
70 of Cs will directly enter the collar
Air duct
Cs Disperser
Thus a Cs dispenser temperature (2x) of 450 C is
required to maintain ½ a monolayer on a 1 cm2
collar under nominal conditions.
38
Initial Test of the Inclined Geometry
Insulating ring
Collar
Outlet Aperture
Standard Configuration
New Configuration

• In order to increase source efficiency we have
  moved the effective ionization surface closer to
  the outlet aperture and inclined it 16 degrees to
  better intercept particles streaming from the
  plasma core.
• We also allowed the collar to be biased to
  enhance H bombardment energy

39
60 mA Achieved!
60 mA pulse averaged 70 mA peak current
10 mm
Typical values from both configurations
40
Emittance

• The beam envelope seems to contain two particle
  distributions
• The usual bright core which comes as a result of
  volume production and charge exchanged surface
  produced ions.
• The wide shoulders could emerge from direct
  surface produced species entering the beam
  without charge exchange with the cooler neutral
  species.
• 5 mA of electrons have been observed in the beam
  while running a He plasma

10 mm
41
Bias Dependence No Effect
10 mm
RF power 30 kW
Trip to DESY planned
42
Outlook

• The ion source when operated at low-duty factor
  (100-200us, 1-20 Hz, continuously for many weeks with out significant
  problems provided ORNL antennas are used. This
  has served the SNS commissioning needs well!
• A new source conditioning procedure was developed
  based on detailed thermochemical analysis of the
  Cs dispenser. At 7 duty-factor this improved
  the beam attenuation rate from 5 mA /day (average
  of runs 1-5) to 0.5 mA/day by injecting more Cs
  into the source. Beams of 30-35 mA have now been
  sustained for 15 days.
• An initial test of an inclined geometry
  ionization collar has resulted in measured beam
  currents of 60 mA average and 70 mA peak H-
  current with much larger emittance.
• The Direct-Transfer Cs collar has been designed
  which is, for the first time, capable of
  supplying Cs at a rate which precisely matches
  the Cs loss rate from the ionization surface
  insuring optimal coverage.