1. 16th International Conference on Ion Sources
Recent Performance of & Plasma Outage Studies with the SNS H- Ion Source
Martin P. Stockli, B. Han, S.N. Murray, T.R. Pennisi, C. Piller, M. Santana, R.F Welton
Oak Ridge National Laboratory 
Oak Ridge, TN 37830, USA
16th International Conference on Ion Sources
New York, NY, USA
August 25, 2015

2. Abstract
Early in 2014, after several years of producing neutrons with ~1 MW proton beams, SNS started to ramp to higher power levels that can be sustained with high availability. Powers of up to 1.4 MW may be possible despite a compromised RFQ, which requires higher RF power than design levels to approach the nominal beam transmission. Unfortunately at higher power the RFQ often loses its thermal stability, a problem apparently enhanced by beam losses and high influxes of hydrogen. This led to the semi-retirement of the high-performing source #3. The apparently lower beam losses of the other two sources shifted the goal to delivering as much H- beam as possible with the least amount of hydrogen in the source, which led to plasma outages. Ongoing plasma outage studies show that the 13 MHz supply struggles with the ~90% power reflected by the 1-ms long 2-MHz plasma pulses. Possible mitigations are being tested, starting with a 4-ms RC filter for the reflected power signal. Lowering the H2 pressure initially increases the H- beam current due to reduced losses, and since mid-2014 ~50 mA are routinely injected into the RFQ. Subsequent LEBT retuning improves the RFQ transmission by better matching the reduced-divergence beams. Accordingly ~35 mA H- beams exiting the RFQ have become routine. To further support higher powers, under-performing sources are replaced after two weeks while well- performing sources are used for up to 8 weeks, frequently exceeding 3 Ah of H- without showing signs of aging. These new approaches increased the average RFQ output peak current at the end of the pulse by ~2 mA while the standard deviation was reduced from 1.9 to 1.3 mA compared to the prior year, which included the high performing source #3.

3. This talk is all about delivering
Content
Introduction to SNS
Brief Performance History
Recent Performance
How to make MORE beam
Beam Current Limits
The RFQ
The 13 MHz system
Lowering the H2
Plasma outages
A new tune for the 13 MHz!
The external antenna source
Conclusions
This talk is all about delivering
more H - ions!

4. Basic Energy Sciences has created 2 powerful neutron sources at ORNL
Neutron scattering pioneer Clifford Shull in 1946 at ORNL
High Flux Isotope Reactor (HFIR) Intense steady-state neutron flux and a high-brightness cold neutron source
Spallation Neutron Source (SNS) World’s most powerful accelerator-based neutron source 4

5. SNS produces the Science with 18 state-of-the art Instruments
SNS produces the Science with 18 state-of-the art Instruments. 2 more under development.

6. The SNS and HIFR Science Output grows!
We aim for 400 neutron science publications by 2017!

7. The Spallation Neutron Source
smashes a pulsed, 1 MW proton beam on to a Hg target to produce ~21017 neutrons 60 times per second!
ion source
linac
accumulator ring
Hg target
SNS was constructed by a collaboration of
Lawrence Berkeley National Laboratory
Los Alamos National Laboratory
Jefferson National Laboratory
Brookhaven National Laboratory
Argonne National Laboratory and
Oak Ridge National Laboratory

8. SNS Accelerator Complex
Collimators
Front-End:
Produce a 1-msec long, chopped, H- beam
1 GeV LINAC
Accumulator Ring: Compress 1 msec long pulse to 700 nsec
Injection
Extraction
<1 msec RF
Current
1ms
1000 MeV
RTBT
2.5 MeV
HEBT
945 ns
1 ms macropulse
Current
mini-pulse
Chopper system makes gaps
Front-End
LINAC
Liquid Hg Target
1 ms macropulse
1 ms
The front end produces a 60 Hz, ~1 ms long chopped ~40 mA H− beam
The LINAC accelerates it to ~ 1 GeV
The ring accumulates it to ~40 A H− beam

9. The Spallation Neutron Source
1.4 MW
1.34 MW
~850 kW
target failures
lack of spare target
cost saving ~850 kW
For the past 6 years SNS has been running near 1 MW except for cost- and target-issues! We aim for reliable 1.4 MW by 2017!
Much had to be learned to support ~1 MW operations with high availability!

10. Ion Source and LEBT Performance History
Only source #3 makes ~38 mA
Decaying beams before developing proper conditioning
Ramping up the beam with many improvements
Year of ~40 mA MEBT currents
source leaks & contamination
Reduced RFQ transmission
large leak
e-dump fails
The 131th production source was started up early this August. We had 131 opportunities to learn, improve and perfect the operation of our H- ion source and drive up its performance to unprecedented levels.
However, due to the nature of ion sources, especially H- ion sources, many things are not fully understood and not exactly reproducible, and remain challenges!

11. Recent Ion Source and LEBT Performance
1.4 MW
source leak suspected
#2 & #4 preferred for more stable RFQ
e-dump failure
e-dump failure
#3 preferred due to higher performance
12 month Period
7/13-6/14
7/14-6/15
change
Average RFQ input
45.4±2.4
51.8±1.2
+14%
Average RFQ output (end of cycle)
33.2±1.9
35.0±1.3
+5%
   RFQ input currents
 source #2 RFQ output
 source #3 RFQ output
 source #4 RFQ output ?
Source #3 was the best-performing workhorse until March 2014, when it became disliked due to its thermally loading the RFQ and limiting its power.
Source #2 was benched in 2011 after a severe contamination. After being tuned up in 2014, it became the favored workhorse.
Source #4 serves as the alternate for the last 4 years. While its LEBT output may lack a few mAs, in the MEBT it is only ~1mA less than #2.
Despite benching star source #3, the average currents are up and the variations are down!

12. How to Make More H- Beam with lesser Sources?
38 sccm
plasma grows
High-voltage upsets can cause plasma outages which are avoided by increasing the H2.
After all high voltages have conditioned, the H2 flow can be lowered. This increases the H- beam as extraction-area losses are reduced.
28 sccm
plasma steady
After reaching the optimum H- beam current near ~28 sccm, the beam current decreases as the plasma is starved.
18 sccm
plasma starves
The MEBT H- beam current is globally optimized during the final tuning using the
time-averaged charge, integrated over the full pulse-length. This is done during the power ramp-up to production.
This includes a LEBT retune to benefit from reduced-divergence beams obtained with smaller pressures.
Short Answer: Very carefully!

13. LEBT and MEBT Beam Current Limits60 54 48 42 36 30
RFQ Input Current
Transmission
RFQ Input & Output Current =
RFQ Output
Current
RFQ Output Current
Increasing the LEBT beam current increases its emittance, which lowers the RFQ transmission. The trend of the transmissions suggest that we are close to the maximum RFQ output current.
Are we doomed to MEBT currents ≤36 mA?
Maybe not, if old dogs can learn new tricks!

14. Andrei Shishlo pointed out one way!
Courtesy of A. Shishlo
Increasing the RFQ power can drastically increase the MEBT beam!
However, to remain thermally stable, less H2 influx is needed!

15. The SNS Baseline Ion Source and LEBT
LBNL developed the SNS H- ion source, a cesium-enhanced, RF driven, multicusp ion source.
Typically 300 W from a 600-W, 13-MHz supply generates a continuous low-power plasma.
It is tuned by minimizing the reflected power.
The high current beam pulses are generated by superimposing kW from a pulsed 80-kW, 2-MHz amplifier. It is tuned for maximum H- beam current.
2 MHz
There remain issues, but this injector keeps breaking records!
13 MHz
But the H2 needs to be lowered, which caused plasma outages that needed to be understood! CS CP

16. Ramping the 2 MHz with 1.96 MHz
Building up the LRC oscillations
13 MHz
reflected amplitude
forward amplitude
RF build-up was studied with 4-week-old source #2 using a 2 MHz directional coupler on ground.
With 13 MHz plasma the LCR circuit rapidly builds up oscillations before drifting off resonance due the evolving plasma inductance.
Only little RF power is reflected because the plasma absorbs the RF readily.
However, without plasma the LCR circuit builds up to 76 kW, absorbing up to 71 kW. The electric fields generated by the very large antenna current are unable to break down the pure H2 gas.
The 80 kW QEI cannot provide reliable ignition!

17. How to reliably ignite the Pulsed SNS Source?
Unable to reliably ignite a clean H2 source with the 2 MHz RF, the only option is to:
Ignite continuous 13 MHz plasma with a pressure bump (PUFF).
Use the 13 MHz plasma to absorb the 2 MHz plasma.
After a plasma outage go back to 1).
It is that simple!
And we do for >7 years!
However, until July 2007 we operated without 13 MHz because the 2 MHz self-ignited the plasma of the poorly-conditioned, dirty sources which delivered decaying beams.
Persistent operation of old sources depends on continuously maintaining some plasma!

18. Plasma outages start at the end of a 2 MHz pulse!
A Plasma Outage
2 MHz detuned
Seen by the 2 MHz directional coupler on ground
13 MHz
More 13 MHz
2 MHz forward
No plasma
Plasma
2 MHz reflected
Plasma outages start at the end of a 2 MHz pulse!

19. Time Resolved 13 MHz Measurements
In the past the average reflected power shown by the 13 MHz supply was used!
In 2014 FE scope#1 was used to show forward & reflected 13 MHz RF.
Our tunes turned out to be inconsistent.
300W
400W
450W
The biggest surprise was that the 2 MHz plasma reflects practically all 13 MHz RF, which triggered foldbacks of the output power!
COMDEL said time averaging the reflected power is difficult!

20. The CX600
 from 0.12 to 4 ms
R19 from 1.2k to 3.9k
C73 from 0.1 to 1
The CX600 was modified to average the reflected power over 4 ms!

21. The 4-ms Filter works! No more foldbacks of the output power!
300 W; 25 sccm
400 W; 25 sccm
450 W; 25 sccm ms ms ms
No more foldbacks of the output power!
300 W; 25 sccm
300 W; 20 sccm
300 W; 19.3 sccm ms
Lower Hydrogen pressures are possible.
Plasma outages are preceded by a reflected power bulb after the 13 MHz reflected power recovers!

22. Tuning the 13 MHz Matcher
In the past the 13 MHz matcher was tuned with plastic slot drivers until the reflected power from the COMDEL CX600 indicated a minimum.
Adding 2 dials this spring enabled reproducible tunes!
Old matcher
+2 dials
= reproducible tunes!
2 MHz
13 MHz

23. An outage resistant 13 MHz tune
The two 13-MHz tuning capacitors are strongly counter-correlated!
The resonance is less than 1 turn wide!
Measuring the outage flow has shown that tunes with higher reflected power and less bright plasma allow for lower hydrogen pressure.
That has enabled neutron production with ~10% less hydrogen which increased the beam current by 5% due to a higher transmission through the RFQ!

24. MonPS35
Baoxi will be around for the rest off ICIS!

25. Ion Source & LEBT Performance
For the last 4 years the ion source and LEBT were ~99.5% available.

26. Source Replacements
Since late 2011, 6 and 6+ service weeks are customary for high performing sources to deliver more H- ions when desired.
33 production sources were replaced in the past 2.6 years.
30 at the end of their service cycle.
2 prematurely after e-dump failed.
1 prematurely after antenna failed.
This is infant mortality!
The diameter of the most plasma exposed “tapered bend” of the antenna was measured before and after service.
Since 2014 the average measured wear was 40 µm or ~10% of the coating.
There is no correlation between the measured wear and the service duration: r = 0.12.
Plasma light emission data suggest that most wear happens during conditioning!
Service cycle is increased to 10 weeks for 5 A· h of H-!

27. SNS External antenna source
See Rob Welton’s Poster TuePE36
RF-driven (pulsed 2 MHz, kW), Cs- enhanced, multicusp H- sources capable of delivering ~50 mA.
Similar to baseline source except for a water cooled AlN plasma chamber, antenna external to the vacuum and a plasma gun for ignition.
Few issues remain after a 40-day run with stable beam and e-dump after a single cesiation!

28. Summary and Conclusions
We have operated for over one year with source #2 and #4 which yield a lower thermal load to the RFQ.
Paying more attention and spending more time tuning the source and LEBT have increased the H- beam current.
However, the compromised RFQ transmission reduces this gain to 1-2 mA in the MEBT.
Plasma outages appear to occur when the 2 MHz plasma decays before enough of the 13 MHz can be absorbed to restore the 13 MHz plasma.
Breakthroughs allow for lower H2 pressures with higher H- beam currents without frequent plasma outages.
Matching the 13 MHz with more reflected power and less light has enabled operation with lower hydrogen pressures, which decreases the beam emittance and increases the transmission through the RFQ.
Thank you for your attention!