RF Breakdown Workshop

1
RF Breakdown Workshop

• (held at) SLAC
• 28-30 August 2000
• Purpose
• Review understanding / experiments
• Define common terms
• Explore paths for future work
• Timely - power generation and delivery,
  structure completion CLIC / NLCTA

Marc Ross
2
Agenda

• Review of structure testing at CERN, ASTA, NLCTA,
  Haimson Research, Klystron lab test stand
• Material preparation / fabrication procedures
• Superconducting RF Nb cavity fab procedures and
  test results
• Theoretical models of breakdown
• RF structure design
• Discussion of structure fabrication, diagnostics,
  processing strategy
• Attendees
• CERN, Cornell, DESY, LANL (20)
• NLC collaboration labs
• invitations directed primarily to LINAC 2000
  attendees

Open minded brainstorm
3
Questions -

• What does it take to produce a breakdown event?
• What role does field emission play?
• What role does the surface play?
• features
• gas (from bulk or ambient)
• contaminants
• crystalline structure
• What is the underlying frequency dependence?
• (beyond pulse length)

4
After breakdown -

• What causes damage?
• How is the microwave circuit (Z, vg) involved?
• Can there be more than one arc in a breakdown
  event?
• Does the damage depend on surface details?
• How does it depend on the processing strategy?
• What is the path to higher gradient?
• material processing
• high power processing strategy
• impedance/group velocity

5
CTF-II (CERN)

• 30GHz two beam
• Onset of breakdown 290MeV/m acc field
• Breakdown every pulse 320MeV/m
• Breakdown right away 480MeV/m
• Strong backwards dark current
• Very short pulses

6
NLCTA / ASTA

• Observe upstream cell damage and large detuning
  changes in five structures 50 MV/m.
• Difference in upstream and downstream cell damage
  very pronounced.
• Phase errors grow with time, even after
  processing to a higher gradient and backing off.

7
NLCTA Processing

• DDS1 will not process above 73 MV/m.
• Estimate removal of 5 mm of copper per cell at
  the upstream end of DDS1.

8
Time series of DDS1 pulses

9
Pulse length and breakdown location
Time

• Pulse length variable
• Breakdown location toward upstream end
• Events within a multi-breakdown sequence move
  upstream

Location
10
ASTA processing of M2 backward

• High gradients can be achieved without cell
  damage attained gt 100 MV/m in about 20 cells
• No apparent damage
• Will be repeated at NLCTA

11
Superconducting cavity diagnostics and results
(Cornell)

• Identification of field emission location
• follow to breakdown
• SEM search / use of Auger surface analysis vital
• connection of each initial emission point to
  surface contamination
• generation of groups of breakdown sites from an
  initial contaminant

Multi-event site only center has contaminant
12
Niobium / Copper breakdown SEM images
13
Transition from field emission to breakdown site
(SRF)

• local heating
• liberation of gas from surroundings - up to 1
  bar!
• generation of plasma in stationary ion cloud
• uni-polar arc
• local melting
• characteristic times are very short compared to
  (our) pulse lengths (SRF tests at CW)

14
Model questions
Surface field

• How can we explain the pulse length dependence of
  the breakdown limit?
• Is it simply 1/power? (independent of f)

Pulse length
15
Pulse width data

• how does it depend on the definition of the onset
  of breakdown?
• Is the time dependence independent of circuit
  parameters?

CTF2 structure
X-band TM020 single cavity
16
Plans

• Summarize cavity and structure performance to
  date -vs- frequency.
• New structure designs
• Low surface field coupler.
• Reducing radius of the first cell after the
  coupler by 0.3 mm, surface fields on the coupler
  iris are decreased by about 1/3
• Coupler problems observed in CTFII/Haimson tests
• (This change underway)

17
Surface Preparation

• High pressure rinse   
• High pressure rinsing process for the short
  structures - KEK.
• Reduces surface particle count 10x
• SRF use of megasonic cleaning in water
• May work for completed structures   

18
Surface Preparation (2)

• Glow discharge cleaning      
• Try chemo-mechanical cleaning of the cells also
  electropolishing
• surface quality -vs- etch depth. 200 microns
  removed from Nb surfaces for SC cavities
• 2 microns are etched from non-diamond turned
  parts and 10's of nm are etched off of diamond
  turned parts.
• At CERN, final diamond turning is done dry
• NLC group uses kerosene

19
Handling / Assembly

• Measure particle count of N2 blown through old
  and new structures
• Use a vacuum furnace instead of H2 furnaces for
  bonding and brazing
• vacuum furnace cleanliness
• Damage for both diamond and conventionally turned
  cells for both vacuum or hydrogen furnace
  brazing.
• Processing gains not lost after structure exposed
  to air.

20
Structure Measurements

• Measure dark current dependence on pulse length.
• Measure HOM power from DDS3 manifolds - localize
  the breakdown.
• Measure temporal and wavelength dependence of
  emitted light.
• Use better gas detectors and do a RGA analysis at
  different stages of processing. Even at this
  level, do not see pressure increases with
  breakdown in some cases.
• Auger measurements of processed noses
• may be due to other surfaces acting as getters.
• Measure iris profiles and SEM surfaces
• processed X and S band cavities

21
Structure measurements and processing strategy

• Avoid repeated breakdown events
• Develop diagnostic techniques
• Sense onset of damage
• Control high power processing
• Look at dark currents as a breakdown precursor.
• Correlate radiation, dissipated RF energy,
  acoustic

22
Conclusion

• Thorough and open discussion
• Good input from superconducting technology
• Re-examination of contaminant control in x-band
  structure fab
• Coupon / single cell tests
• Re-evaluation follow up workshop 4 to 6
  months