1. Vacuum System Failures at HERA
M.Seidel, DESY
OLAV Workshop, CERN, April 12, 2005

2. Categorization of Failures
Non-Interrupting Failures
getter pump power supply failure
cable failure; radiation damage
pump address or location error (software)
NEG pump heating circuit – shortcut to ground!
vacuum interlock electronics failure (dangerous!)
collimator damage by beam (hard to diagnose)

3. Categorization (2) Interrupting Failures (Leaks!)
beam induced leaks by heating/thermal shock
leaks by electrical failure in feedthroughts
water-vacuum leaks at brazing joint
RF window breaks by heat load
(after long term metallization)
fatigue failure at bellows (wire scanner, collimator)
(not discussed here → systematic problems)

4. Example non-interrupting failure – Proton Collimator Damage
Tungsten Collimator Jaw, 40cm long
melting point: 3422 °C !
p-beam: 920GeV, 100mA
damage was not realized during operation; mechanism not clear
Result: reduced collimation efficiency; hard to diagnose!

5. Interrupting Failure: p-Loss induced leak in VAT sealed flange connection
specially shaped vacuum chamber; VAT sealed;
located in half quad in IR
after uncontrolled beam loss (dump fired too late):
high activation 2mSv/hour
leak at seal
>10-6 mbar l/sec

6. Model for p-Loss induced leak
Newton’s law:
τ loss time approx. 2ms!
yield strength of seal exceeded by compression force for beam energy deposition stored
 very well possible!

7. Leaks from asymmetric heating of Conflat flange connections
leak occurs when heat load at flange is switched off suddenly
laboratory tests show this can happen for ΔT=40°C !!
hard SR penetrates chamber or absorber and heats flange from outside
typical signature
lab-experiments
asymetric
heating

8. Collapse of steel chamber from SR heating in 2004
keyhole chamber that collapsed,
right side of IR
schematic layout HERA IR
upstream absorber
- gap in side-channel opened up for background reasons

9. collapsed chamber chamber collapsed from heating by mis-steered SR fan
Temperature more than 1000°C
orbit feedback failed
inside view

10. New interlock – flat T-sensors
between chamber and magnet poles: PT 100 film sensors; thickness 0.4mm
5 sensors each side;
distance 30cm
already 3 beam dumps by this interlock in 2005/04

11. High Current Electrical Feed-Throughs
Problem: NEG heating circuits or Ti sublimation pump filaments are operated at relatively high currents – 30-50Amps
high electrical power can be dissipated at shortcuts or transitional resistance (connectors)

12. NEG Pump Feed-Through
internal shortcut produced an arc resulting in a macroscopic leak
happened during activation!!!

13. Water-Vacuum Leak in Brazing Connection
tuning cylinder RF cavity; internally water cooled
tuner and cavity

14. Another Tuning Cylinder...
same type of device: massive cracks developed
theory: bubble of air reduced cooling at top of cylinder

15. Summary
most serious trouble happened with heating by particle beam or SR
→ build your system passively safe if possible;
any orbit that circulates will occur!
high current electrical feed-throughs are problematic; especially with a large number of components
power supplies and electronics fail most often;
however, this does not interrupt operation (mostly)