Electron clouds and vacuum pressure rise in RHIC

1
Electron clouds and vacuum pressure rise in RHIC

• Wolfram Fischer
• Thanks to
• M. Blaskiewicz, H. Huang, H.C. Hseuh, U. Iriso,
  S. Peggs,G. Rumolo, D. Trbojevic, J. Wei, S.Y.
  Zhang
• 33rd ICFA Advanced Beam Dynamics Workshop on
  High Intensity High Brightness Hadron
  BeamsBensheim, 19 October 2004

2
Contents

• Pressure rise observations
• Injection
• Transition
• Store
• Pressure instabilities
• Counter measures

3
Pressure rise observations
1st fill with 110 Au79 bunches N0.50109 Oct.
2001
Beam lossesduring acceleration
next fill N0.44109
10-7 Torr abort limit
4
Location of pressure rise problems in 2003/04
Blue sector 10/11 Increased gas density in
cold arcs (with p)
IP10 PHOBOS Be beam pipe (high background)
Run-4 Au-Au Nov. 2003 to Apr. 2004 No of
bunches 61, 56, 45Ions per bunch 0.5-1.1?109
Yellow sector 4 Unbaked stochastic cooling
kicker (high pressure)
Blue sector 8 Unbaked collimator (vacuum
instability)
5
RHIC pressure rise observation to date
Pressure rise observed Yes pressure rise ? 1
decade E-clouds observed directly observed with
electron detector
6
Pressure rise mechanisms

• Pressure rise mechanisms considered so far
• Electron cloud ? probably dominating for
  operational problems
• Coherent tune shift in bunch train
• Electron detectors
• Comparison with simulations
• Ion desorption ? tolerable for operation
• Rest gas ionization, acceleration through beam
• Ion energies 15eV for Au, 60 eV for p
• Visible pressure rise, may lead to instability
  in conjunction with electron clouds (Au only)
• Beam loss induced desorption ? tolerable for
  operation
• Need large beam loss for significant pressure
  rise
• New desorption measurements in 2004 (H. Huang,
  S.Y. Zhang, U. Iriso, and others)

7
Electron cloud observation at injection
Indirect observation coherent tune shift along
bunch train
331011 p total, 0.31011 p/bunch, 110 bunches,
108 ns spacing (2002)
(1) From measured tuneshift, the e-cloud density
is estimated to be 0.2 2.0 nCm-1 (2)
E-cloud density can bereproduced in
simulationwith slightly higher chargeand 110
bunches (CSEC by M. Blaskiewicz)
DQ?2.510-3
W. Fischer, J.M. Brennan, M. Blaskiewicz, and T.
Satogata, Electron cloud measurements
andobservations for the Brookhaven Relativistic
Heavy Ion Collider, PRSTAB 124401 (2002).
8
Electron cloud observation at injection
U. Iriso-Ariz
Direct observation electron detectors
Observation 881011 p total 0.81011
p/bunch 110 bunches 108 ns
spacing Simulation Variation of SEYmax 1.7
to 2.1 Keep R0.6 (reflectivity for zero
energy) Good fit for SEYmax 1.8 and
R0.6 Code CSEC by M. Blaskiewicz
bunches with lower intensity
U. Iriso-Ariz et al. Electron cloud and
pressure rise simulations for RHIC, PAC03.
9
Electron cloud observation at injection
Electron cloud and pressure rise
U. Iriso-Ariz
861011 p total, 0.781011 p/bunch, 110
bunches, 108 ns spacing
e-cloud and pressure
Clear connectionbetween e-cloudand pressure
atinjection
Estimate for heassuming pressurecaused by
e-cloud 0.001-0.02 (large error from
multiple sources)
12 min
total beam intensity
U. Iriso-Ariz et al. Electron cloud
observations at RHIC during FY2003, in
preparation.
10
Pressure rise at injection cold regions
Gas density increase observed in cold regions
with intense proton beams
1.5e13 protons
intensity
1e-7 Torr
pressure in gauge
11
Cold bore pumps and gauges
H.C. Hseuh
CCG and Sorption pump every other 15m 17
CCGs per arc ( one per 30m) Most CCGs read mid
10-10 Torr
Gauge conduit
CCGs with a 1.5mx1 F conduit to cold bore Q
10-9 Torr.l/s, C lt 1 l/s ? P 10-9 Torr
Sorption Pumps with 300g charcoal _at_ 10K, S(He)
2 l/s P (He) 10-10 Torr _at_ Q(He) gt 30 Torr.l
12
Pressure rise at injection cold regions

• Cold surface not very clean
• Before cool down without pumping 1e-1 Torr(in
  almost all cold regions in the past)? About
  10-100 monolayers
• Pumping 500m arcs with turbo for 2 weeks 1e-3
  Torr(10 days after stop pumping 1e-2 Torr)
• Need to install more turbo pumps ? 1e-3 Torr
  initial pressure results in 1 monolayer

13
Pressure rise at transition
Typical transition pressure rise in Au-Au
operation
transition bunch length 4ns (18ns at injection)
beam intensities
1e-7 Torr
IR12
IR4
IR10
14
Presssure rise at transition
Strongly intensity dependent
No correlationwith beam loss?
suggestselectron clouds
15
Pressure rise at transition
Stronger pressure rise for fewer bunches in
IR12(independent of pattern in IR10 same as
Run-3, S.Y. Zhang)
? Behavior can be quantitatively reproduced in
e-cloud simulation (U. Iriso)
16
Pressure rise at transition
With fewer bunches, cloud density and current
into wall is reduced,but electron impact energy
is increased, and can lead to largerdesorption.
(simulation by U. Iriso)
45 bunches/ring
56 bunches/ring
61 bunches/ring
e-clouddensity
current into wall
electronimpact energy
400 eV
500 eV
17
Pressure rise at store IR10 PHOBOS
PHOBOS background increase after rebucketing,
drops after minutes to 2 hours(most severe
luminosity limit in Run-4)
Some thoughts on switch-off U. Iriso and S.
Peggs, Electron cloud phase transitions,BNL
C-A/AP/147 (2004). Can e-cloud codes create 1st
order phase transitions?
18
Pressure rise at store IR10 PHOBOS
Pressure before and after rebucketing (50 bunch
length reduction)
Run-4 physics stores

• Did not find narrow range that triggers problem
  for
• average bunch intensity
• peak bunch intensity
• pressure before rebucketing No good correlation
  with any parameter and duration either

19
Pressure rise at store IR10 PHOBOS
Average bunch intensity at rebucketing/pressure
drop, and duration of increased pressure sorted
by bunch patterns
20
Pressure rise at store IR10 PHOBOS, simulations
G. Rumolo, GSI
12m 40ns
Be pipe
Considered 2 casesAt IP nominal bunch
spacing (216ns) and double intensity At end of
the beryllium pipe normal intensity, spacing of
40ns then 176ns
G. Rumolo and W. Fischer, Observation on
background in PHOBOS and related electroncloud
simulations, BNL C-A/AP/146 (2004).
21
Pressure rise at store IR10 PHOBOS, simulations
G. Rumolo, GSI

• Can calibrate Be surface parameters
• No e-cloud before rebucketing (10ns bunch
  length)
• E-cloud after rebucketing (5ns bunch length)

N. Hilleret, LHC-VACTechnical Note 00-10
Modified to match observation
22
Pressure rise at store IR10 PHOBOS, simulations
G. Rumolo, GSI
Important result After surface parameter
calibration find e-clouds at end of 12m Be
pipe, but not in center? May be sufficient to
suppress e-cloud at ends
Emax400 eV and dmax2.5
Center of Be pipe
End of Be pipe
23
RHIC vacuum instabilities

• In a number of cases vacuum instabilities were
  observed (pressure grows exponentially without
  bound)
• Vacuum instabilities seen
• Only with Au79 beam
• Only at locations with unbaked surfaces
• At injection (previous runs)
• At store, after rebucketing (Run-4)
• May be at transition (growth time large compared
  to transition crossing time)
• Growth times range from 2 to 12 sec

24
RHIC vacuum instability
11.7 sec growth time
Location of unbaked collimator (unbaked due to
scheduling conflict)
25
RHIC vacuum instabilities

• Need feedback mechanisms for instability(gas
  load Q proportional to rest gas pressure P)
• Rest gas ionization by cloud electrons
• Rest gas ionization by beam
• Define parameter (analog ISR instability)

e-cloud beamionization
26
RHIC vacuum instabilities

• Critical desorption coefficient for ionizedand
  accelerated rest gas molecules
• Growth time (very approximate)

c conducdanceL half distance between pumps
r pipe radius
27
RHIC vacuum instabilities

• e-clouds contribute about 20 of effect
• Instability may be possible for Au79 and CO
  like molecules.
• Still signifcant descrepancy between reported h
  and calc. hcrit (higher charge state rest gas
  ions important?)
• Note No h measurements available for ion
  energies below 100 eV.

28
Electron counter measures

• In-situ baking of warm elements (gt95 of
  700m/ring warm pipes baked)? Occasionally
  installation schedules were too tight
• NEG coated pipes? Installed 60m last shut-down
  for test, about 250m of NEG coated pipes
  now
• Optimized bunch patterns? Most uniform along
  circumverence, used this year
• More pumping of before cool-down? Need more
  turbo pumps for full coverage
• Solenoids? Tested, no large scale implementation
  planned near term
• Scrubbing? Tested, no large scale implementation
  planned near term

29
Summary

• All operational relevant dynamic pressure rises
  in RHIC can be explained with electron clouds
  (an abnormal large beam losses can still lead to
  inacceptable vaccum)
• Electron cloud driven pressure rises observed
• With all species (Au79, d, p),
• In warm and cold regions
• At injection, transition and store
• Pressure instabilities only for Au79 and unbaked
  wall(likely caused by rest gas ionization by
  beam and electron cloud)
• Current counter measures
• Complete baking of all elements
• NEG coated warm beam pipes
• Optimized bunch patterns