Title: The Crystal Collimation System of the Relativistic Heavy Ion Collider
1The Crystal Collimation System of the
Relativistic Heavy Ion Collider
- Ray Fliller III
- University of Stony Brook
- Brookhaven National Laboratory
2Collaborators
- BNL
- Angelika Drees
- Dave Gassner
- Lee Hammons
- Gary McIntyre
- Steve Peggs
- Dejan Trbojevic
- IHEP Protvino
- Valery Biryukov
- Yuriy Chesnokov
- Viktor Terekhov
3Outline
- Brief RHIC Overview
- Collimation
- Crystal Channeling
- RHIC Crystal Collimation System
- Channeling Results
- Crystal Collimation and Background Reduction
- Conclusion
4Run Species Integrated Luminosity Energy
2000 Au-Au 7.3 mb-1 (PHENIX) 70 GeV/u
2001 Au-Au 92.6 mb-1 (PHENIX) 100 GeV/u
2002 Polarized protons 100 nb-1 (STAR) 100 GeV
2003 d-Au 27 nb-1 (PHENIX) 100 GeV/u
2003 Polarized protons 2500 nb-1 (STAR) 100 GeV
2004 Au-Au 1368 mb-1 (PHENIX) 100 GeV/u
2004 Polarized protons 3200 nb-1 (STAR) 100 GeV
2005 Cu-Cu 100 GeV/u
5RHIC Capabilities
- Two 3.8 km counter-propagating superconducting
rings BLUE (clockwise) and YELLOW
(counterclockwise). - Can accelerate anything from polarized protons
(250 GeV) to fully stripped gold ions (100
GeV/u), possibility of colliding uneven species. - Six IRs with four experiments (STAR, PHENIX,
BRAHMS, PHOBOS). - Typical store each ring contains 110 bunches of
109 gold ions or 1011 polarized protons.
6Typical RHIC Parameters
- 95 norm. Emittance e15 p mm-mrad
- rms momentum spread sp 0.13
- Bunch length sl 0.19 m
- Energy 100 GeV/u
- Store Length 4 hours
- Beam size at collimator 5.3mm (bPHENIX1m)
7Need for Collimation
Various processes cause particles to enter into
unstable orbits with large betatron amplitudes,
causing beam halo formation. These halo
particles cause The job of the collimation
system is to remove the halo and alleviate these
problems. In addition, it should provide a well
defined location for beam losses in case of
equipment failure.
- Background in experiments
- Excessive radiation in uncontrolled areas of the
tunnel - Magnet quenches in superconducting machines
- Equipment malfunction and damage
8Naive Collimation
Collimator
Beam
Naively, all particles that enter the collimator
are stopped in the collimator.
Most particles hit near edge and scatter out of
the collimator forming secondary halo!
However, that is usually not the case.
9Two Stage Collimation
Since primary collimator acts as a scatterer,
secondary collimators are necessary to increase
energy loss and absorb secondary halo particles.
The number of secondary collimators grows quickly
when background or machine protection
requirements are strict and a high collimation
efficiency is required (see LHC collimation
system!).
10A simpler way to collimate
Use a bent crystal to channel halo away from the
beam core, intercept with a scraper downstream.
Number of secondary collimators can be greatly
reduced.
11Crystal Channeling
Ions properly aligned to the crystal planes are
channeled.
Particles with large incident angles scatter
through the crystal
12Interplanar Potential
Ions properly aligned to the crystal planes see
an average potential. This potential is skewed
by the bending of the crystal.
dp
Curvature shifts minimum
13Critical Angle qc
The channeling condition gives an
angle qc, above which a particle will not be
channeled.
Using a Si crystal with 100 GeV/u Au or 250 GeV p
, qc11 mrad
For 100 GeV p, qc19 mrad
14Channeling Efficiency
The integral of the incoming particle
distribution over the channeling phase space is
the channeling efficiency
For a beam with uniform divergence 2Fgt2qc
15Dechanneling and Volume Capture
16CATCH Simulation
CATCH by Valery Biryukov
17Important Considerations for Crystal Collimation
- Crystal alignment to beam halo.
- Angular divergence of beam halo hitting crystal.
18Crystal Collimator Geometry
19Model of Beam Hitting Crystal
Assuming a Gaussian beam distribution of
- J J(x, x, d) is the particle amplitude
- e is the rms unnormalized emittance
- d is the fractional momentum deviation
- sp is the rms fractional momentum spread
By transforming from J, d to x, x, d and
integrating over momentum
20Angular Alignment
Assuming the distribution extends over the entire
crystal face, the angle between the beam orbit
and particles striking the crystal is
- x0 is the distance between crystal and beam
center - Dx is width of crystal face
- a, b, D, D are lattice functions at crystal
The crystal planes need to be at this angle
relative to the beam orbit! This is proper
alignment!
21Angular Divergence
The equation for angular divergence, sx(x0), is
not very illuminating. However, it depends
strongly on
- a, D large values increase sx(x0)
- sp large values increase sx(x0)
- b, D large values decrease sx(x0)
- Dx large values increase sx(x0) (assuming
particles hit whole crystal face)
By optimizing these parameters, the angular
spread of beam across the crystal face is
minimized.
For those who REALLY want to see the equation,
read my thesis!
22Phase Space at Crystal
When crystal is moved into beam, it needs to be
realigned
23Angular Width Model Optics
Critical Angle
bPHENIX 2 m
model
24Angular Width Measured Optics
- and D affect ellipse
- orientation and shape
Critical Angle
measured (FY2001)
bPHENIX 2 m
25Caveat Emptor!
There are a few holes in the model
- Particle distribution Gaussian in the tails??
- Assumption that particles strike across the whole
face of crystal. - Does not take into account multiple turns.
- Not useful for volume capture predictions.
However, this model gives us a starting place.
26Placement of the Crystal
- Crystal should be placed at a location that has
low a and D and a maximun of b so that - xp is independent of x0
- sx(x0) is reduced
- Channeling efficiency is increased
- Operation of crystal collimator is easier
However, in RHIC all warm spaces have large a!
27RHIC Collimation System
Changed after FY2003
STAR
Scraper can move horizontally, vertically and
rotate in horizontal plane
Downstream PIN Diodes
Upstream PIN Diodes
Hodoscope courtesy of Y. Chesnokov and V.Terekhov
28Vessel Cutaway
Pivot
Inchworm
Crystal
Moveable Stage
29Crystal Vessel
Crystal
Crystal Motion
Beam
30Crystal
31Measuring Crystal Angle
By measuring the deflection of the laser beam,
the crystal angle is measured
- Crystal can rotate approx 6 mrad
- Measurement Resolution 20 mrad
- Angular Step Size 30 nrad
32Lattice Functions
bPHENIX 2 m FY2003
33Synopsis of Data
Run Species bPHENIX Stores Scans
FY2001 Au 5 m 8 27
FY2001 Au 2 m 4 24
FY2001 Au 1 m 12 109
FY2002 p 3 m 11 119
FY2003 Au 2 m 4 20
34Typical Crystal Scan
sx(x0)
Volume Capture
Crystal Aligned
xp
Crystal Channeling
November 12, 2001 Au beam at store.
35Hodoscope Signal
Very noisy compared to PIN diodes. Coincidence
rate is almost useless. Limited use in analysis.
36Comparison to Simulation
- Model Optics
- Location wrong
- dip width too narrow
- efficiency too large
Design optics do not agree well with data.
However, measured optics agrees better.
Simulation used CATCH and one turn matrix.
37Comparison to Simulation
Volume capture region strongly affected by number
of turns in simulation.
38Channeling Angle vs. Position
b1m at PHENIX
Design mrad/mm
Measured Optics mrad/mm
Data mrad/mm
39Beam Divergence
Run bPHENIX sx(x0) mrad sx(x0) mrad sx(x0) mrad sx(x0) mrad
Run bPHENIX Design optics Measured optics Simulation Channeling data
FY2001 5 12.3 39 4
FY2001 2 9.98 19 1 20 1 78 4
FY2001 1 8.91 9 1 11 1 38 3
FY2002 3 10.8 58 3
FY2003 2 9.98 14 1 16 1 28 2
Even using the correct optics, the predicted
angular spread is too small. Multiple turns are
not in the theory! Assumed Gaussian halo
distribution!
40Channeling Efficiency
Run bPHENIX Channeling Efficiency Channeling Efficiency Channeling Efficiency Channeling Efficiency Channeling Efficiency
Run bPHENIX Design optics Measured optics Simulation Measured width Channeling data
FY2001 5 59 19 2 24 3
FY2001 2 71 39 2 37 1 9 1 28 3
FY2001 1 74 75 1 56 3 20 2 19 3
FY2002 3 79 21 2 26 3
FY2003 2 71 52 2 50 1 26 2 26 3
Channeling Efficiency does not match predictions
from the theory. This is because the beam
divergence on the crystal does not match theory.
Using the measured beam divergence (from sx(x0)
) the efficiency agrees well for most cases.
41Channeling Results
- RHIC optics did not match model, so initial
predictions overestimated crystal performance - Simple theory overestimates channeling efficiency
lacking multiple turns, model of halo
distribution too simple. - Simulation agrees with data well.
- Channeling efficiency is understood once optics
and beam halo distribution are understood. - Accurate knowledge of lattice functions and halo
distribution VERY IMPORTANT!
Will low channeling efficiency result in too much
scattering and hurt collimation?
42STAR Background
4 crystal scans with different scraper positions
- xs
Crystal not moved.
43Other Experiment Backgrounds
Only BRAHMS see significant effect
44Placing the Scraper
Scattering from scraper
Scattering from crystal
By using both sets of PIN diodes, we can know
when the scraper becomes the primary aperture!
45STAR Background Reduction
Scraper only
Raw Background
Crystal collimation does not do better than
scraper alone!
46Crystal Collimation vs. Raw Background
Scraper moves closer to beam
Crystal Collimation reduces Background to
uncollimated rate
Au beam, d-Au run, crystal collimation not
always effective in reducing background.
47Crystal Collimation Results
- Crystal can cause background in experiments.
- Scraper position very important.
- Because of low channeling efficiency, crystal
collimation was not successful. - Scraper alone collimated the best.
- Crystal Collimator removed from RHIC.
Traditional two stage collimation system
installed for FY2004 run.
48Summary
- Bent Crystals were used for collimation in RHIC
- Crystal Channeling worked as expected once
lattice functions and halo distribution were
understood. - Collimation was unsuccessful because lattice was
not optimized in area of collimator. - Crystal caused background.
- Tevatron is going to install our vessel (and Ill
be following it there!)
Questions??
49Single Stage Collimation
During d-Au run, backgrounds were reduced by as
much as a factor of 5.
Fill 03094 d-Au run
Vertical Collimator
Closer to beam
Horizontal Collimator
Partially retracting the vertical collimator
increases backgrounds
50Upgraded Collimation System
PIN Diodes downstream of V1 and H1 collimators
are not shown for clarity
- Crystal Collimator removed
- Primary is the same collimator as previous runs,
moved to location reserved for the Crystal
Collimator - Secondary collimators are based on design of
primary - Controls software upgraded to include
manual/automatic control of collimators
51Upgraded Collimation Results
Fill 04436 Au-Au run
PHENIX
Backgrounds reduced by factor of 11, 2x the
pervious run!
STAR
Collimators move simultaneously.
52Summary
- Single stage collimation was adequate during
lower luminosity runs. - Two stage collimation was successful during the
FY2004 Au-Au run. - Two more vertical collimators are installed for
the FY2005 Cu-Cu run.