DESIGN AND FABRICATION OF THE SPEAR 3 VACUUM SYSTEM
Nadine Kurita and J Langton
Stanford Synchrotron Radiation Laboratory, USA
The SPEAR storage ring is a racetrack-shaped light source consisting of 18 optical cells separated by 3.2, 4.8, and 7.6 m
straight sections. Under the SPEAR 3 project, the magnet lattice will be replaced by a double bend achromat (DBA) and the
vacuum chamber will be converted to a copper ante-chamber configuration. The primary motives for choosing copper are high
heat conductivity, low molecular desorption rate, and in-house fabrication experience at SLAC. In total, SPEAR 3 has 54
bellows-separated chamber sections (excluding drift tubes) each with 2-3 discrete photon stops and vacuum pumps. The
chamber sections are e-beam welded from machined copper plates. Under full beam loading (500 mA at 3 GeV) the dynamic
pressure is expected to be < 2nT. In this paper, we discuss design and fabrication issues associated with the SPEAR 3 vacuum
chamber and pump systems.
The SPEAR 3 vacuum system must meet several important performance criteria, namely, (1) safe, reliable
operation with up to 500 mA electron beam current, (2) low beam impedance, (3) dynamic gas pressure < 2 nT and
(4) stable BPM supports [1,2]. In addition, the vacuum chamber is constrained to fit within the compact SPEAR 3
magnet lattice and have sufficient wall strength under full vacuum loading. Given these design goals, a number of
vacuum chamber materials were considered including the ALS design (aluminum/clamshell) , the
ANKA/BESSY-II/SLS design (stainless steel/deep drawn sheet)  and the APS design ( aluminum/extrusion) .
A copper system was also considered due to the high strength, high thermal conductivity and low out gassing rate.
Electrically, the high conductivity also screens power supply ripple. Although previously untested in modern light
sources, copper was a prime candidate because the material properties are well suited to the application and SLAC
had manufacturing experience with the PEP-II HER arc chambers [2,6].
The decision to proceed with copper was largely based on a recommendation from the SPEAR 3 Machine
Advisory Committee to make the vacuum chamber passively safe to dipole radiation. The passively safe
requirement allows arbitrary beam steering at full current with insertion devices (ID) open and beam line front
ends closed. At 500 mA, the radiation power is a formidable 13 kW per dipole, or approximately 70 W/mrad. In
practice, the dipole radiation can strike anywhere in the copper chamber but not on SS ID chambers. With IDs
closed, the copper chamber can withstand ID radiation with up to 50 mA circulating current. In order to open the
beam line front ends, the orbit interlock system must be active with the beam in a roughly |x|<5 mm by |y|<1 mm
Referring to Figure 1, each of the 18 magnet cells in SPEAR 3 contains three bellows-separated
chamber/ante-chamber sections. The outboard dipole chambers are ~4 m long and the center chamber is ~1.8 m
long. The sawtooth configuration
creates two radiation ports per cell: the
first chamber (BM2) has a 22 mrad exit
port for upstream ID radiation and the
third chamber (BM1) has a 24 mrad
exit port for radiation from the first
dipole. In total, radiation exit ports
exist for all 18 ID straight sections and
14 of the 18 'standard' cell dipoles. All
radiation from the second dipole
18.8 mm 13 mm
Figure 1 Three chamber standard cell configuration Figure 2 Standard chamber cross-section
Figure 3 Dipole chamber fabrication in progress Figure 4 Center chamber assembly
terminates on masks or crotch absorbers. No radiation is taken from the eight "matching" cell dipoles that
transition to the 7.6 m racetrack straights. As indicated in the Figs. 1, 3 and 4, the radiation masks and crotch
absorbers are installed through the outboard sidewalls of the chambers or welded in through the top, respectively.
Figure 2 shows the 84 mm x 34 mm hexagonal cross-section of the main vacuum chamber. The radiation slot is
50 mm x 13 mm and the ante-chamber is 18.8 mm high with varying depth. The 13 mm slot height is a trade-off
between maximum vertical beam steering envelope and minimum microwave power transmission through the slot.
Magnet pole tip radius and chamber thickness also impact the slot height (2 mm magnet-chamber clearance).
Factoring in construction imperfections, deflection under vacuum loading, chamber sag and alignment tolerance,
the effective slot height is ~10 mm.
The ±17 mm vertical height of the beam channel was selected to provide maximum beam acceptance and
minimum resistive wall impedance. The main contributors to impedance in the ring are ±6mm ID chambers. All
transitions follow a minimum 5:1 slope where possible. Crotch absorbers and pump ports are placed well away
from the main concentration of image currents. The horizontal chamber radius (±42 mm) was chosen for injection
clearance in the septum region and carried throughout the machine to simplify manufacturing. All horizontal
components have a minimum radius |x|=30 mm with the exception of the septum at x=-25 mm.
2 CHAMBER FABRICATION
The chamber construction sequence is similar to the machine/weld "clam-shell" procedure used for the
aluminum ALS chamber  . Initially, ½-hard Class-I OFE copper plates were used to provide a strong, high yield
strength material. But the ½-hard copper resulted in unacceptable machining distortion which proved difficult to
correct. Tests with 18 mm thick, 1/8-hard plates led to tolerable distortion that could be easily removed with a
semi-automatic straightening procedure. To further control distortions, the machining process was refined to
include multiple rough and finish passes while flipping the copper plates between each cut. Even fine "skin" cuts in
copper change surface tensions and can cause the material to warp. As indicated in Figs.3& 4, the top and bottom
plates of each chamber are separated by sidewalls to minimize waste in the machining process. To prevent
excessive deformation under vacuum loading, the copper is 5 mm thick over the electron beam channel, 15 mm
thick over the radiation slot and 3 mm thick in the pole pockets. Numerical simulations of chamber loading effects
show a maximum deflection of < 500 µm in the widest regions.
After machining, the plates are cleaned with the same multi-step chemical etch/clean procedure used for
PEP-II to remove cutting fluid residue, solvents and hydrocarbons. The finished product is well within
dimensional tolerances and has a mirror-like finish on the inner chamber surface.
For the welding sequence, custom tooling fixtures were used to clamp sidewalls and auxiliary components to
the top and bottom plates. To achieve proper weld strength and chamber flexibility under vacuum loading, the e-
beam weld parameters, weld angles and sidewall thickness had to be carefully adjusted. In addition to the main box
welds, copper-to-SS pump port transitions, photon stops and CuNi inserts at corrector sites were welded into place.
As shown in Figure 4, six parallel water-cooling lines traverse the vacuum chamber for thermal stability. To
monitor overall chamber straightness and weld distortions, a spring-loaded "mouse" equipped with corner cube
reflectors for laser metrology was pulled through the radiation slot of each chamber. Finally, SS conflat flanges
brazed to copper transition pieces were welded to the ends of each chamber section. The finished chambers were
processed under vacuum at temperatures up to 200 .
The low-conductivity CuNi inserts at
Vertical Corrector - Chamber Frequency Response (on-axis)
corrector sites allow fast field penetration of the (from Mafia)
corrector field. CuNi is about 20 times more 0.00
0.1 1 10 100 1000
resistive than OFE copper allowing the –3 dB
rolloff for field penetration to approach ~100Hz CuNi
(Fig.5). Pockets machined in the copper chamber -6.00
above and below the adjacent radiation slot -9.00
improve the transverse AC field uniformity from
the corrector magnets. MAFIA simulations
predict that over a region |x|<5 mm and |y|<1 mm, -15.00
the horizontal field uniformity is better than 1%
out to 200 Hz. Figure 5 Bode plot with and without CuNi insert
To control radiation thermal loads, four
discrete GlidCop TM  masks per three chamber cell remove radiation power < 100W/cm. Three additional crotch
absorbers with crenulated surface construction  intercept power in the range 0.5-10 kW (power densities up to
30 W/mm2). A titanium sublimation pump is mounted beneath each crotch absorber to remove the photon-induced
gas load. Due to tight tolerances on photon beam ray tracing, the final chamber alignment will be registered to the
photon beam stops. The BPMs will be calibrated using beam-based alignment. To control radiation produced by
electron beam loss, a fixed mask will be installed at x=-30 mm where dispersion reaches the highest level
(η=0.45m). This mask is intended to intercept off-energy electrons at one fixed location as they spiral toward the
center of the machine. The dedicated loss
point reduces shielding requirements in the
rest of the tunnel.
Temperature rise on the vacuum
chamber is monitored with ~165
thermocouples operating over the range of
30-200 . Each of the 54 crotch absorbers
is equipped with a thermocouple and klixon
to protect against burn-through. The other
~100 thermocouples will be located on
bellows and at locations subject to high
power loading (RF cavities, scrapers, beam
stoppers, etc). Resistance temperature
detectors (RTDs) will be used for higher Figure 6 Pressure profile in a standard cell
resolution in the 10-70 range.
3 PUMP SYSTEM
The SPEAR 3 pump system uses
150 & 300 l/s noble gas ion pumps (IP)
and 1,500 l/s titanium sublimation
pumps (TSP) for a total pumping
speed of over 5,000 l/s per cell. Within
100-200 A-hr of operation, the
photon-induced desorption coefficient
on the photon stops should be below
η~2×10-6 molecule/photon . Figure
6 shows the N -equivalent pressure
profile for a section of vacuum
chamber at 500 mA after 100 A-hr
processing. On average, the predicted Figure 7 TSP canister and fins to enhance surface area
base pressure is well below the 2 nT
design goal. Each chambers is pre-processed to <200 under vacuum, so no in-situ bakeout is needed.
The primary vacuum pumps are PEP-II style TSP modules with an 8” diameter aluminum housing and
extruded aluminum fins around the interior (Fig. 7). Each TSP is located immediately below a 10o incline crotch
absorber which can accept up to 10 kW of radiation power. The fins inside the TSP increase the condensation
surface area for Ti deposition to improve pumping capacity (approximately 6,500 cm2 per pump). Three separate
Ti filaments are installed for maximum service lifetime. The filaments are mounted horizontally to improve Ti
coverage and to simplify filament replacement. Activating the pumps creates a reactive Ti surface with direct
line-of-sight to the photon absorbers. The line-of-sight construction also permits methane pumping as a result of
CH4 ionization by photoelectrons from the photon stop.
Although calculations indicate a fairly constant pump speed before the Ti layer saturates, measurements
indicate the pump speed soon drops to < 1,000 l/s and holds steady at that level until saturation. The TSP pump
capacity is up to 0.5 T-l CO-equivalent between regeneration cycles. At full operating current, the TSPs should
require flashing every 50-100 A-hr. The pumps can be flashed with beam in the machine with no impact on user
operations. About 500 flashes are possible before filament replacement is required (>5yr lifetime). High TSP
pump capacities have been observed at the ALS where the pumps are typically flashed only about 4 times a
The ion pumps use the diode configuration to pump noble gases. Where possible the ion pumps are located
near radiation masks (high gas load) or at locations of high betatron function to improve electron beam lifetime.
4 BELLOWS, SUPPORTS AND KICKERS
SPEAR 3 will have over 60 bellows modules with low impedance RF liners and active cooling. The bellows
will have flanged couplings equipped with RF seals to provide a continuous electrical connection – even small
discontinuities create HOM fields that contribute to local heating, as much as a few watts per element.
By design, each bellows module allows for vacuum chamber motion up to ±2 mm and 25 mrad. The ±2 mm
lateral offset limit is set by the welded bellows, not the RF shield. As shown in Fig. 8, the RF shield configuration
uses the “double finger” mechanism developed for PEP-II  . The shield bridges the bellows to provide
continuous electrical path for image currents to minimize beam impedance. The shield design employs silver-
plated GlidCop® RF shield fingers which slide on the outside wall of a Rhodium-plated GlidCop® stub. Each RF
shield finger has a mating Inconel 718 spring finger to provide sufficient electrical contact. This design moves heat
away from high stress areas and, because the RF shield does not provide the spring force, it is kept relatively
stress-free. Since the RF shield is shadowed from off-axis radiation, the bellows are not the current-limiting
components under mis-steered electron beam conditions.
Each SPEAR 3 magnet and vacuum chamber is individually supported on a kinematic 6-D strut assembly. In
order to stabilize vertical BPM positions (< 5 µm per o C), critical BPMs are isolated on one side by bellows and
supported on low thermal expansion Invar rods attached to the bottom of the raft. At these locations, the chamber is
allowed to thermally expand away from the support system towards the next bellows module. Figure 9 shows how
the support system fixes the BPM module in (x,y,z) position.
Low impedance slotted-pipe devices developed for the DELTA storage ring will be used for the injection
kickers . The copper kickers will maintain the same cross-section as the standard vacuum chambers with two
conductors separated by four slots parallel to the beam axis to produce an inductive loop. The design is electrically
SPEAR 3 Bellows Module
Figure 8 PEP-II style Bellows Module Figure 9 BPM support system
symmetric with two inputs for current pulses of opposite polarity .
Two 45cm transverse kickers (one horizontal, one vertical) utilize copper plates formed into an elliptical
shape to create 50Ω stripline electrodes with profile similar to the main chamber. The side wall shadows the
vertical kicker from synchrotron radiation. The electrodes are suspended in the chamber by Inconel flex supports
connected to 50Ω feedthroughs. The only mode for heat transfer (beside the feedthroughs and flex supports) is via
radiation. Assuming an emissivity of η=0.5 for the chamber and striplines, the anticipated temperature rise is 50
The SPEAR 3 vacuum system is fabricated with technology new to the light source community but proven at
collider physics facilities. In particular, a box weld clam-shell construction from copper plates was chosen to
withstand high power loads and to minimize molecular desorption and resistive wall impedance. Although
aluminum, copper or stainless steel chambers can be made to work for light source applications, experience with
copper fabrication at SLAC weighed heavily in the decision to use copper. In the SPEAR 3 chamber design, over
95% of the synchrotron radiation beam strikes discrete photon stops. By pre-processing the chamber sections, in-
situ bake-out is not required. The ~100,000 l/s pump system uses standard noble ion pumps and TSPs based on the
PEP-II LER design. In order to reduce overall technical complexity, no NEG pumps are used. Invar BPM supports
limit thermally-induced BPM motion in the vertical plane. The resulting system is expected to yield a based
pressure of <2nT at 500mA (>15hr beam lifetime) and provide BPM modules with less than 5µm/ vertical
Table 1 Properties of common light source vacuum chamber materials
Material Thermal Electrical Yield Skin
Conductivity (W/m- oK) Conductivity (Ω -1/m-1 ) Strength (psi) Depth (mm)
Aluminum (6061-T6) 180 3.5 × 107 ~40,000 85 x f –1/2
Copper (OFE H02) 390 5.9 × 107 ~33,000 66 x f -1/2
Stainless Steel (316L) 16 1.4 × 106 ~30,000 425 x f -1/2
Ed, wherever you are, you owe us a pitcher. SPEAR 3 is supported by DOE Office of Basic Energy Sciences
(Contract DE-AC03-76SF00515) and the National Institutes of Health.
 "SPEAR 3 Design Report ", SLAC Pubs, 1999.
 N.R. Kurita, et al, "Overview of the SPEAR 3 Vacuum System", 1999 Particle Accelerator Conference, New York, NY, p 1363.
 K. Kennedy, et al, "Vacuum System for the LBL Advanced Light Source (ALS)", 1989 Particle Accelerator Conference, Chicago,
Illinois, p 560.
 G. Heidenreich, L. Schulz, P. Wiegand, "Vacuum System for the Swiss Light Source", 1998 European Particle Accelerator Conference,
Stockholm, Sweden, p 2196.
 J.R. Noonan, et al, "APS Storage Ring Vacuum System Performance", 1997 Particle Accelerator Conference, p 3552.
 C. Perkins, et al, "Vacuum System for the PEP-II High Energy Ring", 1994 European Particle Accelerator Conference, London, U.K.
 SCM Metal Products, Inc., “GlidCop Product Information”, North Carolina.
 L.Schulz, "Specification for the SLS Storage Ring Photon Absorber", SLS-SPC-TA-1999-0125, April 14, 1999.
see also, I.C. Sheng, et al, "A Conceptual Design and Thermal Analysis of High Heat Load Crotch Absorbers", 1993 Particle Accelerator
Conference, Washington, D.C., p 1497.
 C.L. Foerster, C. Lanni and K. Kanazawa, "Measurements of PSD from Thick and Thin Oxide of KEK-B Collider Copper Beam
Chambers and a Stainless Steel Beam Chamber", JVST A 19 (4), Jul/Aug 2001, p 1652
and C.L. Foerster et al, “Photon Stimulated Desorption (PSD) Measurements of Extruded Copper and Welded Copper Beam Chambers
for the PEP-II Asymmetric B Factory”, 4th European Particle Accelerator Conference, July 1994.
 K. Kennedy, private communication.
 N. Kurita, et al, "Final Design and Manufacturing of the PEP-II High Energy Ring Arc Bellows Module", 1997 Particle Accelerator
Conference, Vancouver, Canada, p 3639.
 G. Blokesch, et al, "A Slotted-Pipe Kicker for High Current Storage Rings", NIM A 338 (1994) p 151-155.
 J. Sebek, D. Arnett, J Langton and C. Pappas, "SPEAR 3 Injection Kicker", 2001 Particle Accelerator Conference, Chicago, Illinois, p