SLAC-PUB-95-6866
b
P
ABSTRACT
cbzf-7m-)&--Bd.
The Stanford Linear Collider*
Paul
Emma
Stm
ford
Linear
Accelerator Center
Stunford
Culifonzia
94309
The Stanford Linear Collider (SLC)
is
the first and only
high-energy
e+e-
linear collider in the world. Its most
remarkable features are high intensity, submicron sized,
polarized
(e-)
beams at
a
single interaction point. The main
challenges posed by these unique characteristics include
machine-wide emittance preservation, consistent high intensity
operation, polNized electron production and transport, and the
achievement of
a
high degree of beam stzability on
all
time
scales. In addition
to
serving as an important machine for the
study of
2
boson production and decay using polarized beams,
the SLC is also an indispensable source of hands-on experience
for future linear colliders. Each new year of operation has been
highlighted with
a
marked improvement
in
performance. The
most significant improvements for the 1904-95 run include
new low impedance vacuum chambers for the damping rings,
an
upgrade to the optics and diagnostics of the final focus
systems, and
a
higher degree of polarization from
the
electron
source.
As
a result, the average luminosity has nearly doubled
over the previous year with pe* approaching
10'0
cm-2s-1
and an
80%
electron polarization at the interaction point.
These developments as well
a$
the
remaining identifiable
performance limitations will
be
discussed.
I.
1994-95
RUN
SUMMARY
In 1994-95
the
interaction point (IP) beam intensity has
been raised
to
3.5~10~~
particles per bunch (ppb)-an increase
made possible through the design and installation of new low
impedance damping ring vacuum ch'mbers
[l].
The electron
polarization
has
increased to 80% at
the
IP by using
a
100
nm
thin strained lattice
GaAs
photociatilode in the electron source
[2]. A major upgrade to
the
final focus optics allows a
reduction of
the
IP
vertical beta function which can produce
an
IP
rms vertical spot size of 400-600 nm
[3].
Work has
continued throughout the run to improve beam stability via
feedback refinements, optical modillcations and magnet
support
alterations. The resultant number of
2
bosons logged
by
the
SLD
has
increased from
11,000
at 23%
e-
polarization
in 1992, and 52,000 at 63% in 1993,
to
over 100,000
at
80%
in 1994-95. Fig.
1
shows
2
production over this period. Due
to scheduled interruptions and an increased number of various
failures, machine up-time has been somewhat lower in 1994-
95
(-65%) than in 1993
(-75%).
Table
1
lists typical
operating par'meters at the IP along with an estimate of their
variability over the extent of the run. The electron vertical
emittance and positron intensity have been
the
most
Work supported
by
Depuraasnr
of
Energy cantrucr
DE-
.
*
AC03-
76SFO051
S
problematic in terms
of
variability. Detector backgrounds,
which are generally quite low,
also
vary
over the run. They are
typically traced to the production of beam
tails
generated
in
the
main linac.
8l
1991
--
1992-
--
1993
-
--199&- 1995
Fig
1.p'~
per week and integrated
2's
from
1991
to
1995.
Table
1.
Typical
IP
operating parameters for the 1994-95 run.
beam
energy
E
GeV 45.64
e-
intensity
N-
1Ol0ppb 3.3-3.6
n.
POLARIZED
ELECTRON
SOURCE
Since early 1992 the SLC
has
been operated exclusively
with
a
polarized electron beam. The electron polarization at the
source is now >809'0-a significant increase over
the
1992 and
1993 values of 25% and 65% respectively. The polarized
electron source [23 presently uses a strained lattice
GaAs
photocathode which
is
biased
at
120 kV and excited with
circularly polarized light generated by
a
pulsed Tisapphire
laser system. The source intensity is 7-8~10'~
e-
per bunch
(3.5~10~~ at the IP). During the second half of the run, the
cathode quantum efficiency was held below its maximum value
in order to yield the highest possible polarization. Periodic
cathode recesiations
are
performed every
-5
clays
through a
simple computer automated process which requires -20
minutes to complete. The system has been remarkably reliable
with <2% unscheduled downtime. The success
of
the high
energy colliding be'm physics progr,m
at
the SLC is due
in
large
part
to the success of the polarized electron source.
llllllll~
/I
V=lW
I
-I
0.0
I
'
Ill
IIIII(
2 3
4
N/10"
Fig
2.
Damping ring extracted hunch length vs.
e-
intensity for
old
and
the new vacuum chamber. Data pints represent
measurements performed on the new chainher in
1994.
At
the nominal machine repetition frequency of 120 Hz the
electron store time
(-8
msec) is half that of the positron ring
(-16 msec). Consequently, the electron damping time is more
critical. In 1993 a reduction in transverse partition numbers
was
achieved by stretching the ring circumference in order to
shorten the transverse damping time by -15% [SI. Recent
measurements show damping times of 3.3-3.6 msec
horizontally and 4.1-4.2 msec vertically
[9].
With an 8.3 msec
store the typical extracted electron vertical emittance is 2-3
mm-mad while it is possible
to
achieve
c1
mm-mrad with a
16
msec store at a repetition rate of
60
Hz.
In the
past,
effort has been devoted to correcting transverse
emittance dilution
in
the SLC bunch compressors
[10-11].
Skew quadrupoles, skew sextupoles and octupole magnets were
installed in previous years to correct first, second and even
third order anomalous dispersion. The large energy spread
(-1
%)
and the strong bending necessary for
a
potential ten-fold
bunch length compression present severe alignment,
construction and multipole field error tolerances. These efforts
have been, for the most part, successful. However a 10-30%
emittance dilution remains (partially due to
an
increased
compressor voltage-see below). Efforts need
to
continue here.
The form of bunch compression was changed in 1994.
Prior to this, the bunch
was
'under-compressed' to 1.3 mm
with
a 29 MV rf voltage which initiates a <go" longitudinal
phase rotation. Starting in 1994 the bunch is now 'over-
compressed', also to 1.3 mm, but by using an rf voltage of
41 MV for a phase rotation of
>go".
The motivation is to
reduce the end-of-linac energy spread by partial cancellation of
energy spread due to the longitudinal wakefield
in
the linac and
that due to rf curvature [12]. This technique successfully
reduced the end-of-linac energy spread from -0.25% prior to
1994, to -0.12%
rms.
In addition, long low-energy tails in the
bunch distribution are no longer generated. A small
compromise is made in beam transmission through the
compressor beamline where large dispersion and increased
energy spread
(-1%
at 29
MV
and -1.4% at
41
MV)
produce a
510%
beam loss.
IV.
MAIN
LINAC
The main linac challenge is in high current emittance
preservation and stabilization of both the
e-
and
e+
bunches in
the presence of the inevitable quadrupole and accelerating
structure misalignments. The requirements for vertical linac
emittance control have become even more challenging with the
advent of flat beam operation in 1993 where the linac entrance
emittance at
1.2
GeV
is
now:
y~~
=
2-3 mm-mrad,
=
30-
40
mm-mrad
[13].
Beam-based alignment techniques have been
used successfully in the past to control transverse quadrupole
alignment
to
-80
pm
rms [14] and new ideas are under
investigation
to
align the disk-loaded wave guides using beam
generated dipole wakefields of the accelerating structures
as
an
error signal
[15].
Under normal operation, empirical linac
emittance correction
is
accomplished by introducing feedback
controlled trajectory oscillations
[
161 to minimize the
measured emittance of wire-scanner phase space monitors [17]
2
DISCLAIMER
Portions
of
this document may be illegible
in electronic image products. Images
are
produced
from
the best available original
document.
I
or by observing a set of four off-axis screens
[18].
Emittance
dilution in
the
main linac
is
usually controllable to
<GO%
vertically and <30% horizontally at 3.5~10~~ ppb. However,
temperature dependencies in the linac
rf
system can generate
day to night emittance variations which require constant
tuning. Improvements are presently under investigation [19].
A
second challenge is pulse-to-pulse and long term
trajectory stabilization
of
both
the
e-
and
e+
beams [20].
Trajectory jitter not only degrades luminosity but also
complicates and slows tuning schemes which rely on phase
space monitors requiring many tens or hundreds of pulses.
A
large source of
e-
trajectory jitter, identified in 1994, was due
to long range tr,msverse wakefields. With equal
e+
and
e-
linac
betatron phase advance, the jitter in the leading
e+
bunch is
resonantly amplified to
the
trailing
e-
bunch. By introducing a
vertical e+ betatron oscillation initiated in the positron bunch
compressor, the trailing electron bunch is seen
to
accumulate
an
oscillation due
to
the long r,ange walrefield [21].
This problem was significantly diminished by introducing
a lO"/cell separation between
the
horizontal and vertical
betatron tunes within
the
linac.
Thus
the resonant condition is
avoided.
This
linac lattice modification successfully reduced
e-
vertical trajectory jitter from
-GO%
of
the
nominal
rms
beam
size (observed in the
final
focus)
to
-40%.
Some improvement
is also observed in
e-
horizontal jitter. Fig.
3
shows the
initiated
e+
oscillation and its
wc&e
induced
e-
oscillation both
with and without a split tune lattice.
1.01
e+
..
0
-1
.o
1
f
b)
,-,
-
0.30
E
E
-0
>
-0.30
-0.30
1
4.95
79W1
Fig
3.
Vertical
e+
oscillation introduced before the linac (a) and
long range wakefield induced oscillation for
c-
beam
of -300
pin
before (b) and
-100
pm
after (c) installation of split tune lattice.
However, with
the
large mid-linac energy spread introduced
for BNS damping [22] and the new 'split-tune' quadrupole
settings, some increased chromatic emittmce dilution within
the linac is expected (-10%). Efforts are underway
to
develop a
split-tune linac lattice with less chromatic dilution.
In light of previous successes [23], further efforts to
stabilize linac trajectories have centered around modifications
of quadrupole magnet support structures. Measurements of
quadrupole magnet vibrations using
a
geophone indicate
-300nm
rms
vibrations for frequencies above
1
Hz
[24].
Beam response modeling in these conditions predict mjectory
jitter which is -20% of
the
10-SO
pm
nominal vertical be,m
size. Inspection of the
supports
has revealed
a
poorly supported
degree
of
freedom in magnet pitch angle which translates into
a
significant vertical displacement component due to the
longitudinally biased pitch rotation axis
of
the support. In
response, magnet pitch wedges were installed for -2/3 of the
linac quadrupoles.
V.
ARCS
AND
FINAL
Focus
SYSTEMS
Prior to the 1984 run the optics
of
both final focus
systems
(FFS)-e-
&
e+-were upgraded in order to allow
reduction of the
IP
vertical beta function [25]. One new
quadrupole magnet per
FFS
was installed between the
chromatic correction section
(CCS)
and the
final
triplet. This
quadrupole optimally adjusts the betatron phase advance
between
CCS
sextupoles and triplet to reduce
the
dominant 3rd
order aberration
(U3466
coefficient in
TRANSPORT
notation
[26]). In addition, two more quadrupoles--one skew and one
normal-were added
to
the upper transformer section
(UT)
to
provide a
full
compliment of orthogonal tuning 'knobs' for
control of IP
beta
functions; cross-plane coupling and IP beam
waist
positions [27]. Four new wire-scanners per
FFS
were
added for emittance and matching diagnostics within the
FFS
and
a
fifth wire-scanner was installed at
an
IP image point
in
the Center of the first
CCS
bend magnet [28].
The
new
final
focus
beamlines were commissioned in April
and May of 1994 using previously established techniques such
as quadrupole and sextupole beam-based alignment methods
[29-311. The new orthogonal
UT
tuning knobs and image
point wire-scanners were employed very successfully to
achieve
the
desired IP beta functions, coupling correction and
waist positions. Subsequent low current beam collisions
(0.5-
1.0~10'~ ppb) using
a
twice nominal
e-
damping ring store
time to achieve ideal emitt'mces produced vertical IP rms spot
sizes of 400 nm, clearly confirming the expected peFfomce
of the upgrade. The horizontal spot
sizes
observed were also
within the expected value of 1.8-2.0
p.
Fig. 4 shows
a
413 nm vertical beam-beam deflection
scan
[32] measured at
low current and long damping ring store.
Y
I-
I
i
-8
-4
0
4
8
3-95
Aye+
(Pm)
79Oul
Fig
4.
Beam-beam deflection,
8,
vs. separation,
Ay,
fitted with
Bassetti-Erskine formula showing
413
nm vertical rms beam size.
At
higher beam intensities
a
significant emittance dilution
within the arc/FF systems appears which is not yet
3
understood. The dilution is usually most evident for the
electrons in the vertical plane. At 3.5~10~~ ppb the observed
luminosity is typically 20-40% lower than expectations based
on emittance measurements at the end of the main linac.
Numerous machine studies have addressed this discrepancy
[3].
One probable explanation for high current emittance dilution is
collimator generated wakefields. The collimators are used to
attenuate detector backgrounds by clipping thinly populated
beam
tails
and are downstream of the end-of-linac wire-scanners
used for emittance measurements. Recent studies [33] show
clear emittance dilution for some collimators which are
routinely closed to within
-3-5
times the
nns
beam size.
However, these measurements have not been reproducible and
they
are
sensitive to varying orbits
‘and
beam conditions in the
collimator re,’ wns.
The addition of new wire-scanners at the entrance to the
FFS
allowed
a
first direct observation of the emittance at the
end of the 1.2
km,
terrain-following collider arcs.
At
the start
of the run the emittance increase through the arcs was found to
be independent of both beam current and initial emittance. The
vertical increase was
Ay&,,
=
3-4
mm-mrad while
the
horizontal was
AYE,
=
10-12 mm-mrad, both of which are in
fair agreement with
the
expected effect of synchrotron radiation
and cross-plane coupling calculated from measured betatron
oscillation data [34]. Without imperfections the vertical
emittance increase in the arcs is expected to be
-1
mm-mrad.
However, toward the end of the
run,
the arc emittance increase
showed some sensitivity to beam current, especially for
electrons in
the
vertical plane.
It
is
not yet hiown
if
this
apparent change was due to
a
slow degradation of the orbits and
optics of the arcs or if it was related to collimator generated
wakefield.. Careful experiments designed to study current
dependencies in the arcs were attempted but are difficult to
perform satisfactorily due to problems controlling main linac
emittances at varying cumnts.
Spin transport through the
SLC
continues to be controlled
with vertical orbit ‘bumps’ in the
e-
(north) collider arc [35].
The two post-damping ring spin rotator solenoids have
remained switched off. Depolarization in the arcs due to initial
energy sprezl has been reduced in comparison to tlie previous
year by a vertical arc orbit variation method which empirically
reduces the effective number of spin precessions through the
arc from -17 ‘turns’ (full precessions) in 1993 to -10 turns
in
1994
[36].
This improvement, in conjunction with the
reduction in energy spread using over-compression, has reduced
the arc depolruization from
-3%
in 1993 to <l% in 1994-95.
PARAMETER
Intensity
repetition rate
hor. emittance
VI.
FEEDBACK, CONTROLS AND
DIAGNOSTICS
There are approximately
28
different microprocessor
controlled
fast
trajectory feedback loops, as well as several
special function loops, in simultaneous operation around the
SLC [37]. These loops maintain beam trajectories and energies
over
a
broad band of frequencies up to -10 Hz. Beam position
monitors are used to measure trajectory variations around
a
previously determined reference orbit and corrections are
applied with fast dipole correction magnets
or
multiple
UNITS DESIGN 1995
lxlOIO
ppb 7.2 3.5
Hz
180 120
mm-md
30
60
klystron phases in the case of energy corrections. There are
seven loops in the main linac which control both
e-
and
e+
orbits. These loops are ‘cascaded’ through
a
communication
link
so
that loop n+l nominally corrects only trajectory
disturb‘ances incurred after loop n. Furthermore, the loops are
‘adaptive’ meaning they are able to learn the transport map-
the accelerator transfer coefficients-between loops. An added
benefit of adaptive-cascaded feedback is the continual
measurement of the phase advance between points in the
accelerator. This information is recorded every six minutes and
cLan be used to trace and isolate optical errors such as errantly
back-phased klystrons. Efforts continue to improve feedback
performance through step response testing and modeling [38].
The beginnings of significant progress in machine wide
emittance control can be traced to the development and
installation of beam profile wire-scanners in the main linac in
1990-91 [17]. Transverse emittance measurements for both
becams are now automatically made at three different points in
the main linac during colliding beam operations once every
hour. In addition, operator initiated measurements are used to
direct tuning efforts when necessary. There are now
-50
different wire-scanners in use throughout the
SLC
from the
40MeV electron injector to near the final triplet. Most of
these wire-scanners are able
to
measure beam sizes in both
planes as well as the
x-y
correlation. Extensive software
controls have been developed which analyze the beam profile
data
collected and return parameters such
as
emittance, beta-
functions, magnitude and phase of mismatch, coupling
magnitudes and beam tails. These parameters, along with raw
beam size, are available in history plots for any time interval
during the
run.
VII.
PRESENT PERFORMANCE LIMJTATIONS
AND
FUTURE
PLANS
Table 2 below summarizes 1995 peak operating parameters
with respect to the original 1985 ‘design’ expectations. The
design expectations are unrealistic, especially in their
underestimation of linac wakefield emittance dilution at beam
intensities of >7x1010 ppb.
Table
2.
Design and peak
1995
parameters most disparate.
The
intensity difference accounts for
a
factor of
-8
in
luminosity.
mm-mrad
Pm
urn
30 10
1.65 2.1
1.65 0.7
Luminosity
per hour
e-
nolarization
I
enercv snread
I
%
I
0.25
I
0.12
I
1030 cm-2s-1
6.0
0.8
hi1
650
80
80
-
o/o
~ ~~
Enhancement
I
-
I
2.2
I
-1.15
4