THE LATE NGLS:
OVERVIEW OF LINAC DESIGN,
BEAM DYNAMICS
Marco Venturini
LBNL
Sept. 26, 2013
1 ─ M. Venturini, Sept. 26, 2013, SLAC
Outline
Guiding principles for choice of main parameters, lattice design, bunch
compression
 RF vs. magnetic compression
 Single vs. multiple stage magnetic compression
Description of layout, lattice, working point baseline
Preservation of beam quality and beam dynamics issues (single bunch)
 Longitudinal dynamics
 CSR-induced emittance growth
 The microbunching instability
 Transverse space-charge effects in the low-energy section of the linac
Impact of availability of passive de-chirping insertion on machine design
 Lowering degree of RF (velocity bunching) compression
2 ─ M. Venturini, Sept. 26, 2013, SLAC
Requirements informing choice
of linac design
All bunches exiting the linac have same design characteristics,
are adequate to feed any of the FEL beamlines (1keV photon /energy)
 Different kinds of beam tailored to specific FEL beamlines are a speculative
possibility. Not investigated yet.
As high as possible peak current consistent with:





Flat current profile
Flat energy profile
Minimal degradation of transverse emittance (both slice and projected)
Sufficiently small energy spread
Sufficiently long bunches to support two- (three-?) stage HGHG external-laser
seeding
2.4 GeV beam energy
Q=300 pC/ bunch
3 ─ M. Venturini, Sept. 26, 2013, SLAC
RF vs. magnetic compression
At cathode of proposed gun bunch current is very low I ~ 5-6A
 Substantial compression is needed
Magnetic compression:
 Energy chirp at exit of last compressor
 CSR effects
 Low-frequency SC RF structures would be needed for acceptance
of very long initial bunches
RF compression in injector (velocity bunching):
 Less than ideal current profile
 Space-charge effects, emittance compensation
Adopted approach: do both RF and magnetic compression
 Right balance depends on various factors (e.g. how much chirp can be
removed after compression)
 RF compression to 40-50A range has shown overall best results
4 ─ M. Venturini, Sept. 26, 2013, SLAC
Single vs. multiple stage magnetic
compression
Overall magnetic compression ~ 10 or higher.
One-stage compression:
 Minimizes microbunching instability
Two-stage compression:
 More favorable to preservation of transverse emittance
 Better beam stability
Three-stage compression
 Adds complexity; may aggravate microbunching instability
Adopted approach:
 Two-stage compression with flexibility for single-stage compression
(disabling second chicane).
5 ─ M. Venturini, Sept. 26, 2013, SLAC
Machine layout,
highlights of linac settings
Linearizer off
Magnetic compressors are conventional
C-shaped chicanes
 BC1 @ 215MeV (Sufficiently high to reduce
CSR effects on transverse emittance)
 BC2 @ 720MeV (There may be room for
optimizing beam energy)
Potential harm from large angle (36 deg)
between linac axis and FELs (CSR)
6 ─ M. Venturini, Sept. 26, 2013, SLAC
Large dephasing
to remove
energy chirp
Baseline beam out of injector
(used in Elegant simulations of linac)
Out of injector beam
(ASTRA simulations)
flat core 
curvature 
head
Physics model in Elegant
simulations (next 4 slides)
includes:
• 2nd order transverse dynamics
• Ideal (error free) lattice
• Longitudinal RF wakefields
(using available models for TESLA
cavities)
Ipk~45 A
Not included:
• LSC, RW wakes, transverse RF
wakes
head
slice e┴ ≤0.6 mm
proj. e┴ =0.72 mm
• CSR
relatively long tail
is a signature of velocity
compression 
7 ─ M. Venturini, Sept. 26, 2013, SLAC
Elegant tracking:
Longitudinal dynamics through BCs
BC1 exit (factor ~2 compression)
head
BC2 exit (~5 compression)
Curvature of energy profile,
to cause current spikes,
harm radiation coherence
if we compressed much more
Ipk~90 A
Ipk~500 A
head
flat current profile as desired 
(current not very high but adequate)
8 ─ M. Venturini, Sept. 26, 2013, SLAC
substantial portion
of bunch is in the tail 
Elegant tracking: Longitudinal dynamics
through linac and Spreader
Exit of linac
head
Entrance to FEL beamlines
CSR long. wake in
spreader helps
somewhat with
energy chirp removal
head
Energy profile relatively flat
within beam core 
Note: tracking done through
fast-kicker based spreader
9 ─ M. Venturini, Sept. 26, 2013, SLAC
Flat core is
> 300fs long
Projected emittances through spreader
Careful lattice design keeps
projected emittance almost
unchanged by the exit of
spreader (<0.8mm)
(two-stage compression)
horizontal
horizontal
vertical
vertical
x/z and x’/z sections
Slice* x-emittance (exit of spreader)
head
= .
head
10 ─ M. Venturini, Sept. 26, 2013, SLAC
*slice is 5mm
10
10
Aside on setting of linearizer
wakefields (RF, CSR)
generate energy chirp
w/ positive quadratic term within bunch
Exit of BC1
Turning on linearizer
would add to positive
quadratic chirp, pushing beam
tail forward upon compression,
and causing current spike*
Exit of BC2
Exit of linac
head
head
Elegant simulations for baseline working point; linearizer off
11 ─ M. Venturini, Sept. 26, 2013, SLAC
*Details depend on machine settings
One-stage compression causes 25%
growth of projected emittance
Longitudinal phase space is comparable
to that of 2-stage compression
BC1 at beam energy ~ 250MeV;
 BC2 off
Linearizer on (20MV), decelerating mode
Reduced dephasing of L3S (20 deg)
head
Exit of spreader
Projected emittances through spreader
horizontal
head
angle rad
R56 m
BC1
0.0955
0.0856392
BC2Voltage0. MV phase
Indeterminate
deg
E MV Acc. Grad MV m
L1
201.6
30.
174.591
13.8763
L2
549.998
0.
549.998
12.619
HL
20.
180
20.
8.25971
L3
1706.67
20.
1603.75
13.0524
12 ─ M. Venturini, Sept. 26, 2013, SLAC
vertical
no. modls
2
6
1
18
Transverse space
charge effects in
low-energy section of
linac
Space charge affects:
, −matching
y
x
w/o space charge
(solid lines)
 some effects in section between
Laser Heater (~95MeV) and BC1
(~210MeV)
 IMPACT simulations (Ji Qiang)
 Emittance growth not large (~10%)
 but a portion of it is slice rather than
with space charge
(dashed)
Exit of injector
Entrance of L1
emittances
with space charge
(dashed) y
projected emittance growth.
 Possible remedy: Increase beam
x
energy at exit of injector
 2 vs 1 cryomodules?
E=94 MeV
13 ─ M. Venturini, Sept. 26, 2013, SLAC
w/o space charge
The microbunching instability can damage
the longitudinal phase space
Linear gain for 2-stage compression
Seeded by shot noise and perturbations at the source
(e.g. non-uniformity in photo-gun laser pulse)
Consequences
 Slice energy spread (penalty on lasing efficiency)
 Slice average energy (penalty on radiation spectral purity,
in particular in externally seeded FELs beamlines)
Modeling primarily by IMPACT; simulations w/ multibillion macroparticles to minimize numerical noise.
Current profile
Longitudinal phase space
2-stage
compression
head
10keV
10keV
5keV
not Venturini,
fully compressed Sept. 26, 2013, SLAC
14Note:─beamM.
Compare various
degree of
heating (rms)
5keV
head
Microbunching seeded by shot noise
Two-stage compression:
 Slice energy spread is minimum for sE=15keV
Slice energy along bunch
heating
 Variations of slice energy are on the order of
the energy spread (~200keV). Too big?
sE =15keV
heating
One-stage compression:
 Instability is effectively suppressed for
sE=10keV heating
IMPACT simulations
final
E
keV
Slice* energy spread vs. Heater setting
450
400
350
300
250
200
150
100
DE~200keV
Two-stage
compression
5
10
15
20
Laser Heater keV
E
15 ─ M. Venturini, Sept. 26, 2013, SLAC
*Slice is 1mm ~ coop length
Microbunching seeded by sinusoidal
current perturbation at cathode (I)
Initial current profile w/ perturbation
Amplification of modulation
depends strongly on period
of perturbation
Current profiles at exit of linac (Two-stage compression)
5% perturbation, 3.4ps period
5% perturbation, 0.8ps period
head
16IMPACT
─ M.simulations
Venturini, Sept.z26,
2013, SLAC
(mm)
head
z (mm)
Microbunching seeded by sinusoidal current
perturbation at cathode (II)
Slice energy along core of bunch
(exit of Spreader)
5% amplitude
perturbation on current at
cathode
 0.8 ps period
sE=15keV heating
Energy profile for two-stage
compression shows
~200keV ripple
(comparable to instability
Seeded by shot noise)
Energy profile for one-stage
compression remains
Relatively smooth
IMPACT simulations
17 ─ M. Venturini, Sept. 26, 2013, SLAC
17
Specs for Heater with sE~15keV heating
power are not too demanding
PM Undulator gap vs. e-beam energy @LH
~0.16 MW laser peak power
 for ~15keV rms energy spread
~2.2 /laser pulse
~2.2 W laser average power @1MHz
(at LH undulator)
Dedicated laser system
Commercially existing, high-repetition rate, shortpulse, high-power laser
Laser peak power* requirement for sE=12keV
Accurate simulation
of 3D laser-beam
interaction w/ collective
forces (“trickle” effect)
still missing.
18 ─ M. Venturini, Sept. 26, 2013, SLAC
*Neglecting diffraction effects
lu
5.4 cm
lL
1.064 mm
Eb
94 MeV
s┴
160 mm
How could availability of passive “dechirping”
insertions affect the linac design?
Longitudinal Phase Space
L3 on crest
add 5-m long
de-chirper
(r = 3 mm)
…or 35-deg off
crest
P.Emma
Save on no. of cryomodules in last linac section (or allow for
higher beam energy)
1.

5m long, r=3mm corrugated pipe would do the dechirping job (L3S on crest)
Allow for compression through the spreader lines (a bit far
fetched…)
2.

Different FEL lines with differently compressed bunches
Increase amount of magnetic compression relative to RF
compression as a way to increase beam quality
3.

Deliver beams with more compact current profile and possibly higher peak current
19 ─ M. Venturini, Sept. 26, 2013, SLAC
19
Tracking the origin of the long bunch tail:
longitudinal dynamics in the injector
(kinetic E)
Fig. from C. Papadopoulos
Current profile
Energy
profile
s 1.2mm
E MeV
0.58
0.56
0.54
head
head
0.52
0
1
2
z mm
3
20 ─ M. Venturini, Sept. 26, 2013, SLAC
4
20
A walk down the injector (1):
half-way through the gun
Fig. from C. Papadopoulos
E MeV
Energy
profile
s 2cm
1.00
0.95
0.90
0.85
0.80
0.75
Current profile
head
head
14 16 18 20 22 24 26
z mm
21 ─ M. Venturini, Sept. 26, 2013, SLAC
21
A walk down the injector (2):
past the exit of the gun
Fig. from C. Papadopoulos
E MeV
s 7cm
Energy
profile
22 ─
Current profile
1.255
1.250
1.245
1.240
1.235
head
Space-charge
65
70
75induced
z mm
M. Venturini, Sept. 26, 2013, SLAC energy chirp
head
22
A walk down the injector (3):
right before the buncher
Fig. from C. Papadopoulos
s 70cm
Energy
profile
Current profile
E MeV
1.26
1.25
1.24
head
head
1.23
695
700
z mm
705
23 ─ M. Venturini, Sept. 26, 2013, SLAC
23
A walking down the injector (4):
right after the buncher
Fig. from C. Papadopoulos
Energy
profile
s 1m
Current profile
E MeV
1.36
24 ─
1.34
1.32
energy chirp
1.30
imparted by
buncher (@
995
about zero-crossing)
head
head
1000 1005
M. Venturini, Sept. 26,
z 2013,
mm SLAC
24
A walking down the injector (5):
ballistic compression begins
Fig. from C. Papadopoulos
Energy
profile
s 1.4m
Current profile
1.37
1.36
E MeV
1.35
1.34
1.33
1.32
1.31
head
head
1.30
1392 1394 1396 1398 1400 1402 1404 1406
z mm
25 ─ M. Venturini, Sept. 26, 2013, SLAC
25
A walking down the injector (6):
a tail in current profile develops
Long tail is associated
with 2nd order chirp
Fig. from C. Papadopoulos
E MeV
Energy
profile
s 2.19m
26 ─
1.350
1.345
1.340
1.335
1.330
1.325
1.320
1.315
Current profile
head
head
2186 2188 2190 2192
M. Venturini, Sept. 26, z2013,
mmSLAC
26
With moderate ( ≃ 2)
RF compression,
beam is close to parabolic.
A 650MHz booster
for the APEX injector?
Option of very low RF compression
 enabled by availability of passive dechirpers (we could
afford making more magnetic compression)
~10A peak current, ~1.2cm FW bunch length (300pC)
Bunches are too long for a 3.9GHz linearizer
 choose 1.3GHz rf frequency for the linearizer (same as in
Linac structures)
 injector booster at 650MHz
Snap-shot of NGLS baseline beam @0.4m
downstream the buncher (IMPACT
simulations)
Long. phase space at exit of linac
Possible layout for injector, first linac Section.
27 ─ M. Venturini, Sept. 26, 2013, SLAC
27
Passive insertion used for dechirping
(Very) preliminary study using LiTrack and parabolic model
of beam
 layout with three magnetic BCs (BC1 functionally replacing
most of the RF compression in the injector)
 simulations show improvement in longitudinal phase space
 transverse emittance could suffer from low-energy
compression
Conclusions
Delivered beam meets FEL design requirements
 I=500A flat current profile over about 300fs core

Relatively long tail is harmless but wastes a good fraction of charge
 Relatively flat energy profile in core

Nonlinear energy chirp in the beam tail
 ex=0.6 mm (slice) preserved; ex=0.8 mm projected (two-stage compression)
 ex=1 mm (projected) for 1-stage compression
CSR in spreader not harmful at this current
 CSR longitudinal wake helps with energy chirp removal from beam core (but adds some
nonlinearity on energy chirp)
The microbunching instability seeded by shot noise is effectively suppressed by
heating to sE = 10keV in one-stage compression mode
 In two-stage compression, heating to sE = 15keV yields ~150 keV final slice rms energy
spread (acceptable) but also slice average energy variations of the same magnitude.
 Beam current at cathode should be smooth within a few %’s, or much less depending on
spectral content of noise
Availability of reliable dechirper-insertion would open up interesting possibilities
 Reduce RF compression for better beam quality.
28 ─ M. Venturini, Sept. 26, 2013, SLAC