yujong.kimPSI.ch, http:www.PSI.chkim_y, http:FEL.WEB.PSI.ch

1
31st International FEL Conference, Liverpool,
August 24, 2009
Critical Issues in the Coherent Single Spike
Mode Operation with Low Charges
Yujong Kim, H. Braun, T. Garvey, M.
Pedrozzi J.-Y. Raguin, S. Reiche, T. Schilcher,
and V. Schlott PSI, CH-5232 Villigen PSI,
Switzerland
PSI XFEL-2009-51
yujong.kim_at_PSI.ch, http//www.PSI.ch/kim_y,
http//FEL.WEB.PSI.ch
2
Outline

• Acknowledgments
• Introduction to the SwissFEL Project
• Limited Site and Operation Modes
• Optimized Injector for Various Single Bunch
  Charges
• Single Spike Mode Operation with 2 pC for XFEL _at_
  0.1 nm
• Nominal and Single Spike Mode Operations with 10
  pC for XFEL _at_ 0.7 nm
• Critical Issues in Single Spike Mode Operation
  with Low Charges
• Tight RF Jitter Tolerances
• Random CSR Kicking and Orbit Fluctuation
• Wavelength or Bunch Length, Charge, Beam
  Diagnostics
• Energy Chirp and Wide Bandwidth
• Summary

3
Acknowledgements
Y. Kim sincerely give thank to following
persons - K. Floettmann of DESY - M. Borland
of ANL - Y. Ding, J. Frisch, D. Dowell, and P.
Emma of SLAC - J.B. Rosenzweig and C. Pellegrini
of UCLA for valuable discussions, helpful
information, and encouragements on this work!
3
3
4
SwissFEL - Limited Site
2008-2014 250 MeV Injector Test Facility -
Commissioning will be started in October,
2009 2012-2016 RF Gun Short C-band Linac
In-Vacuum Undulator based 5.8 GeV SwissFEL
Facility initial (2016-2020) operation modes
100 Hz with two micro-bunches (0.1 nm to 7
nm) upgrade (after 2020) 400 Hz with three
bunches longitudinal single spike mode (0.7 nm
to 7 nm)
total facility length lt 800 m undulator length lt
70 m total available length for linac lt 530 m
See details from our talks and posters MOPC05
seeding MOPC06 EEHG seeding _at_ 250 MeV
injector MOPC07 low charge operation MOPC36
projected emittance measurement method MOPC62
transverse beam size effects in OTR
spectrum MOPC63 gun laser MOPC64 chirp and
bandwidth control with C-band linac MOPC65 THz
pump source TUPC35 commissioning results of Low
Emittance Gun phase-II (LEG-II) TUPC36 advanced
beam diagnostic section TUPC37 design concepts
of compact SwissFEL linac TUPC38 construction
status and overview of 250 MeV injector test
facility WEPC58 tolerance study hard X-ray
beamline THOA01 undulators for the SwissFEL
PSI - West
available site length ? 950 m
LEG Test Facility
SwissFEL Facility
4
5
SwissFEL - Required Parameters Modes
To get saturation at 1 Å with 5.8 GeV (or 7 Å
with 3.4 GeV) with 50 m undulator.
assumed in-vacuum (NdFeB with diffused dy)
undulator parameters K 1.2, ?u 15 mm, ltßgt
15 m, total length 70 m for 1 Å
6
SwissFEL - 250 MeV Injector Test Facility
6
6
6
7
SwissFEL - Summary of Optimized Injector
ASTRA Simulation Results CTF3 RF Gun based
Injector for various Charges
Final beam parameters are at the exit of the 2nd
S-band structure (130 MeV - 172 MeV). Gun max
gradient 100 MV/m, assumed Kave 0.4 eV. For
only Q ? 2 pC, eprojected ? eslice ? ethermal
For a much higher Q, eprojected ? eslice ??
ethermal due to the nonlinear space charge force.
We can get an excellent emittance with a lower
charge for the single spike mode operation!
7
7
7
8
Full Coherent Single Spike Mode
Re-studied - J.B. Rosenzweig et al., NIMA 593
(2008) 39
temporal profile of FEL photon for 3 different
bunch lengths
Shot-to-shot fluctuation in saturation length due
to shot noise startup. Here they assumed that
there is no error or jitter in linac.
?z ? 0.3 fs for 1 Å ?z ? 2 fs for 7 Å
for the SwissFEL project due to many
difficulties (diagnostics, RF tolerances), we
chose 7 Å and higher charge of 10 pC.
8
8
8
9
Single Spike with 2 pC - Dancing Bunch
For 0.1 nm with 2 pC, if tolerances is loose,
operation becomes really unstable
gun timing error 10 fs (rms) bunch charge
error 0.5 (rms) injector S-band RF phase
error 0.02 deg (rms) injector S-band RF
voltage error 0.02 (rms) injector X-band RF
phase error 0.04 deg (rms) injector X-band RF
voltage error 0.04 (rms) BC1 BC2 dipole
power supply error 10.0 ppm (rms) LINAC1
S-band RF phase error per klystron 0.02 deg
(rms) LINAC1 S-band RF voltage error per
klystron 0.02 (rms) LINAC2 S-band RF phase
error per klystron 0.02 deg (rms) LINAC2 S-band
RF voltage error per klystron 0.02 (rms)
very dynamic peak current slice parameters!
Loose Tolerances for 2 pC
under loose tolerances for 2 pC, we performed 300
times start-to-end simulations to see the
fluctuation of the longitudinal phase space at
the entrance of undulator. beam chirp and
position are very dynamically dancing ! under
this situation, peak current, beam arrival time,
and slice beam parameters are also very
dynamically dancing, which induce unstable XFEL
photon beams.
100 fs
9
9
9
9
9
9
10
Single Spike with 2 pC - More Stabilized
For 0.1 nm with 2 pC, we need much more tighter
tolerances for stable operation
gun timing error 1 fs (rms) bunch charge
error 0.5 (rms) injector S-band RF phase
error 0.001 deg (rms) injector S-band RF
voltage error 0.001 (rms) injector X-band RF
phase error 0.004 deg (rms) injector X-band RF
voltage error 0.004 (rms) BC1 BC2 dipole
power supply error 1.0 ppm (rms) LINAC1 S-band
RF phase error per klystron 0.001 deg
(rms) LINAC1 S-band RF voltage error per
klystron 0.001 (rms) LINAC2 S-band RF phase
error per klystron 0.001 deg (rms) LINAC2
S-band RF voltage error per klystron 0.001
(rms)
Ultra-Tight Tolerances for 2 pC
under tighter tolerances for 2 pC, we performed
300 times start-to-end simulations to see the
fluctuation of the longitudinal phase space at
the entrance of undulator. beam position dancing
was reduced and chirp dancing was dramatically
damped. under this situation, peak current and
slice parameters are almost constant. Therefore,
XFEL power becomes more stable. But there is some
small fluctuation in beam arrival time.
40 fs
10
10
10
10
10
10
10
11
SwissFEL - Layouts for 200 pC and 2 pC
ASTRA up to exit of SB02 ELEGANT from exit of
SB02 to consider space chare, CSR, ISR, and
wakefields !
Optimization-XIV with C-band RF Linacs for
Effective Chirp Control 400 Hz
Optimization-VIII with New Injector for 2 pC
Single Spike Mode
11
11
12
SwissFEL - Two Layouts for 10 pC
New Injector Layout with Laser Heater BC1 _at_
450 MeV
Optimization-X with New Injector for 10 pC
Nominal Mode
Optimization-XI with New Injector for 10 pC
Single Spike Mode
12
13
SwissFEL - Operation Modes with 10 pC
Even Q and Ipk are different, saturation length
is shorter than 50 m for (0.7 nm to 7 nm)!
Nominal Mode
FEL photon beam for Optimization-X rms photon
pulse length 2.27 fs wavelength 0.7 nm with K
1.05 rms bandwidth 0.08 for 0.7 kA pulse
energy 6 µJ no of photon per pulse
2.01010 saturation length 49.7 m for 0.7 kA
electron beam for Optimization-X nominal
mode beam energy 3.4 GeV Q 10 pC peak current
0.7 kA rms electron bunch length 6.2 fs
total compression factor 240
13
13
14
Nominal Mode with 10 pC - Requirements
Note that the total S-band and X-band klystrons
in injector linac are six and two,
respectively. Here we assumed that orbit feedback
system is working in estimating RF tolerances.
14
14
14
14
14
14
15
Single Spike with 10 pC - Requirements
For stable coherent single spike mode (0.7 nm to
7 nm) with 10 pC, 3.4 GeV beam.
parameters
tolerance (rms) tolerance source
gun laser arrival timing error 1 fs
(rms) saturation length, arrival time single
bunch charge error 1 (rms) saturation
power injector S-band RF phase error per
klystron 0.005 deg (rms) power, wavelength,
arrival time injector S-band RF voltage error per
klystron 0.005 (rms) arrival time injector
X-band RF phase error per klystron 0.005 deg
(rms) power, saturation length injector X-band
RF voltage error per klystron 0.025
(rms) arrival time BC1 dipole power supply
error 7.5 ppm (rms) arrival time LINAC1 S-band
RF phase error per klystron 0.015 deg (rms)
wavelength, power LINAC1 S-band RF voltage error
per klystron 0.010 (rms) wavelength, arrival
time BC2 dipole power supply error 7.5 ppm
(rms) arrival time LINAC2 S-band RF phase error
per klystron 0.017 deg (rms) wavelength LINAC2
S-band RF voltage error per klystron 0.011
(rms) wavelength

• beam arrival time error ?Tarrival ? 5 fs
  (zero-to-max) electron bunch length order.
• saturation power error ?Psat/Psat ? 100
  (zero-to-max) against any optics damage.
• wavelength error ??/? ? 0.01 (zero-to-max)
  against intensity lowering due to collimator.
• saturation length error ?Lsat/Lsat ? 15
  (zero-to-max) to get saturation with a given
  undulator length
• (undulator length margin 30-40).
• Note that the total S-band and X-band
  klystrons in injector linac are six and six,
  respectively.
• Here we assumed that orbit feedback system is
  working in estimating RF tolerances.

15
15
15
15
15
16
Single Spike with 10 pC - Performance
median 1.0 µs rms variation 5.5 fs
median 83 GW (80 core slices) rms variation
86 still some fluctuation when we include
all errors together!
from 300 times S2E simulations under RF jitter
tolerances
median 20 m rms variation 4.5
median 1 nm rms variation 0.006
16
16
16
16
16
17
Single Spike with 10 pC - Performance
median 3.4 GeV rms variation 0.003
median 0.69 µm rms variation 6.5 small
but gives a big impact in current due vertical
chirping in single spike mode
from 300 times S2E simulations under RF jitter
tolerances
median 0.109 µm rms variation 3
median 9.1e4 rms variaiton 2.1e-4
17
17
17
17
17
18
Single Spike with 10 pC - CSR Orbit Kicking
Under RF jitter tolerances, random RF jitters
generates random CSR orbit kicking in the
horizontal plan. There is no good way to
compensate it because the CSR orbit kicking is
random. Since its rms orbit fluctuation is larger
than 100 of electron rms beamsize in undulator,
there is a big impact on FEL lasing.
300 S2E simulations with RF Jitter Tolerances
electron rms beam size in undulator 13 µm can
we get stable lasing?, maybe, no.
change error 1 (rms) laser arrival timing
error 20 fs (rms) injector S-band RF phase
error 0.04 deg (rms) injector S-band RF
voltage error 0.04 (rms) injector X-band RF
phase error 0.16 deg (rms) injector X-band RF
voltage error 0.16 (rms) BC power supply
error 10 ppm (rms)
18
18
18
18
18
19
Single Spike with 10 pC - CSR Orbit Kicking
Under RF jitter tolerances, random RF jitters
generates random CSR orbit kicking in the
horizontal plan. There is no good way to
compensate it because the CSR orbit kicking is
random. Since its rms orbit fluctuation is close
to 100 of electron rms beamsize in undulator,
there is some impact on FEL lasing.
300 S2E simulations with Required Tolerances
electron rms beam size in undulator 13 µm can
we get stable lasing?, maybe, some lasing.
change error 1 (rms) laser arrival timing
error 1 fs (rms) injector S-band RF phase
error 0.005 deg (rms) injector S-band RF
voltage error 0.005 (rms) injector X-band RF
phase error 0.005deg (rms) injector X-band RF
voltage error 0.025 (rms) BC power supply
error 7.5 ppm (rms)
19
19
19
19
19
19
20
Nominal Mode with 10 pC - CSR Orbit Kicking
Under same RF jitter tolerances for the single
spike mode with 10 pC, we checked status of CSR
kicking for the nominal mode with 10 pC. Clearly,
its CSR orbit kicking is ignorable during the
nominal mode, and lasing will be OK.
median 2.5 GW (80 core slices) rms variation
5 very stable saturation power!
300 S2E simulations with Required Tolerances
electron rms beam size in undulator 8.5 µm can
we get stable lasing?, certainly, good lasing.
change error 1 (rms) laser arrival timing
error 1 fs (rms) injector S-band RF phase
error 0.005 deg (rms) injector S-band RF
voltage error 0.005 (rms) injector X-band RF
phase error 0.005deg (rms) injector X-band RF
voltage error 0.025 (rms) BC power supply
error 7.5 ppm (rms)
20
20
20
20
20
20
21
Other Issues and Summary
By the help of the excellent emittance and higher
beam brightness, we can use a low charge to
generate the fully coherent single spike at X-ray
region. But to realize the single spike mode, we
have to carefully choose wavelength or bunch
length, and single bunch charge after considering
RF tolerances, CSR orbit kicking, and beam
diagnostics. Bandwidth of the single spike mode
is always wide due to the vertical energy
chirp. We expect that the single spike mode will
be easy at a longer wavelength ( 1 nm) due to is
loose RF tolerance and a longer bunch
length. PSI chose the single spike mode with 10
pC as an upgraded mode and we will continuously
try to find the best wavelength or bunch length
and charge to relax RF tolerances. It seems that
current LCLS 20 pC operation is between our
nominal mode and the single spike mode, which CSR
kicking is controllable. And RF tolerances of
LCLS is somewhat looser than other compact XFEL
projects because there are much more RF stations
in LCLS linac.
21
22

Appendix
22
23
250 MeV Injector Test Facility
23
23
23
23
24
Used Thermal Emittance
LCLS Core Slice Emittance Measurements Q 20
pC, E 135 MeV Method QM scan with LOLA E
135 MeV Q 20 pC Ecathode 115 MV/m ?laser
253 nm ( 4.899 eV) ?Tlaser 3.8 ps (FWHM) Laser
profile pseudo-Gaussian shape sz 1.3
ps Ipeak 5 A
Kave 0.4 eV was used for our old CTF3 RF gun
based injector optimizations! Kave 0.63 eV was
used for our four RF gun based injector
optimization (in this talk)!
24
25
Optimized Injector for 2 pC
CTF3 RF Gun may generate an ultra-low emittance lt
0.06 µm for 2 pC !
25
26
Optimized Injector for 2 pC
for 2 pC, slice projected emittances eth
0.042 µm ensc contribution to eslice is
ignorable !
Projected Emittance along Injector
Slice Emittance at 172 MeV
INSB01-RAC
INSB02-RAC
GUN
eprojected 0.054 µm

eslice 0.044 µm
CTF3 RF Gun may generate an ultra-low emittance lt
0.06 µm for 2 pC !
26
26
27
2 pC - 300 S2E Simulations
For 0.1 nm with 2 pC, if tolerances is loose,
operation becomes really unstable
gun timing error 10 fs (rms) bunch charge
error 0.5 (rms) injector S-band RF phase
error 0.02 deg (rms) injector S-band RF
voltage error 0.02 (rms) injector X-band RF
phase error 0.04 deg (rms) injector X-band RF
voltage error 0.04 (rms) BC1 BC2 dipole
power supply error 10.0 ppm (rms) LINAC1
S-band RF phase error per klystron 0.02 deg
(rms) LINAC1 S-band RF voltage error per
klystron 0.02 (rms) LINAC2 S-band RF phase
error per klystron 0.02 deg (rms) LINAC2 S-band
RF voltage error per klystron 0.02 (rms)
Note that tolerances above are better than
current LCLS situation (phase 0.04 deg, dV/V
0.04) in S-band.
Loose Tolerances for 2 pC
S03 X01 BC1
LINAC1 BC2
LINAC2 5.8 GeV for 0.1
nm with 2 pC
27
27
27
27
27
27
28
2 pC - 300 S2E Simulations
median 1.5 µs rms variation 16 fs
median 13 GW (80 core slices) rms variation
130 with loose tolerances, power
fluctuation is very strong.
poor performance with loose tolerances to
generate single spike at 0.1 nm with 2 pC!
median 19 m rms variation 12
median 0.1 nm rms variation 0.008
28
28
28
28
28
28
28
28
29
2 pC - 300 S2E Simulations
For 0.1 nm with 2 pC, we need much more tighter
tolerances for stable operation
gun timing error 1 fs (rms) bunch charge
error 0.5 (rms) injector S-band RF phase
error 0.001 deg (rms) injector S-band RF
voltage error 0.001 (rms) injector X-band RF
phase error 0.004 deg (rms) injector X-band RF
voltage error 0.004 (rms) BC1 BC2 dipole
power supply error 1.0 ppm (rms) LINAC1 S-band
RF phase error per klystron 0.001 deg
(rms) LINAC1 S-band RF voltage error per
klystron 0.001 (rms) LINAC2 S-band RF phase
error per klystron 0.001 deg (rms) LINAC2
S-band RF voltage error per klystron 0.001
(rms)
Ultra-Tight Tolerances for 2 pC
S03 X01 BC1
LINAC1 BC2
LINAC2 5.8 GeV for 0.1
nm with 2 pC
29
29
29
29
29
29
30
2 pC - 300 S2E Simulations
median 1.5 µs rms variation 4.5 fs
median 13 GW (80 core slices) rms variation
12 with tighter tolerances, fluctuation is
damped.
great performance but ultra-tight tolerances! we
will not use 2 pC to generate single spike !
median 19 m rms variation 3.9
median 0.1 nm rms variation 0.001
30
30
30
30
30
30
30
31
Optimized Injector for 10 pC
ASTRA Simulation Results with Space Charge
CTF3 RF Gun may generate a low emittance lt 0.1 µm
for 10 pC !
After considering targeting slice emittance (
0.18 µm) at the end of linac, we chose thermal
emittance 0.072 µm a low peak current of 3 A
at the cathode.
31
31
31
32
Optimized Injector for 10 pC
Projected Emittance along Injector
Slice Emittance at 167 MeV
INSB01-RAC
INSB02-RAC
GUN
Excellent Slice Emittance ! And Wide Uniform
Range !
Good Invariant Envelope Matching ! Emittance
Damping in Booster Linac !
Mismatching Parameter ? at 167 MeV
Q 10 pC E 167.654 MeV, ?? 0.021 ?x 105
?m, ?y 105 ?m, ?z 316 ?m ?nx 0.095 ?m, ?ny
0.095 ?m Ipeak 3 A, ?n,core,slice 0.078
?m Excellent Projected Slice Emittance
! Excellent Uniformity in Current Profile
! Better Twiss Mismatching Parameter !
32
32
32
33
Efforts on Chirp Bandwidth
From our recent full S2E simulations with ASTRA,
ELEGANT, and GENESIS codes (Y. Kim and S.
Reiche), we confirmed that we can effectively
minimize the bandwidth of XFEL photon beams by
optimizing energy chirping of electron beams.
Optimization-III V S-band based Linacs Linac
Length 650 m Optimization-VI VII C-band
based Linacs Linac Length 540 m, 510 m
Optimization-V chirp for Ipk 2.7 kA
Optimization-III, VI, VII chirp for Ipk 1.6 kA
Saturation Length lt 50 m !!!
wavelength 0.1 nm _at_ FEL1 no of photon per pulse
1.01011 saturation length 40 m with 2.7
kA saturation length 48 m with 1.6 kA
BW 0.05 for Ipk 1.6 kA
BW 0.1 for Ipk 2.7 kA
33
33
34
SwissFEL - RF Development Milestone
year requirements
compression factor operation
conditions 2010 ?s ? 0.1 deg (rms)
1
for injector gun ?V/V ?
0.1 (rms)
with 22 A, 200 pC 2011
?s ? 0.06 deg (rms) 16
for injector BC1
?V/V ? 0.06 (rms)
with 350 A, 200 pC
2012 ?s ? 0.04 deg (rms)
75 for XFEL
BC1BC2 ?V/V ? 0.04 (rms)

with 1.6 kA, 200 pC 2014 ?s ?
0.02 deg (rms) 125
for XFEL BC1BC2
?V/V ? 0.02 (rms)
with 2.7 kA, 200 pC
2016 ?s ? 0.01 deg (rms)
240 for
XFEL BC1BC2 ?V/V ? 0.01
(rms)
with 0.7 kA, 10 pC after 2020 ?s ?
0.005 deg (rms) 2400
for XFEL BC1BC2
?V/V ? 0.005 (rms)
with 7 kA, 10 pC These are
requirements for S-band RF for about 1 minute.
Requirements of X-band are four times of S-band.
34 for LCLS 1 nC case
Nominal Operation Modes
90 for LCLS 250 pC case
Upgrade Mode
34
34
34
34
34
34
35
10 pC - Criteria of Jitter Tolerance

• To determine error tolerances for the single
  spike mode operation with 10 pC, at the
• entrance of undulator (3.4 GeV), following
  FEL performance should be satisfied
• when there is an error in a single machine
  component
• beam arrival time error ?Tarrival ? 5 fs
  (zero-to-max) electron bunch length order.
• saturation power error ?Psat/Psat ? 100
  (zero-to-max) against any optics damage.
• wavelength error ??/? ? 0.01 (zero-to-max)
  against intensity lowering due to collimator.
• saturation length error ?Lsat/Lsat ? 15
  (zero-to-max) to get saturation with a given
• undulator length (undulator length margin
  30-40).
• Please note that in rms ( zero-to-max/3.0), they
  are about
• ?Tarrival ? 1.7 fs (rms), tighter than
  European XFEL (12 fs) due to a shorter bunch.
• ?Psat/Psat ? 33 (rms), looser than European
  XFEL (5)
• ??/? ? 0.003 (rms), tighter than European XFEL
  (0.007) due to smaller photons
• ?Lsat/Lsat ? 5 (rms), looser than European XFEL
  (0.5)

35
35
35
36
Single Spike with 10 pC - Jitter Sensitivity
error source RF phase of S-band linac (S03) in
injector error range -0.06 deg to 0.06 deg
with 10 steps monitoring point at the entrance
of undulator _at_ 3.4 GeV For ??/? ? 0.01
(zero-to-max), S-band RF phase error ? 0.005 deg
(rms). This S-band RF phase tolerance is
challenging !
wavelength change due to 10 S-band phase errors
long. phase space peak current fluctuation due
to errors
wavelength 1 nm energy 3.4 GeV undulator
period 40 mm beta-function 10 m K 1.6
30 fs
injector S-band RF phase errors and LINAC1 RF
phase and voltage errors, and LINAC2 RF phase and
voltage errors are sensitive jitter sources to
photon beam wavelength.
36
36
36
37
Single Spike with 10 pC - Jitter Sensitivity
error source RF voltage of S-band linac (S03)
in injector error range -0.10 to 0.10 with10
steps monitoring point at the entrance of
undulator _at_ 3.4 GeV For ?Tarrival ? 5 fs
(zero-to-max), S-band RF voltage error ? 0.005
(rms). This S-band RF voltage tolerance is
challenging !
arrival time change due to 10 S-band voltage
errors
arrival time fluctuation due to 10 S-band voltage
errors
60 fs
BC1 chicane power supply error, gun timing error,
injector S-band RF voltage and phase errors are
sensitive jitter sources to photon beam arrival
time.
37
37
37