Christoph Steier

SLAC – LCLS-2 accelerator physics meetingOct. 17, 2013

RECAP OF NGLS ELECTRON COLLIMATION DESIGN STUDIES

NGLS Electron Collimation

Motivation(Gun) Dark Current

Collimation System LayoutInjector KickerEnergy CollimationBetatron Collimation

Simulation of Collimation EffectivenessDark CurrentTouschek ScatteringGas Scattering

Collimator HardwareDevelopment PlanSummary

Motivation High duty factor accelerators have main beams with considerable

power (MW in our case) Even small fractional losses can have substantial effects

Demagnetization of permanent magnet undulators Quenches of s/c cavities and s/c undulators Heat damage to vacuum envelope Activation

Collimation system is essential Needs to localize ‘routine’ losses away from sensitive areas Needs to prevent equipment damage in case of malfunction until

MPS stops beam For routine losses, experience elsewhere has shown gun dark

current to be dominant source – our calculations so far confirm this

NGLS Layout

APEX-based e- injector (1 MHz, = 0.6 m)300 pC/bunch (0.3 mA max. current)1.3-GHz CW SRF @ 16 MV/m (27 CM’s)Two bunch compressors + heater (500 A)Beam spreader using RF deflectors (9 FELs)Three (initial) very diverse FEL designsDiagnostics and collimation sections720-kW main beam stops (3)

injector linac spreaderFELs (1-9)

beam stops

compressorse- diagnostics

e- diagnostics

exp. halls

collimation

APEX: Injector Dark current status

Use integrated Fowler -Nordheim formula to fit with instantaneous formula (E=E0*cos(ωt))

Fernando's Dark Current measurements

Resulting current profile of dark current

Longitudinal Phase Space at Injector exit

Transverse Distribution of Dark Current in APEX

Solenoid magnet

Single-cell RF cavity

Multi-cell RF cavity

Laser pulse

Gun

Bun

cher

Dark current “Hotspots”Butterfly shape due to large energy spread

<750 keV,Identical to NGLS

>750 keV,Similaro to NGLS,Not superconducting

FLASH gun for comparison

Dark current losses in injectorSimulated different initial distributions (spots or uniform)

After transport in the injector, about 10% (spots) to 15% (uniform) of the dark currentsurvives.

C. Papadopulos

Electron Collimation

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

R56 = -94 mmsd = 0.44%

BC2720 MeV

R56 = -76 mmsd = 0.48%

GUN0.75 MeV

Heater94 MeV

R56 = -5 mmsd = 0.02%

L0j 0

Ipk = 47 Asz = 0.85 mm

L1j = -20.0°Ipk = 47 A

sz = 0.85 mm

Lhj = 180°

V0 = 0 MV

L2j = -23.2°Ipk = 90 A

sz = 0.44 mm

L3j = +34.8°Ipk = 500 A

sz 0.08 mm

SPRDR2.4 GeVR56 = 0

sd 0.04%300 pC; 2012-04-18 & 2012-07-02

3.9

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

sd = 0.44%

BC2720 MeV

sd = 0.48%GUN0.75 MeV

Heater94 MeV

sd = 0.02%

L0Ipk = 47 A

L1, LhIpk = 47 A

L2j = -23.2°Ipk = 90 A

L3j = +34.8°Ipk = 500 A

SPRDR2.4 GeV

sd 0.04%

3.9

EnergyCollimator1.5 mm

EnergyColl.

15 mm

EnergyColl.

8 mm

EnergyColl.

2.5 mm

DarkCurrentKicker 10

4 -tronColl.’s

10 mm (x)2 mm (y)

Assumed apertures for machine +/- 18 mm radius pipe almost everywhere No restriction (except collimator) in LH, BC1/2, FEL chicanes Undulator chamber +/- 15 mm (x), +/- 3 mm (y)

Motivation for Dark Current Kicker First dispersive place to collimate - laser heater chicane (100

MeV) 15% of 8 A at 100 MeV corresponds to 120 W

Anything not captured there quickly gains more energy towards bunch compressor

FLASH stays below 100 W losses in bunch compressor due to radiological concerns Coordinating with EHS – cost implications for shielding

Some of the other 85% is lost in injector cryo-module XFEL guidance is <0.1W/m to avoid cavity quenches,

simulation shows about 1 W/m for uniform emission case – <0.1 W/m for other more realistic distributions

Gaining factor 10 safety margin necessary – Dark Current Kicker

Dark Current Deflector Dark current produced in every injector RF

bucket (186 MHz) – useful beam only 1 MHz FLASH kicker reduces dark current intensity

by factor of >3 NGLS:

kick main bunches and compensate with DC magnet high repetition rate (1 MHz) and fast rise

and fall times Just after the gun (0.75 MeV).

Reference: ALS camshaft kicker (1.5 MHz, rise/fall times of 20 ns, >70 rad@1.9 GeV)

Simulations: scaled version of ALS kicker could reduce by factor of >10

FLASH: F. Obier

ALS: S. Kwiatkowski

Dark Current Kicker Simulation

Gun

Bun

cherCryomodule

>750 keV, Cold<750 keV, Warm

Plan to collimate in this region

collimatorkicker

Need to collimate kicked beam without scraping main beam: Collimator R=10 mm

ConditionDark current (μA) @ injector exit

15 mm ap. across the inj. (ie doing nothing) 1.228

10 mrad Kicker 0.69510 mrad Kicker + 10 mm coll. 0.34515 mrad kicker + 10 mm coll. 0.057

ALS kicker kicks >70 rad at 1.9 GeV - 180 mrad at 750 keVFactor 2 shorter, factor 3 larger opening, about 30 mrad possible

Dark Current Deflector

Dark current produced in every injector RF bucket (186 MHz) – useful beam only 1 MHz

FLASH kicker reduces dark current intensity by factor of >3 NGLS:

kick main bunches and compensate with DC magnet high repetition rate (1 MHz) and fast rise and fall times Just after the gun (0.75 MeV).

Reference: ALS camshaft kicker (1.5 MHz, rise/fall times of 20 ns, >70 rad@1.9 GeV)

Simulations: scaled version of ALS kicker could reduce by factor of >10

New shape

153 mm

54 mm

21 mm

64 mm

H. Qian, S. de Santis, S. Kwiatkowski

Electron Collimation

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

R56 = -94 mmsd = 0.44%

BC2720 MeV

R56 = -76 mmsd = 0.48%

GUN0.75 MeV

Heater94 MeV

R56 = -5 mmsd = 0.02%

L0j 0

Ipk = 47 Asz = 0.85 mm

L1j = -20.0°Ipk = 47 A

sz = 0.85 mm

Lhj = 180°

V0 = 0 MV

L2j = -23.2°Ipk = 90 A

sz = 0.44 mm

L3j = +34.8°Ipk = 500 A

sz 0.08 mm

SPRDR2.4 GeVR56 = 0

sd 0.04%300 pC; 2012-04-18 & 2012-07-02

3.9

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

sd = 0.44%

BC2720 MeV

sd = 0.48%GUN0.75 MeV

Heater94 MeV

sd = 0.02%

L0Ipk = 47 A

L1, LhIpk = 47 A

L2j = -23.2°Ipk = 90 A

L3j = +34.8°Ipk = 500 A

SPRDR2.4 GeV

sd 0.04%

3.9

EnergyCollimator1.5 mm

EnergyColl.

15 mm

EnergyColl.

8 mm

EnergyColl.

2.5 mm

DarkCurrentKicker 10

4 -tronColl.’s

10 mm (x)2 mm (y)

Assumed apertures for machine +/- 18 mm radius pipe almost everywhere No restriction (except collimator) in LH, BC1/2, FEL chicanes Undulator chamber +/- 15 mm (x), +/- 3 mm (y)

Location of Collimators (LHS, BC1)

• Based on low impedance version of ALS collimators (as well as other places) 50 cm is reasonable length for collimators

• With safety margin for finalized mechanical design (impedance calculation) – desirable to reserve 1 m

• Enough space available in BC1, working to increase space in LH, downstream of undulator

Location of Collimators (BC2, MCS)

• Enough space seems available in BC2• Generous space available in FODO section after main

LINAC (MCS, which is just after L3S)• MCS, SLS collimation section of NGLS design much more

compact than XFEL• No requirement to transport energy chirped bunchtrains• No need for very high beta functions (bunchtrain power)• No separate need for R56 variability, …• Spreader angle and achromats in SLS provide natural place for energy collimation

with secondary showers kept away from undulators

Location of Collimators (SLSx)

• Current simulations are based on spreader lattice from October• Baseline change to RF spreader since then• General achromat layout and space similar – current collimator layout should work – will verify

• Space at first collimator OK, at second one a little tight. • Beta functions at second collimator very small – better spaces later in

arc (need trade-off analysis of required MPS speed vs. secondaries escape rate)

Technical Details of Tracking Started from CDR MAD file (sharepoint)

Translate (automated) with mad2elegant (does not accept matching routines, but bare lattice) Needed to remove all CSR (just turning switch off is not enough) – otherwise dark

current gets lost in first CSR element Translate (automated) with mad2at Added beamline apertures (see before) and collimators to resulting files

Will slowly add all apertures/collimators to baseline MAD files Imported ASTRA distributions (astra2elegant, Matlab)

Need to carefully consider phase matching between different distributions, energy scaling, …

Important to use elegant fiducialization correctly In elegant always need to track two bunches (fiducialization reference + dark

current) Tracked CDR beam (and gaussian approximation of it) to determine collimator settings

No loss of nominal beam (or 6 sigma particles) + 10-20%

Collimator Location + Setting LHEATCOL

|x|<1.5 mm BC1COL

|x|<15 mm BC2COL

|x|<8 mm CXL3ED_1

|x|<10 mm CXL3ED_2

|x|<10 mm CYL3ED_1

|y|<2 mm CYL3ED_2

|y|<2 mm SPREADCOL1

|x|<2.5 mm SPREADCOL2

|x|<5 mm

Tracking Gun Dark Current

Dark Current losses well controlled• Most losses on Laser Heater Collimator• Followed by BC1 and BC2• Remaining losses in warm section around laser heater• Losses in Linac 1 below XFEL quench criterium of 0.1 W/m

• Dark current kicker will help• No losses beyond BC2 (and in undulator)

Trajectories, Loss Power

Power densities [W/m] on right are for 8 A dark current from gun:• 10-100 W on collimators

• Likely need for reduction (deflector)• Up to 1 W/m around laser heater

• Would like to reduce• 10s mW/m in Linac1

• Tesla used threshold 10 mJ/cm3 over 20 ms for 25 MeV/m – extrapolating their shower calculations this is safe by factor of >10

Removing collimators (start to end)

• When removing collimators earlier in accelerator, undulators remain protected from dark current (until very last energy collimator is pulled)

• Of course, Linac does not and loss power gets much higher (because collimation does not occur at lowest possible energy)

• Encouraging with regards to protection from Touschek+Gas Scattering in Linac+Spreader

Post Linac Collimation (Gas Scattering)

Test of post linac collimation by artificially increasing (20-50x) divergence of beam at points along the linac

In vertical plane, combination of two (90 degree apart) collimators and energy collimators protects undulators

Rough estimate of pressure requirements on next slide, plan to quantify further with monte carlo and tracking of scattered particles

Estimate of gas scattering loss rates• For electrons one can simplify the formulas for gas Bremsstrahlung

lifetime (in the approximation of <Z2> ~ 50):

• In the same approximation, the elastic gas scattering lifetime becomes:

For NGLS:• Assume 1% energy acceptance (logarithmic dependence) relative

losses of 10-9 for 100 nTorr due to inelastic scattering over full length

• Assuming 7mm ID vacuum chamber relative losses of 10-8 for 100 nTorr due to inelastic scattering

• 1-10 mW for nominal beam power (ALS total beamloss power about 30 mW) – No concern

Post Linac Collimation (Gas, Touschek Scattering)

Test of post linac collimation by artificially increasing (20x) energy spread of beam at points along the linac

For energy error originating within LINAC (inelastic gas or Touschek scattering), very small betatron amplitudes

First momentum collimator in spreader effectively removes scattered beam – very small amplitudes in undulator

Touschek losses In Rings - Bruck’s formula for Touschek

lifetime – valid for flat beam Only complicated part is to calculate

momentum aperture/acceptance For NGLS with its round beams and

changing energy not sufficient Multiple approaches: Monte-Carlo, … We are using approach used by

Xiao/Borland for APS-ERL studies: Based on analytic Piwinski formula:

Still needs calculation of momentum acceptance – because of tight collimator settings (dark current), acceptance is pretty small in parts of line.

Above: APS-ERL example – dependence of Touschek loss-rate in full energy arcs on Momentum ApertureBelow: Momentum Aperture of NGLS with baseline collimation.

Touschek losses (2) Scattering rate based on

analytic Piwinski formula:

Scattering rates with NGLS momentum acceptance + design beam parameters: Integrating local scattering

probability leading to loss on a collimator of up to few 10-6

(<10 W on spreader collimator) – Acceptable

Verified calculation on ALS example – agree well with measured lifetimes

Collimator Design Main issues that determine space requirements

for each collimator (necessary for CDR): Heat load / beam power / power density

<=1 ms MPS -> similar to 3rd generation light sources (kJ) – consistent with XFEL scaling

Impedance heating -> similar to rings Wake fields, effect on beam:

Need to not spoil beam quality Radiation showers, secondary particle

transport, activation: Use of collimator pairs where possible Considered for local shielding and tunnel

wall thickness

XFEL collimator damage In XFEL design collimator damage sets requirements for large

beta functions, one driver for length of collimation section (energy acceptance, R56 tunability, fixed (set of) collimator apertures …)

Scaling of XFEL considerations to NGLS Our assumption is 1 ms MPS, i.e. 1000 bunches

XFEL was 80 – 90 bunches We assume 0.3 nC, XFEL is 1 nC Gun (750 keV)

No concern, low power, very large beam LH, BC

Beam is enlarged a lot due to dispersion Post LINAC

2.4 GeV vs. 20 GeV – total deposited energy is factor 2.2 higher in XFEL – but shower is deeper

Normalized emittance (0.6 vs 1.4 mm mrad) – absolute emittance is factor 3.6 larger in NGLS

NGLS beta functions at collimators factor 10 below XFEL Potentially worse in spreader

Overall seems similar -> Need detailed quantitative analysis But faster MPS response possible (desirable?), i.e. current solution is

feasible

Protector absorbers between cryomodules

At CD-0 design had distributed collimators along length of LINAC and large beta functions to make them effective

Based on tracking of gun dark current and gas/Touschek scattering estimates we do not believe we need those

It was proposed (by reviewers) that local fixed absorbers might be a good idea to localize most of losses (for fault conditions like quadrupole PS trip, …) away from cavities Also provides well defined spots for where to place discrete, fast

loss monitors for MPS Marco incorporated those in new layout However, looking at geometry in more detail, they naturally appear

just downstream of cryomodule (70->35 mm) Still need to verify that location is appropriate and consider

potential impact for designing transition

500 A, 0.6 um, 150 keV, 10 m beta, 2.4 GeV, 3.3 m segment, 4.4 m break, self-seeded (Lux1.5), 25% safety factor on length (Lux1.5x1.25), 60-um Nb3Sn SCU insulator at 80% (0.48 mm diam.)

chamber gap is 2 mm less than magnetic gap

7.5 mm

6.0 mm

Note that 10-mm gap (XFEL) is only ~20 m longer!

Smaller Magnetic Gap and Impact on Undulator Length(Emma)

Sel

f-see

ded

undu

lato

r with

bre

aks,

etc

Estimate of gas scattering loss rates• For electrons one can simplify the formulas for gas Bremsstrahlung

lifetime (in the approximation of <Z2> ~ 50):

• In the same approximation, the elastic gas scattering lifetime becomes:

For NGLS:• Assume 1% energy acceptance (logarithmic dependence) relative

losses of 10-8 for 100 nTorr due to inelastic scattering over full length

• Assuming 4mm ID vacuum chamber relative losses of 10-8 for 100 nTorr due to inelastic scattering

• <20 mW for nominal beam power (ALS total beamloss power about 30 mW) – Still no concern

Effect of smaller undulator gap on darkcurrent collimation

• Smaller undulator gap means vertical collimation is necessary in addition to energy collimamation

• Reducing YCOL from +/-2 mm to +/- 1 mm is sufficient• Losses on YCOL get much bigger – too high ?

– Also tighter tolerances on orbit, collimator position, … - probably OK

To do list + work in progress Further characterize transverse dark current distribution from APEX.

Refine models. Study how to reduce dark current and what final level might be achievable.

Study secondary particles, escaped particles after the collimators. Continue study of sensitivity to lattice errors, changes in initial distribution, collimator misplacements, …

Do trade-off study between cost for shielding/mitigation of activation and complexity and operational impact of collimation system

Carry out tracking of scattered particles (Monte Carlo of gas/Touschek). Potentially benchmark calculations with FLASH measurements.

Finish Collimator hardware reference design Shower simulations, detailed thermal simulations. calculate short and long range wakefields.

Differences NGLS vs. LCLS-2

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

R56 = -94 mmsd = 0.44%

BC2720 MeV

R56 = -76 mmsd = 0.48%

GUN0.75 MeV

Heater94 MeV

R56 = -5 mmsd = 0.02%

L0j 0

Ipk = 47 Asz = 0.85 mm

L1j = -20.0°Ipk = 47 A

sz = 0.85 mm

Lhj = 180°

V0 = 0 MV

L2j = -23.2°Ipk = 90 A

sz = 0.44 mm

L3j = +34.8°Ipk = 500 A

sz 0.08 mm

SPRDR2.4 GeVR56 = 0

sd 0.04%300 pC; 2012-04-18 & 2012-07-02

3.9

CM01 CM2,3 CM04 CM09 CM10 CM27

BC1215 MeV

sd = 0.44%

BC2720 MeV

sd = 0.48%GUN0.75 MeV

Heater94 MeV

sd = 0.02%

L0Ipk = 47 A

L1, LhIpk = 47 A

L2j = -23.2°Ipk = 90 A

L3j = +34.8°Ipk = 500 A

SPRDR2.4 GeV

sd 0.04%

3.9

EnergyCollimator1.5 mm

EnergyColl.

15 mm

EnergyColl.

8 mm

EnergyColl.

2.5 mm

DarkCurrentKicker 10

4 -tronColl.’s

10 mm (x)2 mm (y)

CM01 CM2,3 CM04 CM15 CM16 CM35

BC1E = 250 MeVR56 = -55 mmsd = 1.4 %

BC2E = 1600 MeVR56 = -60 mmsd = 0.46 %

GUN0.75 MeV

LHE = 95 MeV

R56 = -14.5 mmsd = 0.05 %

L0j = *

V0 =94 MVIpk = 12 A

Lb = 2.0 mm

L1j =-21°

V0 =223 MVIpk = 12 A

Lb =2.0 mm

HLj =-165°

V0 =55 MV

L2j = -21°

V0 =1447 MVIpk = 50 A

Lb = 0.56 mm

L3j = 0

V0 =2409 MVIpk = 1.0 kA

Lb = 0.024 mm

LTUE = 4.0 GeV

R56 = 0sd 0.016%

2-km

100-pC machine layout: Oct. 8, 2013; v21 ASTRA run; Bunch length Lb is FWHM

3.9GHz

Summary Have completed tracking with energy + betatron collimators in

CDR lattice Energy collimators sufficient to protect superconducting cavities

+ undulators from gun dark current Dark current kicker appears necessary to minimize activation of

collimators and protect injector s/c cavities Betatron collimation (and post linac energy collimation) effective

in stopping Touschek+Gas scattered particles before undulators Space requirements for collimators are workable within current

layout No apparent show-stoppers remain for CD-1, will finish work in

progress – main area is detailed collimator hardware design

Thanks to Hiroshi Nishimura, Christos Papadopoulos, Fernando Sannibale, et al.

Backup Slides

ΔE = 13.5 MV/m

State of the art25 MV/m BCP cavities

State of the art35 MV/m EP cavities

Elegant Tracking

AT / Elegant comparison

• At first, results did not agree at all … Doubted my AT modifications• However, reason turned out to be intricacies of how elegant tracks

(fiducialization, no reference particle) and how data from astra was transferred

• Now good agreement – small remaining discrepancies are different modeling of apertures, small differences in import of large energy offset coordinates from ASTRA

JH ScrapersSector 1

JH ScrapersSector 3

ALS Routine Stored Beam Losses

• New scrapers localize losses away from beamline source points and undulators

• Installed+work very well

DESY – FLASH / XFEL