Frank Zimmermann, LHC Electron Cloud, GSI Meeting 30.03.2006 Frank Zimmermann introduction electron build up pressure rise heat load & scrubbing

Frank Zimmermann, LHC Electron Cloud, GSI Meeting 30.03.2006

Frank Zimmermann

introduction electron build up pressure rise heat load & scrubbing instabilities incoherent effects simulation needs

Overview of LHC Electron-Cloud Effects & Present Understanding


introduction• CERN ISR (70s) & KEK PF (late 80s) experience

→ 1997: 1st LHC ECLOUD simulation, crash program• 1999: e- cloud seen with LHC beam in SPS, PS & even PS-

SPS transfer line• 1999: e- cloud at both B factories• ~2002: e- cloud at RHIC & Tevatron → observed in all proton

rings with LHC-like parameters (though for 1/5 LHC bunch charge or 10x bunch spacing)

• 2004: DAFNE, 2006: CESR • ?: SNS & J-PARC

truly astonishing if this problem will not occur in LHC


blue: e-cloud effect observedred: planned accelerators


e- build up


schematic of e- cloud build up in LHC arc beam pipe,due to photoemission and secondary emission

[F. Ruggiero]


LHC strategy against electron cloud

1) warm sections (20% of circumference) coated by TiZrVgetter developed at CERN; low secondary emission; if cloud occurs, ionization by electrons (high cross section ~400 Mbarn) aids in pumping & pressure will even improve2) outer wall of beam screen (at 4-20 K, inside 1.9-K cold bore) will have a sawtooth surface (30 m over 500 m) to reduce photon reflectivity to ~2% so that photoelectrons are only emitted from outer wall & confined by dipole field

3) pumping slots in beam screen are shielded to preventelectron impact on cold magnet bore4) rely on surface conditioning (‘scrubbing’); commissioning strategy; as a last resort doubling or triplingbunch spacing suppresses e-cloud heat load

uniquevacuum system!


R. Cimino, I. Collins, 2003; CERN-AB-2004-012

probability of elastic electron reflection seems to approach 1 forzero incident energy and is independent of *max

yield


data from SLAC: R.E. Kirby, F.K. King, “Secondary Emission Yields from PEP-II Accelerator Materials”, NIM A 469, 2001

dependence of secondary emission yield on impact angle

Copper -different surface

finish andsurface chemistry - large variation

in behavior, CERN data not

available

model


[M. Furman, 1997]

cos17.01

cos121exp

maxmax

maxmax

EE

2

0

0

EEEEEE

Eelastic

Present Model of Secondary Emission Yield ,,,, prediffusedpelasticptrueptot EEREE

secondary electrons consist of true secondaries and elastically reflected;since 2003 we assume that elastic reflection is independent of (no data)

sp

pptrue EEs

EEsE

)(/1/

,max

maxmax

[Kirby, 2001;Henrist, 2002;Furman, 1997]

true secondaries:

elastic reflection:[Cimino, Collins, et al., 2003]

this quantum-mechanical formula fits the data well for E0~150 eV

M. Furman includes rediffused electrons and finds that they increase theheat load by 100%


R=1, R=1,

Illustration of present secondary-yield model


e- cloud diagnostics @ SPS

MBA chamber

Copper strips

Holes (transparency 7%)

Beam

B field

Collecting stripsBeam “pipe” (< 30 K)Thermal shielding (80 K)

cold strip detector Motor

Moving plate

RF contacts

variable aperture strip detector

shielded pick ups quadrupole strip detector COLDEX

+ WAMPAC1-4+ pick-up calor.+ SD1-2 + RGAs… J.M.Jimenez, V. Baglin,

N. Hilleret et al.

IcAIsA

M

e- GUN

e-

e-

e-

BeamSPS chamber

=Ic+Is

IcIbeam

= Ic

in-situ max


benchmarking ECLOUD code with SPS measurements

surface conditions (max, R) and detector properties are uncertain constrain parameters by benchmarking multiple measurements change distance between trains & use relative measurements

two differentbunch train spacings

two differentpressures(40 ntorrand 4 ntorr)

Daniel Schulte

ECLOUDsimulation


three curves intersect at max=1.35, R=0.3;flux at later times (=0.3 mA) max=1.2 was reached

flux: (1) ratio 1&2 trains, (2) two spacings, (3) absolute

Daniel Schulte

note:resultssensitiveto pressure,chambergeometry,etc.,variation: max~1.4-1.3R~0.1-0.7

ECLOUDsimulation


pressure rise


vacuum pressure rise

warm

cold

measured e-flux

pressure rise observationsRHIC

0

2

4

6

8

10

0.0E+00 2.0E+10 4.0E+10 6.0E+10 8.0E+10 1.0E+11Protons bunch intensity

P/P

Duty cycle ~ 45 %after 17h integratedtime of LHC beam

0

2

4

6

8

10

12

14

0.E+00 1.E+10 2.E+10 3.E+10 4.E+10 5.E+10 6.E+10 7.E+10 8.E+10 9.E+10 1.E+11Proton bunch intensity

P/P

Duty cycle ~ 45 %after 17h integratedtime of LHC beam

SPS

dipole field

no field

TEVATRON

threshold ~4x1010 ppbvacuum increase in most straights


e- flux@wall vs. intensity, 25 ns spacing, ‘best’ model

R=0.5

calculation for 1 batch

max=1.7

max=1.5

max=1.3max=1.1

vacuum pressure with electron cloud

17 hr running at 3 mA/m gives CO pressure corresponding to100-hr beam lifetime (N. Hilleret, LHC MAC December 2004)

ECLOUDsimulation


' tot desorption yield

strongly bound molecules,varies with e- dose!, cleaning rate is a function ofmaterial, cleanliness, temperature

“recycling desorption yield”, varieswith surface coverage, pressure,sticking coefficient

' usually

'

'

SndtdA

SnCndtdnV

BS pumping speed

hole pumping

surfacecoverage

e- fluxVincent Baglin

Cn

in equilibrium:

Vincent Baglin;see W. Turner,PAC93


AT-VAC (V.Baglin, N.H.) has simulated the LHC pressure evolution. According to Noel’s lab measurement, for E> 30 eV, the e- recycling yield is large.

Therefore, under electron bombardment the BS will have a bare surface without any monolayers. Monolayers will be only on the cold bore.

BEAM SCREEN COVERAGE VERSUS ELECTRON DOSE

1.E+10

1.E+11

1.E+12

1.E+13

1.E+14

1.E+15

1.E+13 1.E+14 1.E+15 1.E+16 1.E+17 1.E+18 1.E+19 1.E+20 1.E+21

ELECTRON DOSE ( e -/cm2)

CO

VE

RA

GE

(mol

./cm

2 )

H2 COVCO COVCO2 COVCH4 COVH2 EQ COVCO EQ COVCO2 EQ COVCH4 EQ COV

300.01.0

E moy (eV):P (W/m/ap.):

1 monolayer = 1E15 molecules/cm2

N. Hilleret,LHC MAC Dec 2004

Information from Vincent Baglinpressure effects

1 min 17 hours

for 2E16 e/m/s i.e. 3 mA/m

max~1.3

1 hour


Pressure increase due to e-cloud. Level is a linear function of the electron flux. It depends only on the electron dose

GAS DENSITY VERSUS ELECTRON DOSE

1.E+12

1.E+13

1.E+14

1.E+15

1.E+16

1.E+17

1.E+13 1.E+14 1.E+15 1.E+16 1.E+17 1.E+18 1.E+19 1.E+20ELECTRON DOSE (e-/cm2)

GA

S D

EN

SIT

Y (

m-3

)

H2

CO

CO2

CH4

H2 EQ

CO EQ

CO2 EQ

CH4 EQ

300.01.0

E moy (eV):P (W/m/ap.):

for 2E16 e/m/s i.e. 3 mA/m

1 min 17 hours (assuming 2 stripes of 3 mm each)

N. Hilleret,LHC MAC Dec 2004

Information from Vincent Baglin

e- fluxdose

100-hr lifetime H2

100-hr lifetime CO2

and cleaning rate a depend on the e- energy; if the energy decreases from 300 eV down to 100 eV, the eta decreases by a factor 3, similarly, the cleaning rate decrease as well. V.B. expects the pressure of Noel's plot will be about the same for 300 eV or 100 eV.

SDD

SP

a1

00

1 hour

shortest lifetime ~ 10 hr


heat load


arc heat load vs. intensity, 25 ns spacing, ‘best’ model

calculation for 1 train

R=0.5

computational challenge! higher heat load for quadrupolesin 2nd train under study

max=1.7

max=1.5

max=1.3max=1.1

max=1.3-1.4 suffices

BS cooling capacityinjection

low luminosityhigh luminosity

ECLOUDsimulation


heat load in COLDEX (prototype LHC vacuum chamberin the SPS)

heat load - constant !?(possibly consistent with conditioned state)

threshold at ~7x1010

p/bunch

V. Baglin

favored interpretation:very fast conditioning (?)

estimated SEY

simulatedheat load


is “scrubbing” needed in LHC? still lacking experimental data, e.g., on max

uncertainty in heat load prediction of factor ~2 also incomplete understanding of scrubbing

(COLDEX data vs. prediction, RHIC, DAFNE) if max~1.3 reached in commissioning, no scrubbing

is needed for heat load and fast instabilities pressure should be ok too according to N. Hilleret one concern: long-term emittance growth and poor

lifetime (observed in SPS after scrubbing) we still believe we need to prepare a scrubbing

strategy in case it turns out to be necessaryto go to max~1.3 (e.g., tailor train spacings & train lengths at nominal bunch intensity)


max=1.7

max=1.5

max=1.3max=1.1

nominal Nb

stability limitat injection

the challenge is to decrease max to 1.3 with a stable beam

nominal filling patterntop energy

ECLOUDsimulation


instabilities


INP Novosibirsk, 1965 Argonne ZGS,1965 BNL AGS, 1965

Bevatron, 1971 ISR, ~1972 PSR, 1988

AGS Booster, 1998/99 KEKB, 2000 CERN SPS, 2000


coupled-bunch instability

extrapolating instability threshold from SPS to LHC

electrons protons

crp

CB 2 SPS: 26 GeV/c, ~40 m; LHC: 450 GeV/c, ~100 m

→ CBI is ~7 times weaker in LHC


single-bunch “TMC” instabilityfast growth above e- densitythreshold; slower growth below

= 1 x 1011 m-3= 2 x 1011 m-3

= 3 x 1011 m-3

“Transverse Mode Coupling Instability (TMCI)” for e- cloud ( > thresh)

Long term emittance growth ( < thresh)

E. Benedetto LHC, Q’=0,at injection


HCrQ

py

sthre

12,

Nyy

zeb

Nyy

zebyezy

rNrNc

H,,

,

24

24141

ebzy

Ny

p

sthre rNCr

Q

,,

22

estimate ofthreshold density

pinch enhancement

assume only vertical pinchsecond term is much larger

→synchrotron tunechanges if z and ||

are held constant

→

TMCI e- threshold density

2

,||,2

2

1

zrev

Nrmsc

s mfQ

12

4

1,

22

,||,2

,ebzy

Ny

pzrev

Nrmse

thre rNCrmf


attempt at extrapolating TMCI threshold from SPS to LHC using analytical estimate

SPS: C~6900 m, z~0.3 m, ~40 m, 26 GeV/c, c~1.8x10-3

LHC: C~26700 m, z~0.011 m, ~100 m, 450 GeV/c, c~3.2x10-4

2/12/12/5

2/1

z

cthr C

→ threshold LHC ~ 1/3 threshold SPS

→ threshold LHC ~ 1/2 threshold SPSwithout pinch enhancement factor H:


simulated emittance growth vs. electron density

no field

dipole

no field

dipole

SPS 26 GeV/c LHC 450 GeV/c

rise time ~1/s rise time ~1/s

rise time ~1/s rise time ~1/s

threshold LHC ~ SPS

E. Benedetto

fast growthslow growth

HEADTAILsimulations


incoherent effects


KEKB: Emittance increase with current below threshold & reduced luminosity w/o instability sidebands

RHIC: transverse instabilities, emittance growth, and beam loss, especially near t

SPS: Poor beam lifetime & bunch-length shrinking after scrubbing

TEVATRON: Fast beam emittance growth and short beam lifetime observed simultaneously with the ECE pressure rise. But little coherent motion seen on Schottky monitor. Longitudinal quadrupole oscillation.

Experimental Indications


KEKB e+ beam blow up, 2000(H. Fukuma, et al.)

threshold of fastvertical blow up

slow growthbelow threshold?

beam current

IP spot size


Fourier power spectrum of KEKB BPM data

• LER single beam, 4 trains, 100 bunches per train, 4 rf bucket spacing• Solenoids off: beam size increased from 60 m ->283 m at 400 mA• No excitation

V. Tune Sideband Peak

J. Flanagan et al.


KEKB Sidebands and Spec. Lum.

• Sidebands disappear at around a bunch current of 0.8 mA.

• Specific luminosity of 2-bucket and 4-bucket spacing bunches do not merge at that point, however.– Possible indication of the

presence of an incoherent component below the sideband threshold (non-linear focusing by cloud leading to non-Gaussian beam tails, e.g.)

Sideband Peak Height

Specific Luminosity

4-bucket spacing

2-bucket spacing

SidebandThreshold

J. Flanagan et al.


evolution of longitudinal profile during beam loss near t

RHIC beam loss at transition

J. Wei


TEVATRON

emittance growth >34 mm mrad/hr(> 100% hr); beam lifetime ~24 hr(normally ~1000 hr)

vertical emittance vs. time vertical Schottky power vs. time

Schottky power -12 dBm(normal instability signals between0 and 10 dBm)

X.L. Zhang et al.

http://www-bd.fnal.gov/cgi-mach/machlog.pl?nb=tev05&action=view&page=-2498&button=yes&invert=no


poor lifetime in the SPS after scrubbing

Poor beam lifetime with LHC beam in the SPS on August 13, 2003 (can it be explained by electron cloud?)

Courtesy G.Arduini

at 26 GeV/clifetime 10-20 minutes, decreasingalong bunch train

not a problem per sein SPS, but it would bein LHC at injection

origin not understood


poor lifetime in the SPS after scrubbing, cont’d

two nominal batches at 26 GeV/c,225 ns spacing between batches;

bunch intensity in store e- cloud signal

LHC beam signal

e-cloud on shielded pick up

J.M. Laurent, J.M. Jimenez, ~2002

E. Shaposhnikova et al., 11/11/04

both patterns are similar and show similar dependence on batch spacing


typical“TMCI”instabilitythresholdat injection

e- central density vs. Nb, 25 ns spacing

R=0.5

calculation for 1 train

max=1.7

max=1.5

max=1.3 max=1.1

challenge: how to go from max=1.7 to 1.3?scrubbing should be done at nominal Nb (stripes)

ECLOUDsimulation


simulated e- density evolution during a bunch passage in an LHC field-free region on log scale

E. Benedetto

HEADTAILcode

bunch tail


e- density on horizontal axis at different time stepsduring a bunch passage, for the LHC

E. Benedetto

high local density, high tune shift, varying with x,y,z


average e- density inside circle of variable radius

E. Benedetto

HEADTAILcode

bunch tail

high local density, high tune shift, varying with x,y,z

rotationin e- phasespace


tune footprint obtained by tracking through a frozen e- potential at z=+2z by a frequency-map analysis of HEADTAIL simulation

E. Benedetto,Y. Papaphilippou


e- distribution in dipole measured by SPS strip detector

approximation forHEADTAIL code

E. Benedetto


e- density evolution in a dipole field

SPS LHC

x

y

E. Benedetto


evolution of on-axis e- density for the SPS

E. Benedetto


simulated emittance growth for 1 and 10 e-beam Interaction points per turn with & w/o synchrotron motion

E. Benedetto

HEADTAILcode

=2x1011 m-3


horizontal invariant of a proton vs. turn number

G. Franchetti,E. Benedetto

HEADTAILcode

two mechanisms: • resonance crossing and trapping → halo growth• linear motion may become unstable → core growth

Ts


incoherent emittance growth due to e- cloud simulated either by HEADTAIL (weak-strong mode) or by analytical field model G. Franchetti,

E. Benedetto

→ emittance growth is not a numerical artifact→ analytical model allows accessing longer time scale


vertical phase space and frequency spectrum of particlemotion at different z positions along the bunch

E. Benedetto,G. Franchetti

singleinteractionpoint=1014 m-3 linear

instability,hyperbolicfix point

chaoticmotion


simulation needs


e-cloud build up code

e-cloud SB/CB instabilitycode

self-consistent code

optics codee.g., MADX

beam sizesapertures, B fields, … cloud density,

local growth rates,around the ring or‘from DR to IP’ (M. Pivi),“ECLOUD TWISS TABLE”, incl. 3D e- motion

wake/impedancecode, e.g., HFSS,MAFIA, GdfidL

E(x,t), B(x,t) ‘ecloud wake’,generalizedimpedance

ion codevacuum code

ionizationE(x,t)

ion desorption,ionization

electrondesorption,scrubbing,“e- pumping”

regularinstabilitycode beam-

beamcode

s.c.code

e

beam motion,losses

future ‘complete’ e-cloud simulation?CARE-HHH-2004


summary & conclusions simulations based on SPS benchmarking lead to optimistic heat load prediction; max~1.3 sufficient to reach nominal & ultimate (max~1.3 was obtained in SPS after ~1-2 days at 25-ns spacing) fast instabilities also under control for max~1.3

~5x1011 m-3, but slow growth <1%/s !?! uncertainties:

(1) LHC vacuum chamber is different from SPS; COLDEX either shows no conditioning or it conditions too fast to notice

(2) RHIC, Tevatron & KEKB experience (3) poor lifetime in SPS resembling e-cloud build up pattern(4) dynamic vacuum & detector background in LHC

incoherent slow emittance growth remains concernwe identified two mechanisms causing halo or core blow up: periodic crossing of resonance or unstable region may

explain


thanks to

Gianluigi Arduini, Vincent Baglin, Giulia Bellodi,Elena Benedetto, Giuliano Franchetti, Noel Hilleret,

Bernard Jeanneret, Miguel Jimenez, Laurent Tavian, Kazuhito Ohmi, Francesco Ruggiero, Giovanni Rumolo,

Daniel Schulte, Elena Shaposhnikova, Jie Wei, and Xiaolong Zhang

for important contributions & discussions & help

Documents

Frank Zimmermann, LHC Electron Cloud, GSI Meeting 30.03.2006 Frank Zimmermann introduction electron build up pressure rise heat load & scrubbing