CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the...

24
MULTIPLE SCATTERING CM34 @ RAL Timothy Carlisle 1

Transcript of CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the...

Page 1: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

MULTIPLE SCATTERING

CM34 @ RAL

Timothy Carlisle

1

Page 2: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

INTRO.

MICE performance predicted using the cooling formula (CF):

G4MICE ≠ CF

(see prev CMs)

Simulation disagrees with C.F by up to 20%.

MS calc. typically approx.:

CF uses Rossi-Greisen (1941).

Somewhat crude.

Rutherford scattering law - simplest description of MS.

2

2 3

0

0.014 GeV

2

tn nd dE

dz E dX Em X

p

XXrms

plane

0/MeV14

2 N

Page 3: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

RUTHERFORD MS

Rutherford tends to infinity at θ=0

Apply limits:

Lower cut-off by Uncertainty Principle:

(Thomas-Fermi model of the atom)

Upper by rel. size of nucleus:

Integrate, giving:

3 Rutherford Form. I -with hard limits for scattering

Page 4: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

DERIVING THE PDG FORM

1941: Rossi-Greisen

1974 - PDG - log correction to R-G (by Highland) • removes strong path length dep.

• *Z dep. remains however*

)ln(038.01MeV6.13

00 XxXXzp

rms

plane

PDG values for X0 include e-

R-G implicitly incl. atomic e- scatters.

However...

Heavy particle scatters off e- have a rel. small max. angle.

Limited by kinematics.

Not incl. by R-G/PDG

4

4

Definition of X0: Log terms cancel..., substitute X0 into diff. eqn.

Page 5: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

SCATTERING OFF ATOMIC ELECTRONS

e- collisions are significant, increasingly at low Z.

Include e- by using different limits for e- scatters.

... integrate to get:

However

- Also need to include screening of the nuclear potential by e-s.

Rutherford Form. II -with sep. hard limits for nuclear & electron scatters

5

Page 6: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

ATOMIC SCREENING

N

e- screened nucleus

Wenztel-Moliere Form.

• Atomic e-s reduce nuclear charge seen by muon.

•1927: Wentzel assumed exponential scattering

potential, used Born Approximation.

later corrected by Moliere

6

Integrate to get:

Page 7: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

COMPARISON OF FORMULAE

7

Mean Sq. Angle of scattering / X0 (Z). Muons at 200 MeV/c.

electrons (W-M)

Total – Wentzel-Moliere

nuclei (W-M)

Total – Rutherford w. hard limits

PDG

R-G

Page 8: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

SIMULATION OF MULTIPLE

SCATTERING 8

Page 9: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

SCATTERING IN SIMULATION CODE

Geant4: several different scattering models!

Urban (default), Wentzel-VI & CoulombScattering model

Compared with MuScat in *, Urban ‘worst’ of the 3.

Urban used by LHC sim. exps., quoted as 1% precision.

Used by MAUS – now builds with the latest version G4.9.5.p01.

ELMS

Uses knowledge of photoabsorption cooeffs to describe atomic structure.

Only for LH2, only useable through ICOOL at present...

Can we build a Monte Carlo using the Wentzel-Moliere Z-section?

9

* Geant4 models for simulation of multiple scattering V N Ivanchenko et al 2010 J. Phys.: Conf. Ser. 219 032045

YES!

Page 10: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

COBB-CARLISLE MODEL

New Monte Carlo Model, based on Wentzel-Moliere X-section

Simulates muons fired through block of absorber material.

Predicts .

Integrate over angle and thickness

Gives Nnucleus & Nelectron scatters.

E.g. for 10.9cm LH2 ~ 36,000 scatters (total).

Nuclear scattering angles distributed as:

Integrate over angle, solve:

Sum for all collisions = total scat. angle.

Can compare with MuScat results...

10

Page 11: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

MUSCAT ABSORBERS

11

Page 12: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

COMPARISON FOR 10.9CM LH2

c2~12

c2~15

vs. MuScat

12

Probability /

radian Cobb-Carlisle

Model MuScat

ELMS

Geant4.7.0p01

Page 13: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

c2~130

vs. MuScat

Probability /

radian

13

COMPARISON FOR 0.15CM AL

Cobb-Carlisle

MuScat

Geant4.7.0p01

Page 14: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

0

100

200

300

400

500

600

700

1 10

c2

log Z

ELMS

Geant4.7.p01

Cobb-Carlisle

LH2 Li

Fe

Al

C

Be

14

COMPARISON WITH MUSCAT

Page 15: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

MS MEASUREMENTS IN STEP IV [CM32]

Use MAUS to evolve tracks to

downstream face of absorber.

Use tracker hits as starting values.

Scattering angle is the angle between

two vectors.

Model tracker resolution by

smearing.

Tracker code status?

q

qp

qpspace

rms

.

cos

p

MeV/c58.4

MeV/c52.1

MeV/c05.2

mm44.0

mm54.0

Pz

Py

Px

y

x

MICE Note

#90

15

Page 16: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

AFC + LH2

Smear

][mradspace

10 20 30 40 50 60 70 80

No

Smear

16

Page 17: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

SUMMARY

Longstanding disagreement between Geant4/ICOOL & CF.

New Monte Carlo model. Simple, uses 4 parameters. Doesn’t incl. energy loss correlations (unlike

ELMS).

V. good agreement with LH2, less so at higher Z.

Unclear why! Thomas-Fermi model less accurate at low Z...

Need to explain atomic structure better encapsulated by .

Eq. Emittance lower than CF, ~20% less for LH2 – consistent with Geant4.

New Geant4.9.5.p01 to be tested shortly. How close to MuScat?

Step Length dependencies fixed?

Step IV can & should measure MS. Complement MuScat measurements.

Test of physics models.

Resolution covered @ CM32 – to be revisited.

17

Page 18: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

EXTRAS 18

Page 19: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

19

Page 20: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

20

Page 21: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

21

1)1ln(

1

1 2

22

,

c

F

cplanerms1990 - Lynch & Dahl expression

• “much better approximation...agrees with Moliere scattering to 2% for

all Z”

• ...Derivation / comparison with Highland / Moliere not

supplied!

• doesn’t seem to have replaced Highland…

Page 22: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

22

G4MICE: Step IV 63mm LiH

Note: known Step Length problems at 1mm, as

used in these simulations [old Geant4 version].

Page 23: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

23 DzLiH = 6.3 cm

DzLH2 = 57.6 cm

Note: known Step Length problems at 1mm, as

used in these simulations [old Geant4 version].

Page 24: CM34 @ RAL Timothy Carlisle · 2018. 11. 15. · INTRO. MICE performance predicted using the cooling formula (CF): G4MICE ≠ CF (see prev CMs) 2 0 Simulation disagrees with C.F by

EQ. EMITTANCE & SCATTERING ANGLE (STEP LENGTH)

Pencil beams sent

through absorber block

24