X-rays in MICE RAL 27/10 -04 Rikard Sandström Geneva University 1 Outline • Presentation of the newest RF background model – Introduction – Improvements • Absorber design • Phase and time information – Reusing the background • New results – Particles leaving vacuum windows – Particles reaching the trackers • Summary 2 Introduction • The RF background problem in MICE is still present. – But narrowing it down! • Together with Yagmur I created a way for the user to generate background, store it and reuse it independent of the tracker simulated. – We now have a standard background defined and used. • Many improvements were made along the way. 3 Progress made past weeks • Too much progress to report! This is from our homepage: – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – check rf background output Done (October 18)Rikard generate up-to-date rfBGMultipleSpectraFile for background production Done (October 15)Rikard commit standard rfBGMultipleSpectraFile Done (October 14)Rikard fix track visualization in interactive mode Done (October 15)Rikard add support for generating only background with no beam Done (October 15)Rikard add time delay when using bg bank file Done (October 14)Rikard change input bank to SpecialVirtual planes in BGPlane Done (October 12)Rikard move generatorBank output to EndOfEventAction Done (October 12)Rikard clean up SteppingAction Done (October 12)Rikard add individual time offsets for histogram peaks in BGPlane Done (October 11)Rikard implement looping over input background hits in BGPlane and PrimaryGeneratorAction Done (October 8)Rikard replace generatorBank output with Interface function in BeginOfEventAction Done (October 7)Rikard update MICEPrimaryGeneratorAction and MICEBGPlane to handle reading rf background from Sim.out Done (October 7)Rikard update MICEPrimaryGeneratorAction to handle multiple background planes Done (September 30)Rikard add class for background planesDone (September 30)Rikardcommit PhysicsList update Done (September 13)Rikard add torispherical window to AbsorberVessel Done (October 6)Rikard clean up AbsorberVesse lDone (September 30)Rikard implement spherical shell window in AbsorberVessel Done (September 23)Rikard add flat vacuum windows with no body Done (September 13)Rikard fix absorber vessel geometry Done (September 13)Rikard fix hydrogen composition Done (August 26)Rikard add rfBG parameters to handle reading from Sim.out Done (October 7)Rikard remove integer cast for rfBGParameters.fBGDirection Done (October 7)Rikard add circle radius offset for torispherical window to AbsorberParameters Done (October 6)Rikard update RFBackgroundParameters Done (September 30)Rikard add get/set methods to VirtualHitBank Done (October 7)Rikard add static WriteSim to generatorBank Done (October 7)Rikard fix energy output in generatorBank Done (October 7)Rikard add rfBG parameters to handle reading from Sim.out Done (October 7)Rikard add torispherical absorber parameter to dataCards Done (October 6)Rikard • …and much much more done by Yagmur, Malcolm, Chris and others. 4 Levels of description & understanding OK in z,t Not considered x,y Requires more studies All good OK Improved in Geant4.6? 5 Improved absorber representation • Absorbers and vacuum windows supports different geometrical shape. – Default is spherical. • All optional absorber shapes have the central window thickness and central liquid hydrogen thickness set to the latest design. • Flange sizes might need an update. 6 RF phase calculations in Matlab • G4MICE now supports fixed E-field. – Working and tested. • G4MICE also support for time dependent field. – Needs more testing to confirm working. In the meantime, calculate with Matlab (faster): • The phases for the background electrons assumes phases optimized for a mu+ at 200 MeV/c on axis. – This causes a symmetry breaking in z! – Assumes phase difference between neighboring cavities is constant. – Phase diff = 2.0498 rad = 1.621 ns. – This gives the muon an energy gain of 10.8 MeV per set of four RF-cavities, including energy loss in Be windows. – Energy loss is calculated using STAR data for ionization and bremsstrahlung. 7 RF phase calculation, mu+ 8 Accelerating e- in the RF, intro • • • • The electrons are assumed to have zero kinetic energy when emitted from beryllium windows. The electrons emitted at the peak values of the Efield only (+ and - respectively). They are accelerated using the same Matlab model as used for the muon. This results in 1. energy when leaving the RF-system 2. travel time for leaving the RF-system 9 Accelerating e- in the RF, 2 cavs Downstream direction Upstream direction 10 Accelerating e- in the RF, 4 cavs Downstream direction Upstream direction 11 Comments on Matlab results • The RF phases are set such that – electrons have higher energy in upstream direction. – some electrons turn around if starting with downstream direction. • The situation is better with 466 mm cavities than with 430 mm cavities – In the first case the “turn around” electrons are stopped in the Be windows as they turn -> Only 2/3 of BG upstream! • Worth considering: – If we optimize for a slightly different pµ, can we reduce the background by the change in RF phases? – If we allow e- to be emitted off crest, how much worst could it get? 12 Generating the background • Generating the background as calculated in Matlab at red locations. • Extracting data at green locations. 13 Feeding Matlab into G4MICE 1. The time of emission from each Be window is calculated using the phases. 2. The time of arrival is the travel time + the time of emission. 3. The downstream RF system is offset in time by looking at a reference muon in G4MICE. 4. Background from different RF periods is achieved by repeating with an integer RF period offset in time. – Number of periods is chosen with respect to the flight time of the muon. 5. The electrons are distributed evenly over 21 cm in radius just outside the outer beryllium windows. – r = R*sqrt(rand[0,1]) 6. Particles are assumed to be parallel to beam line initially. 7. Emission rate is given by the measured 40 kHz/cm2 at 8 MV/m measured at Lab-G. 8. Only flip B-field simulated so far. 14 Uncertainties • Rate of e- emission – Largest uncertainty is area considered for emission. (Bessel functions…) – We use worse case here, tool for scaling already exists. – Will be studied experimentally at Fermilab this winter. • Off crest emission – If the particles were allowed to be emitted somewhat off crest E peak, energies might change. 15 Generated RF e-, E_kinetic, upstream MeV 16 Generated RF e-, E_kinetic, downstream MeV 17 Generated RF e-, time ns mu enters RF system mu leaves RF system 18 Background event, few ns 19 Reusing the background Background bank generated, works, supports looping and scaling: • Due to the small fraction of particles making it to the trackers, hits at the tracker entry is saved and can be reused. – Excellent optimization. • The user is responsible for setting the expected time when the muon is arriving at the cooling channel. – Different input beams -> different t0 and z0. • Should a simulation run out of such background events, looping over the background bank is supported. – Random event from BG bank might be implemented. • Background can be scaled down from nominal value, should later be possible to scale up too. • Malcolm has produced thousands of BG events, Yagmur has merged and filtered to one file. – Available on our homepage. 20 Particles leaving vacuum windows • Using mice2root instead of my old readout code. – Very nice but not problem free. • Due to a bug momentum and energy is not saved correctly to output for photons in vacuum windows. – Simulation itself handles them fine though. Propagation to trackers OK. • No problem with e-, but… – Not part of the standard output Malcolm is generating -> Must run locally -> Low statistics. – Low statistics data shows expected 1-1 correspondence between what is leaving the vacuum window and what is reaching the tracker. -> OK to use track ref planes (high stat) to investigate e- from absorbers. 21 e- reaching the upstream tracker Still not the 1000 requested MeV 22 e- reaching the downstream tracker Very few MeV 23 Photons reaching the upstream tracker Follows Landau distribution Reaches full input energy MeV 24 Photons reaching the downstream tracker MeV 25 Results, rates • Assumed emission rate: – Given the 40 kHz/cm2 of emitted electrons off a cavity at 8 MV/m, 201.25 MHz frequency, and a conservative assumption of area to be considered, we have 7.92 e- hitting the absorber per energy peak (and period). – 8 energy peaks * 2 linacs * 7.92 = 126.72 e- per period in total. • Very few particles generate a hit at the tracker entrance but some do. • Data for particles reaching trackers given as per total generated, and scaled to frequency: BG, Upstream, tracker per entrance generated Upstream, [MHz] Downstream, per generated Downstream, [MHz] e- 5.76e-6 0.147 1.72e-7 0.00438 Photon 8.315e-4 21.21 6.934e-5 1.768 26 Future plans • Close to go to latest release of Geant4. – This will hopefully solve some issues we have had with physical processes, step length dependence etc. • Renormalize BG rate using new data which shall be taken experimentally at Fermilab. • Run a few events with non-flip B-field. • Regenerate the background bank file if necessary. • Solve a long list of bugs assigned to me by Chris et al… 27 Summary • Standard background defined, generated and used by all. • Rates in upstream tracker (worst) are 21.21 MHz for photons, 147 kHz for electrons. • Future experiment might change assumption of emission -> Rescale rate! 28
© Copyright 2026 Paperzz