Interactions of hadrons in the SiW ECAL (CAN-025)

1. Interactions of hadrons in the SiW ECAL (CAN-025) Philippe Doublet - LAL Roman Pöschl, François Richard - LAL

2. Outline • Introduction • The SiW ECAL (in 2008) • Beam test setup at FNAL • MC simulations • Algorithm to find interactions • Classification • Optimisation • Results • Conclusions

3. Introduction • Studying interactions of hadrons naturally supports the development of Particle Flow Algorithms (PFA)with a betterknowledge of hadronicshowers • Wewant to investigate the hadronicshower structure with the fine 3D sampling of the ECAL • Our goal : analysis and comparison of interactions of pions in the SiW ECAL using test beam data samples and Monte Carlo simulations • We compare observables withvariousmodels of interactions in Geant4

4. The SiW ECAL in 2008 Fullyequipped ECAL 3 x 3 wafers of 6 x 6 pads Sensors = Si pixels of 1 cm x 1 cm trackingpossibilities Absorber = W 30 layers in 3 differentstacks : 1.4 mm of W 2.8 mm 3.6 mm ≈ 24 X0≈ 1 λI ≈ half of the hadrons interactinside the ECAL volume 9 Si wafers Picture of the fullyequipedSiW ECAL

5. Test beamat FNAL in 2008 • 3 CALICE calorimetersinstalled : SiW ECAL, Analogue HCAL, TailCatcher (TCMT) • Triggers :scintillators, Cherenkovcounters • Muon cutsaddedon the basis of simulated muons : < 0.6% remaining • Ask for only one primarytrackfoundwith the MipFinder • Events left : Beamlineat FNAL

6. Monte Carlo simulations • For comparisons, differentphysicslistsweresimulated • QGSP BERT isused as reference for optimisation : no differencebetweenphysicslistsisseenatthislevel

7. A look at interactions of hadrons • Picture of a generic interaction in the calorimeters : • A primarytrackentersthe detector (« MipFinder ») • The interaction occurs • Secondariesemergefrom the interaction zone

8. Visual example 2D profiles of an eventat 10 GeV in the SiW ECAL High energydepositionwhen the interaction starts Interaction layer confirmed by visual inspection Large number of secondariescreated Equation to besatisfied: Ecut (sort of naïve cut)

9. New cutneededatthesesmallenergies The previouscut (Ecut) willfailatsmallerenergies : fluctuations no more negligible. Need a new criterion : relative increase in consecutivelayers 25% of the eventswith an endpoint in the ECAL are seenwhere 10% could not befoundusing the othercriteria  Use both in combination F and F’ values after the knownendpoint of the MC particle in the ECAL F = F’ =

10. Visual example : a new kindisfound Using the new criterion, one finds a new kind of event Here, local energydeposition Quantified by the relative increasein energyand a decrease : Secondary proton (from MC) +

11. Classification • High energydeposition « FireBall » • Increase continues + veto for backscattering « FireBall » Works here and meant for smallenergies Event view of the « FireBall » type at 10 GeV

12. Classification • High energydeposition « FireBall » • Increase continues + veto for backscattering « FireBall » • Local increase  « Pointlike » • Remark :delta rays are naturallyincluded in « Pointlike » but contributelessthan 4% Event view of the « Pointlike » type at 2 GeV

13. Classification • High energydeposition « FireBall » • Increase continues + veto for backscattering « FireBall » • Local increase  « Pointlike » • Others = non interacting • « MIP » • « Scattered » • Remark :delta rays are naturallyincluded in « Pointlike » but contributelessthan 4% 4 cm ! Event view of the « Scattered » type at 2 GeV

14. Optimisation of the cuts (with MC) • Method: use MC to optimise 3 parameters • Standard deviationof « reconstructed – true » layer • Interaction fraction = fraction of eventswith interactions found • Puritywith non interactingevents= fraction of eventswith no interaction found • Graphs: • Ecutvariedfrom 1 to 20 by steps of 1 unit • Fcutvariedfrom 1 to 10 by steps of 0.5 unit

15. Interaction fraction : defininginteracting and non interactingevents Simulatedevents Interaction layer knownfrom the endpoint of the primary Energy per cell / energy in the last layer before interaction for each layer Interactingevents are selectedwithek > 1.2 x ek-1 (thus « Scattered » eventswill not betaken) Otherevents are non interactingeventsand used to calculatepurity Interaction fraction = fraction of interactingeventsfound  shouldcontain « FireBall » + «  Pointlike » Purity = fraction of non interactingeventsfound shouldcontain « MIP » + « Scattered »

16. Exampleat 10 GeV Ecutvaried Fcutfixed Areas of interest Results :choice to merge all Fcuts for simplicitysince changes have littlesystematics Fcutvaried Ecutfixed

17. Efficienciesafter optimisation • The efficiency to find the true interaction layer within ±1 and 2 layersis the result of the optimisation. • It iscomparedwithanothermethod.

18. Rates of interactions Interaction rates similarbetweenphysicslists Small systematicswithEcut and Fcut in ±1

19. Meanshower radius Gives an idea of the lateral extension of the shower Discrepancy for r > 50 mm MC normalised to number of data events MIP peak Broad peak for interaction classes

20. Separation per class of event : 8 GeV Discrepancy MIP and FireBall have a correct behaviour Data vs QGSP_BERT at 8 GeV

21. Separation per class of event, evenat 2 GeV Tendency to have the samebehaviour MIP and FireBallstill have a correct behaviourdespitesmallerstatistics Data vs QGSP_BERT at 2 GeV

22. Spotteddiscrepancyat 8 GeV : QGSP and FTFP Discrepanciesseenbefore No discrepancywith FTFP_BERT (uses Fritiof model at 8 GeV)

23. Longitudinal profiles Build longitudinal profiles withpseudolayers : = 1 pseudolayer in first stack = 2 pseudolayers in second stack = 3 pseudolayers in thirdstack Withenergiesextrapolatedlinearly Colors are for varioussecondary contributions (from MC table)

24. Longitudinal profiles : FireBall 2 GeV : Large differencies in total Alsodifferencies in showersubcomponents 8 GeV : Recoverspreviousanalysis of pions in the ECAL

25. Longitudinal profiles : PointLikes 2 GeV : Energydeposited by differentsecondaries Remark : good behaviour of LEP 8 GeV : BERT physicslistsperform best here

26. MIPs and Scatteredevents Good behaviours but unfound interactions in the end of the ECAL are seen

27. Conclusions • Interactions of hadrons in the SiW ECAL atenergiesfrom 2 GeV to 10 GeVare found and classifiedinto4 kinds, usingenergydepositionand highgranularity • Efficiencies to reconstruct the interaction layer within ± 2 layers are > 62 % • Systematiceffectshave been checked and are small, O(1%) (muons, physicslist, cuts) • The CAN note isalmostcomplete and ready for circulation within the collaboration

28. Backup slides Efficiency to select eventswith one particle Cutsagainst noise Systematics due to the physicslist

29. Efficiency of the MipFinder • Efficiencies to find the correct number of particlesentering the ECAL • Efficiencies : 99% with one track, 80% withtwotracks (muons) • 12% of irreducible background for overlaid muons (enter the samecell)

30. 2D correlationsbetweenreconstructed and true layer 2 GeV 10 GeV Horizontal axis = Reconstructed layer Vertical axis = True (MC) layer (given by the endpoint of the primaryparticle) Good at 10 GeV, more difficultat 2 GeV : smallerdepositions, but fluctuations

31. Standard deviation :Reconstructed layer – True (MC) layer Cutstoosmall Cutstoo large Measure of the standard deviationwithdifferentEcut/Fcut Good cuts

32. Cutsagainst noise • Efficiency (interaction fraction) and purity for eachenergies • Calculatedwithdifferentcuts on the minimum cellenergy(mipcut) • Not sensitive • Error bars are systematicsfrom(Ecut±1,Fcut±1)

33. Systematics due to physicslists • Efficiency (interaction fraction) and purity are calculated for all physicslists • Error bars are systematics due to (Ecut±1,Fcut±1) • Differencies are < systematics due to (Ecut,Fcut)