Beam halo muons in the muon system

1. Beam halo muons in the muon system Erik van der Kraaij CERN LCD

2. Sample from L. Deacon • Studied the occupancy (without detector simulation) in the transverse plane through the interaction point. • Sample was cleaned by Martin of duplicates • Sample normalized to 2.4*104 • Taking into account that the detector sees the flux of two beams. Erik van der Kraaij, CERN LCD

3. Distribution in Inner Detectors • The distribution shows the occupancy per 3x3 cm2 integrated over 156ns. • Highest entry is ~ 55 muons per 9 cm2 per train. • That is ~6 muons per cm2 per train. Erik van der Kraaij, CERN LCD

4. Zoomed out: radius 0.3 – 2 m • The distribution shows the occupancy per 3x3 cm2 integrated over 156ns. • Inner part blanked out to better see the scale. • At R = 0.7 m the highest density is ~5 muons per cell (=3x3cm2) per 156 ns. • Per nanosecond this is 3% cell occupancy. Would need multi- readout capability with nsec precision in endcaps Erik van der Kraaij, CERN LCD

5. Zoomed out: radius 2.0 - 7.5 m • At R = 4 m occupancy drops to ~0.1 muon per cell per train. • For the barrel a cell is hit from the side: area of ~ 1x3 cm2. The muon can hit an entire ‘row’of cells in the barrel. • With a timestamping of 10-20ns, the chance of one row being hit is 0.3% Erik van der Kraaij, CERN LCD

6. Energy distribution Muon energy at the entrance to detector • At least 25% has no more than 40 GeV energy. • Long tail with high energy muons. Erik van der Kraaij, CERN LCD

7. Energy vs radial entry point • Dependence on radius is not large. Higher energy muons a bit more focussed in inner region. Erik van der Kraaij, CERN LCD

8. First reconstruction tests on halo muons Applied the tools from the muon reconstruction algorithm • Extrapolate the track in one endcap to the track in the second endcap and try to match them. • Started simple with a particle gun in Mokka at fixed energy of 20, 30 or 50 GeV. • Origin at (1m,1m,-100m), shot parallel to the z-axis with a “thetasmearing = 1degree” Did not get far with the lower energies as they often do not make it through both endcaps. • Averaging the detector density to 7 g/cm3, with 14 m long detector and 2 MeV/cm2 energy loss for a muon: • A muon loses on average 20 GeV when traversing the entire detector including the HCAL barrel. Erik van der Kraaij, CERN LCD

9. Event example 20 GeV muon • At these energies many do not make it through the detector. Erik van der Kraaij, CERN LCD

10. Event example 20 GeV muon • At these energies many do not make it through the detector. Erik van der Kraaij, CERN LCD

11. Event example 30 GeV muon • All make it through, with small deflection. • Some energy depositions visible, not much. Erik van der Kraaij, CERN LCD

12. Event example 50 GeV muon • All make it through, with small deflection. • Slightly more energy depositions visible. Erik van der Kraaij, CERN LCD

13. How to find & reco halo muons • Reconstruct the two tracks in both endcaps • Extrapolate the direction of the track in one endcap to the other side of the detector and measure its distance and angle to the track in the second endcap Erik van der Kraaij, CERN LCD

14. Halo muon algorithm – step 1 1000 events of 50 GeV muons, coming in through the barrel HCAL Clustering algorithm results mostly in clusters with at least 8 hits • Due to problems with pseudolayers the innermost hit is not included. • For ~100 events the trail is divided in multiple smallclusters • The trail on the incoming side has more hits (higher energy) Red: clusters on ‘incoming side’ Black: clusters on other side #hits in cluster Erik van der Kraaij, CERN LCD

15. Halo muon algorithm – step 2 Extrapolating direction of track in first endcap to the other endcap: • Obtain distance of closest approach to the hit in the most inner layer (of the track) • Peak at ~10cm • For ‘normal muons’ a cut of 20cm is applied • Obtain the angle between the directions of the two tracks • Not bigger than a few degrees • For normal muons a cut of 0.2 is applied Distance [mm] Angle [rad] Erik van der Kraaij, CERN LCD

16. Conclusion • Density of muons in the endcaps is too high: • 3% occupancy for 3x3cm2 per ns at innermost radius of muon system endcap (R = 0.7 m) • Have a simple version of a halo muon ID working • For 50 GeV muons relating the first endcap track to the second endcap track is very similar to ‘normal’ muons. • Unfortunately it only works for muons with considerable energy, because the muon must traverse both endcaps for the algorithm to work. • As most muons have low energy, an approach starting in muon endcap and working further inwards in the HCAL would be more efficient Erik van der Kraaij, CERN LCD

17. Backup slides

18. Halo muon algorithm – step 2 Track selection: • An ‘endcap track’ is a cluster of at least 7 hits • ~110 events do not have two good tracks on both side of detector Extrapolating direction of track in first endcap to the other endcap: • Obtain distance of closest approach to the hit in the most inner layer (of the track) • Peak at ~10cm • Obtain the angle between the directions of the two tracks • Not bigger than a few degrees Distance [mm] Angle [°] Erik van der Kraaij, CERN LCD

19. Origin of muons (swapped conf.) SWAPPED configuration SWAPPED configuration Erik van der Kraaij, CERN LCD

20. Pandora clustering For now have two issues with the clustering in pandora: • Hits are added to clusters. But in the absolute first iteration all hits are individual clusters of 1 hit. • Such a cluster can have no direction fitted to it, so gets a ‘standard direction’: its position vector • If the trail of hits is parallel to the beam pipe, the next hit will never be found: the next hit is looked for along the clusterdirection, i.e. the position vector • Solved this a dirty way: for now, the standard direction is (0,0,1). So parallel to the beamline. This way a 2nd hit can be found, and the clustering begins and finishes, finding horizontal clusters. Erik van der Kraaij, CERN LCD

21. Pseudolayers • Muon hits barrel layer #1 - #6, then endcap layer #1 - #3 • In the bookkeeping of clusters in Pandora, layers are redefined such that these endcap layers do come “later” then the barrel layers. Erik van der Kraaij, CERN LCD

22. Pseudolayers • Muon hits barrel layer #1 - #6, then endcap layer #1 - #3 • In the bookkeeping of clusters in Pandora, layers are redefined such that these endcap layers do come “later” then the barrel layers. pseudolayer #105 Erik van der Kraaij, CERN LCD

23. Pseudolayers • Muon hits barrel layer #1 - #6, then endcap layer #1 - #3 • In the bookkeeping of clusters in Pandora, layers are redefined such that these endcap layers do come “later” then the barrel layers. pseudolayer #106 Erik van der Kraaij, CERN LCD

24. Pseudolayers • Muon hits barrel layer #1 - #6, then endcap layer #1 - #3 • In the bookkeeping of clusters in Pandora, layers are redefined such that these endcap layers do come “later” then the barrel layers. pseudolayer #107 Erik van der Kraaij, CERN LCD

25. Pseudolayers • Muon hits barrel layer #1 - #6, then endcap layer #1 - #3 • In the bookkeeping of clusters in Pandora, layers are redefined such that these endcap layers do come “later” then the barrel layers. pseudolayer #119 Erik van der Kraaij, CERN LCD

26. Tested incoming halos at different spots • Halo muons which come in horizontally at the edge of the HCAL barrel have that the two most inner yoke layers have the same pseudolayer. • Halo muons which come in horizontally at the edge of the YOKE have that all but the most outer hit are in the same pseudolayer. • Muonclustering algorithm does not cluster hits in same layer. • Difficult to redefine this, as it would then simply cluster everything in a full event, with say some punch through. • Will have to think about this… Erik van der Kraaij, CERN LCD

27. Example Muon coming in horizontally ~ at R = hcal_outer_edge layer #105 - #120 are clustered Erik van der Kraaij, CERN LCD

28. Example Muon coming in horizontally ~ at R = hcal_outer_edge “first” layer #105 is single cluster Erik van der Kraaij, CERN LCD