1. HF Calorimeter PMT Upgrade R&D Yasar Onel for University of Iowa, Fairfield University, INFN, Trieste

2. HF Detectors Readout Readout Box IP  11 m away from IP. Fact: High energetic muons or residues of late hadron showers may reach to HF Readout Box. 11/19/20082CMS Upgrade Workshop, Nov 19-21, 2008

3. HF PMT Specs Hamamatsu R7525 PMT has 8 stages dynode 25 mm diameter bialkali photocathode borosilicate glass window maximum QE at 450 nm (23%) The glass window can be source of Cerenkov light itself 11/19/20083CMS Upgrade Workshop, Nov 19-21, 2008

4. HF PMT Events These events are due to Cerenkov radiation from particles directly hitting the PMT window. The glass window is plano-convex. – 1mm thick in center – 6.1mm thick at the edges 11/19/20084CMS Upgrade Workshop, Nov 19-21, 2008

5. Energy GeV Pion Beam HF PMT Events (continued) The PMT window events were seen in HF 2004 data. (Left figure) Very high energy events were also seen in TB2004 analysis (right figure) by A. Moeller. See a talk from the analysis: http://omega.physics.uiowa.edu/HEP/Files/Talks/AbnormallyHighEnergyEventsinHF. pdf http://omega.physics.uiowa.edu/HEP/Files/Talks/AbnormallyHighEnergyEventsinHF. pdf Then it was confirmed in 2007 by using a HF mock up by Jim Freeman et al. We also performed two test with a stand-alone HF PMT at Fermilab and at CERN in 2008. Energy GeV Muon Beam 11/19/20085CMS Upgrade Workshop, Nov 19-21, 2008

6. Muon Energy Spectrum 150 GeV Muon @ Tower 14 The  200 GeV peak is from the muons interacting with HF PMT. The  4 GeV peak is from the muons interacting with HF body. Large tail are events are due to the muons which creates more p.e. in due to the interactions with PMT window. HF Test Beam 2004 This is potentially problematic for forward jet tagging, MET, and luminosity precision. We need an upgrade studies right away! 11/19/20086CMS Upgrade Workshop, Nov 19-21, 2008

7. R&D PROPOSALS 11/19/20087CMS Upgrade Workshop, Nov 19-21, 2008

8. HF PMT System Upgrade We propose to upgrade HF PMTs with two photo-sensors (PMTs) QE of 45% peak or more Thin glass window less than 1 mm thick, and walls cannot generate Cerenkov light (for PMT, a metal envelope) We consider mini-PMTs (2 per tower, total 3456) or multi-anode PMTs (total 1728) 11/19/20088CMS Upgrade Workshop, Nov 19-21, 2008

9. One of the Candidates Hamamatsu 16- Anode PMT (R7600U-200) QE peaks at 400 nm with 45% (ultra bialkali) within the range of 300-650 nm. The metal envelope eliminates the Cerenkov light in the walls. A very thin window (0.6 mm) which also eliminates Cerenkov light production due to head on muons. It gives possibility to assign two channels per tower (or equivalently per light-guide.) 11/19/20089CMS Upgrade Workshop, Nov 19-21, 2008

10. Benefits of Mini /Multi-Anode PMT The HF upgrade plan includes changing PMTs with new generation PMTs. The replacement will help to tag “PMT Events” have better energy resolution Increase muon signal efficiency 11/19/200810CMS Upgrade Workshop, Nov 19-21, 2008

11. Tagging “PMT Events” with Mini-PMT or Multi-Anode PMT 2 channels of PMT per tower can recognize and tag signal event by event when these PMT events occur correct false signal directly by measuring the true signal. If one of the channels in one tower gives large pulse while the other does not, one can correct the signal with the smaller one, given that they cross-calibrate each other. Light Guide Cross-Sectional View Single channel mini-PMT 2 signal channels as “bow-tie” anode connections A traversing muon will only hit one channel, therefore this structure will help to tag “PMT events” 11/19/200811CMS Upgrade Workshop, Nov 19-21, 2008

12. Better Energy Resolution The energy resolution in HF is determined by the photo-statistics (even for hadronic showers since the signal is mostly generated by  0 s). In current HF PMT the light output is 0.25 p.e./GeV (4 GeV per p.e.). By using new PMTs which has refined bialkali photocathode p.e. yield can be increased by a factor of  1.9 (0.48 p.e./GeV). This gives  33% better energy resolution. Hamamatsu QE of  45% peak or more 11/19/200812CMS Upgrade Workshop, Nov 19-21, 2008

13. Increase in Muon Signal Efficiency By using high QE-PMT 1 p.e./muon can be increased to 2 p.e./muon. This will increase the efficiency of HF to see a muon signal up to  80%. The benefit is more ability to remove PMT events which are not interacting with HF fibers. 11/19/200813CMS Upgrade Workshop, Nov 19-21, 2008

14. Short Term Solutions to “PMT Events” Before doing a major upgrade, along with our Fermilab colleagues, we propose to use radiation hard scintillating crystals to tag these “PMT events”. HF PMT Scintillating Crystal Due to the scintillation light from traversing particles (which are the cause of “PMT events”) -the light output, therefore energy reading, will be increased. - also due to the decay time of the crystal the pulse shape will become wider. These features may help to eliminate “PMT events”. Two crystals, namely BGO and LySO, were chosen to be tested. Expectations are * : LySO is faster than BGO (decay time  LySO ~40 ns for LysO and 300 ns for BGO). Light output of LySO is 6 times larger than BGO. * http://www.hep.caltech.edu/~zhu/talks/ryz_080526_calor.pdfhttp://www.hep.caltech.edu/~zhu/talks/ryz_080526_calor.pdf 11/19/200814CMS Upgrade Workshop, Nov 19-21, 2008

15. Continued… We had beam tests with the proposed R&D in the year of 2008 at Fermilab and CERN. 3 LySo and 3 BGO crystals were installed in CMS for beam collisions. 11/19/200815CMS Upgrade Workshop, Nov 19-21, 2008

16. TEST BEAM ACTIVITIES 11/19/200816CMS Upgrade Workshop, Nov 19-21, 2008

17. Test Beam 2008 CERN/FNAL Setup Beam Quartz Calorimeter HF Crystal Setup Small PMT Test Setup 11/19/200817CMS Upgrade Workshop, Nov 19-21, 2008

18. Test Beam 2008 CERN/FNAL Setup-II The crystals were tested at CERN/Fermilab 2008 beam tests. The main objectives to find the effect of them to muons. Timing, pulse shape, and energy response were also subject of beam tests. The mini PMT was tested as part of CERN 2008 beam test. The main objectives were to find response to beam, to identify if glass window and or side walls are sensitive for Cerenkov light production. 165 cm steel Quartz Plate Veto Beam Mini PMT Trigger Muon Veto HF PMT Crystal Wire Chamber 11/19/200818CMS Upgrade Workshop, Nov 19-21, 2008

19. Test Beam Participants at CERN 2008 Yasar Onel 1, Jim Freeman 2, Rick Vidal 2, Taylan Yetkin 1, Elif Asli 1 Albayrak 1, Anthony Moeller 1, Burak Bilki 1, Warren Clarida 1, James Wetzel 1, Ferhat Ozok 1, Nasuf Sonmez 3, Kerem Cankocak 1, Aliko Mestvirishvili 1, David Winn 4, Aldo Penzo 5 1: University of Iowa 2: FNAL: 3: Ege University 4: Fairfield University 5: Trieste, INFN We also had shifters from Boston U. (A. Clough, M. Carleton, P.D. Lawson 11/19/200819CMS Upgrade Workshop, Nov 19-21, 2008

20. Analysis Requirements Required only beam triggered events. Remove beam contamination and applied wire chamber cuts. The ADC counts are converted to charge by using non- inverted conversion table (CMS HF QIE conversions). The fCToGeV conversion is done by using 0.25 p.e./GeV from HF test beam and PMT gain (the resulted conversion is an approximate one). (fC2GeV ~ 0.02 GeV/fC) 11/19/200820CMS Upgrade Workshop, Nov 19-21, 2008

21. Muon Energy Spectrum Comparison 150 GeV Muon @ Tower 14 E Long Calibration with 100 GeV e- 2004 Test Beam. Beam is parallel to the fibers 2008 Test Beam. Beam is parallel to the PMT 225 GeV Muon @ HFPMT 165 cm steel absorber Calibration with 0.25 p.e./GeV The energy spectrum from HF PMT events are consistent with each other. The peak of the signal from PMT glass is ~200 GeV. N p.e. ~370 sin 2  C eV -1 cm -1 ~340 eV -1 for muons with  1, PMT glass with n  1.5 and 6.1 mm thickness. 11/19/200821CMS Upgrade Workshop, Nov 19-21, 2008

22. Direct Hit to HF PMT (muons) 225 GeV mu-, without crystal 225 GeV mu-, LySO (0.15mm) LySo signal is ~600 GeV (400 GeV above Cerenkov signal) Average Pulse Shape 11/19/200822CMS Upgrade Workshop, Nov 19-21, 2008

23. 16 GeV proton beam at Fermi Test ConfigurationApproximate Mean, ADC Count w/o Crystal 2.5  0.07 (1) LYSO 650  5 (260) BGO 90  3(36) ADC Counts  100 as amplifier factor  1 1 16GeV proton, NoXtal, No Lucite16GeV proton, 3mm LySO, No Lucite 16GeV proton, 3mmBGO, No Lucite LySO/ BGO ~7.2 11/19/200823CMS Upgrade Workshop, Nov 19-21, 2008

24. Summary of Crystal Tests The signal with and without crystal configurations were compared. Signal timing were found to be consistent with manufacturer specs. The signal from the “PMT events” is changed with the crystals and the ratios found to be consistent with expectations. The average pulse height distributions are found to be wide which is due to the decay time of scintillating crystals. As a result of these conclusions 6 crystals were installed in HF for LHC beam collisions to make further tests. 11/19/200824CMS Upgrade Workshop, Nov 19-21, 2008

25. (Run 50532 – 175 GeV e - )‏ Typical Pulse Shapes from Mini-PMT The trigger was a scintillator with a mini-PMT. The wide pulse shape is typical. As expected the particles hitting mini- PMT window are producing Cerenkov signal (note the very fast charge collection) 11/19/200825CMS Upgrade Workshop, Nov 19-21, 2008

26. Other Approaches Two veto counters, one is based on quartz and the other one based on Lucite, were also tested. HF PMTQuartz Plate Veto Lucite Rod Veto HF PMTQuartz Plate Veto Lucite Rod Veto If the readout system allows, veto counters at the back of HF can be used 11/19/200826CMS Upgrade Workshop, Nov 19-21, 2008

27. Charge Collection with Mini-PMTs Signal in mini PMT which is used in trigger Signal in mini PMT The mini PMT registers the signal for window hitting particles. 11/19/200827CMS Upgrade Workshop, Nov 19-21, 2008

28. Charge Collection with Mini-PMTs Signal in mini PMT which is used in trigger Signal in mini PMT 300 GeV Pion Beam 175 GeV Electron Beam The mini PMT registers the signal for window hitting particles. 11/19/200828CMS Upgrade Workshop, Nov 19-21, 2008

29. SIMULATION ACTIVITIES 11/19/200829CMS Upgrade Workshop, Nov 19-21, 2008

30. Simulation of “PMT Events” CMS detector simulation cannot produce “PMT events” in its current form. HF simulation is based on shower library from a stand-alone Geant4 simulation. A substantial effort is put to implement these feature to CMSSW by A. Moeller, S. Banerjee. The goal is to study the effect of “PMT events” on physics, especially for forward jets and MET. Also possible filtering algorithms are being studied and simulation will help to test these ideas. 11/19/200830CMS Upgrade Workshop, Nov 19-21, 2008

31. Geant4 HF Simulation in CMSSW The original HF simulation only had two depths, one for long fibers, and another for short fibers. The PMT windows were added to the simulation in the form of discs behind HF (~ 3 < η < 3.2). The hits recorded in these discs are represented by two additional depths, one for PMTs corresponding to long fibers, and another corresponding to short fibers. – There are now four total depths for HF. The simulation currently uses simplified Geant4 detector/interaction processes. GFLASH is considered to replace this approach soon. PMTs in RBXs behind HF. There are two RBXs for each HF wedge, with a total of 48 PMTs per wedge. 11/19/200831CMS Upgrade Workshop, Nov 19-21, 2008

32. Simulation Studies Single particles as well as simplified jets (single quark jets) are sent to the HF region to test the new simulation and filter algorithms. Next few slides show a glimpse of these studies. 11/19/200832CMS Upgrade Workshop, Nov 19-21, 2008

33. 11/19/200833 Energy response, 100 GeV electrons, η=3.401 For electrons at η=3.401, the energy is closer to what is expected. The energy in depth 2 still needs improvement. 49.6 GeV (expected 50GeV) 10.0 GeV (expected 16 GeV) CMS Upgrade Workshop, Nov 19-21, 2008

34. 11/19/200834 Energy response, 100 GeV pions, η=3.401 The energy in depths 1 and 2 is higher for pions as well. As η=3.401 is now beyond the range of the RBX, the number of hits in depths 3 and 4 has decreased. 30.1 GeV (expected 35 GeV) 21.5 GeV (expected 30 GeV) CMS Upgrade Workshop, Nov 19-21, 2008

35. 11/19/200835 20 GeV P T Jets, 3.0 < η < 4.5 Events do occur in depths 3 & 4 for jets. 10.4 GeV 0.4 GeV CMS Upgrade Workshop, Nov 19-21, 2008

36. 11/19/200836 Time of Flight (100 GeV Pions) The TOF peaks at about 53 ns for depth 1 & 2. But for depths 3 & 4, the peak is at about 47 ns. – Timing information may allow us to eliminate abnormal events. Abnormal events from TB04 arrived later that normal. CMS Upgrade Workshop, Nov 19-21, 2008

37. 11/19/200837 Time of Flight (20 GeV P T Jets) The timing information for the jets is similar to that for pions, although the time is slightly later in depths 1 & 2. Using the TOF information to eliminate abnormal events appears to be a possibility. CMS Upgrade Workshop, Nov 19-21, 2008

38. 11/19/200838 Timing Based Cuts 10.39 GeV 10.28 GeV Original With Cuts Only hits with TOF of more than 50 ns were used to reconstruct the energies. CMS Upgrade Workshop, Nov 19-21, 2008

39. Summary During beam tests HF is found to give “PMT Events” which are due to energetic particles hitting the PMT glass window. We proposed HF PMT system upgrade as long term solution as well as other alternatives for short term solutions. – Two channel readout per tower by using mini-PMTs or multi-anode PMTs – Scintillating crystals – Back vetoes We started to have beam tests for upgrade studies for HF at Fermilab and CERN. We are planning to continue these studies during 2009. – The crystals were tested at Fermilab and CERN and results found compatible with expectations. As a result 6 crystals were installed in CMS for LHC beam collision tests. – Mini-PMTs were tested for HF for the first time and analysis of data is currently being finalized. We plan to do more tests with 2 channel readout system in 2009. Simulation studies for CMSSW to incorporate the “PMT Events” are underway. We definitely need this to study effect of these on physics. 11/19/200839CMS Upgrade Workshop, Nov 19-21, 2008