1. GLAST LAT Project S. Ritz 1 GLAST Large Area Telescope: Adding Converter to the Blank TKR Planes? Bill Atwood, Tune Kamae, Steve Ritz 4 September 2002 Gamma-ray Large Area Space Telescope

2. GLAST LAT Project S. Ritz 2 Outline  Statement of the issue and proposal  Memo from the PI to the IS  Investigating the need: on-board albedo gamma rejection  Summary of 2001 PDR studies  New studies using GLEAM  Potential negative impacts of additional material  Summary

3. GLAST LAT Project S. Ritz 3 Statement of the Issue, Based on BFEM Studies (Kamae, August 2002) “Anomaly found: Lower portion of the tracker Was hit by 2-3 times higher flux from behind.” “Scaling to LAT: For these events coming from the periphery, multiply by ~5 to scale to LAT. This means ~ 250Hz of albedo gamma triggers. and a good fraction of them to be downlinked” “Study showed: Low energy e-/e+ produced in interactions in CAL “evaporate” to Tracker. Their Typical energy is a few MeV.” “If the standard 3.6%RL Pb is added to the bottom 3 trays, trigger by gammas decrease to about 40%. T. Kamae’s Recoomendation”

4. GLAST LAT Project S. Ritz 4 Memo from the PI Dear Steve, I am writing to you in your role as Instrument Scientist to request that you carry out an evaluation of the proposal that Tune Kamae has made to add radiators to the last two planes of the LAT tracker. My understanding is that the intention of this proposed change is to reduce the trigger rate due to soft events that boil out of the calorimeter and cause triggers in the tracker. Given the current state of the tracker design and the need to redesign certain aspects of the mechanical structure of the tracker, now is the time to consider whether the project should implement the change that Tune has proposed. However, the timescale is relatively short for making a decision. After discussion with Bill Althouse, I agree that we need to be able to decide within the next 10 days. I realize that this allows very little time for a thorough analysis and suggest therefore that you, Tune Kamae and Bill Atwood consider the possible benefits of adding radiators and whether, from a science point of view or with regard to implementation of on-board triggers, if the proposed change has any likely negative impacts. I would like to receive a report from you on this subject by Sept 5. Regards, Peter

5. GLAST LAT Project S. Ritz 5 Analysis Review the statements on –The L1 trigger rate (250 Hz) –“a good fraction to be downlinked” –low energy particles produced in interactions in CAL “evaporate” to Tracker Compare with previous studies Compare with new studies using GLEAM. Consider the potential negative impacts of additional material in the TKR.

6. GLAST LAT Project S. Ritz 6 (1) L1 Trigger: Albedo Gamma Rate ? The following four slides are from the January 2002 PDR presentation, showing –the energy spectra of the background fluxes (orbit max and orbit average) –the L1T rates Bottom line: the 250 Hz estimate by Tune for the albedo gamma rate agrees well with the PDR study.

7. GLAST LAT Project S. Ritz 7 Implemented Orbit-max Background Fluxes total Integrates to ~10 kHz/m 2 LAT-TD-00250-01 Mizuno et al Note by Allan Tylka 12 May 2000, and presentations by Eric Grove AMS Alcaraz et al, Phys Lett B484(2000)p10 and Phys Lett B472(2000)p215 Comparison with EGRET A-Dome rates provides a conservative ceiling on the total rate. orbit-max fluxes used for trigger rate calculations

8. GLAST LAT Project S. Ritz 8 Implemented Orbit-average Fluxes Integrates to ~4.2 kHz/m 2 orbit-avg fluxes used for downlink and final background rejection calculations

9. GLAST LAT Project S. Ritz 9 Orbit Max L1 Rates allchimemax albedo_p_max albedo gamma CR e- maxalbedo e+e- flux (kHz/m 2 )9.94.22.60.920.0432.2 L1T (Hz)13,1347,4193,501242791,893 L1T frac10.560.270.020.010.14 L1V Throttle (Hz) 5,51028111,67919037793 L1V Throttle frac 10.510.300.030.010.14 Notes: with the ACD throttle on the TKR trigger, the total max rate is <6 kHz. more on this later albedo gamma rate is for zenith pointed – more on this later, as a function of rocking angle. These rates are high (by ~30%), due to an implementation error in the CAL-LO trigger. They will be updated for the collaboration meeting.

10. GLAST LAT Project S. Ritz 10 chime albedo albedo CRe albedo p  e+e- chime albedo albedo CRe albedo p  e+e- 5 kHz line total: 13.1 kHz L1T unthrottled L1T with Throttle total: 5.5 kHz Orbit Max L1 Rates 1 kHz line 100 Hz line

11. GLAST LAT Project S. Ritz 11 (2) “A good fraction to be downlinked” ? The following 5 slides are from the PDR presentation outline, showing –the strawman onboard filter set –the albedo gamma downlink rate –the effects on the rates rocking away from zenith-pointed Note that, although the albedo gamma flux was included in all background studies, there was not a particular effort required to remove the albedo gammas: the filters that were used to reject other backgrounds were also effective against the albedo gammas. Remember, any particle hitting the CAL can send energy up into the TKR. Bottom line: a good fraction need not be downlinked. With the strawman filter, the albedo gamma rate to the ground is 2 Hz.

12. GLAST LAT Project S. Ritz 12 On-board Filters select quantities that are simple to calculate and that do not require individual sensor calibration constants. Filter scheme is flexible – current set is basis for flight implementation. order of selections to be optimized. Grouped by category for presentation purposes: –ACD info: match track to hit tile, count # hit tiles at low energy Background 100 MeV  outside tile boundary no tile hit inside tile boundary Rate after ACD selections is 180 Hz orbit-avg (360 Hz orbit-max) [cm]

13. GLAST LAT Project S. Ritz 13 On-board Filters (II) –CAL info: most of the residual rate at this point is due to albedo events and other upward-going energy events. Require track-CAL energy centroid loose match, fractional energy deposit in front layer reasonably consistent with downward EM energy flow. If no CAL energy, require track pattern inconsistent with single-prong. –TKR info: low-energy particles up the ACD-TKR gap easily dealt with: project track to CAL face and require XY position outside this band; for low CAL energy, require TKR hit pattern inconsistent with single prong. X (cm) Y(cm) Rate after CAL selections is ~80 Hz orbit-avg (130 Hz orbit-max)

14. GLAST LAT Project S. Ritz 14 On-board Filters Results After all selections, orbit-average background rate is 17 Hz. chime albedo albedo CRe albedo p  e+e- 5 Hz line 2 Hz line 1 Hz line composition: Additional margin available: much of the residual rate is due to high-energy proton and electron events with CAL E>5GeV -- if apply ACD selections onboard to higher energy, rate can be cut in half (to 8 Hz), with ~5% reduction in Aeff at 10 GeV. 16.5 Hz total rate

15. GLAST LAT Project S. Ritz 15 Effects of Rocking: Albedo Gammas cos (  ) full background flux L1T with Throttle frontback albdeo gamma As we rock, the spike spreads in  : At zenith, earth horizon is at 113 degrees. Study what happens when observatory rocks to 35 and 60 degrees off zenith. 35 degree rock60 degree rock Front Back cos(   

16. GLAST LAT Project S. Ritz 16 Albedo Gamma Rates L1T rate [Hz]L1T rate with Throttle [Hz] After filters [Hz] After fiducial cut [Hz] zenith25019022 (no cut) rock 35°26020033 (no cut) rock 60°30025081 (<45°) 3 (<53°) Notes: rates for other backgrounds will be reduced somewhat by the same angle cut, not taken into account here. small incremental L1T rate not a problem calculating the gamma direction only happens at a relatively low rate, if needed (after other filters), so incremental CPU load not a problem. can reduce the downlink contribution to whatever we need with a tighter fiducial cut.

17. GLAST LAT Project S. Ritz 17 Which Filters Remove Albedo Gamma Events? The TKR info provides the greatest reduction of albedo gamma events (note, these selections are used to reject other backgrounds, too). Require: –at least one track to be found and, if no CAL energy, require the bottom of the best track NOT point to the ACD skirt gap: 60 Hz –track to be inconsistent with a single prong (no extra hits anywhere near the best track) OR CAL energy > 350 MeV: 28 Hz –if there is any CAL energy, the bottom of track must reach at least down to the 3 rd TKR layer above the CAL: 10 Hz NOTE: both gammas and e+/e- will “evaporate” up into the TKR from the CAL. The last cut in the above list removes the gammas. Placing more material closer to the CAL will kill this distinguishing characteristic for the subset of upward gammas that convert in the extra material. The remaining reduction to 2 Hz comes from simple CAL shape cuts and the ACD info (tracks exiting through the ACD). Again, these are used to remove other backgrounds, too.

18. GLAST LAT Project S. Ritz 18 New Studies Using GLEAM A sequence of runs using the current GLEAM simulations based on Geant 4 have been undertaken to directly investigate with high statistics the concern over the  albedo flux. Specifically 3 runs have been completed and are labeled thus: 1) SIGNAL: 50K - 100 MeV  ’s, normal incid. over the area of the LAT 2) ALBEDO: 100K “Albedo  ’s” (E -2 spectrum with -.405 < cos(  ) < -.395) 3) SIDE ALBEDO: 100K “Albedo  ’s” (E -2 spectrum with -.05 < cos(  ) <.05) The following plots report the results

19. GLAST LAT Project S. Ritz 19 Where do the 3-in-a-row Triggers Start? What is the distribution of the upper most layer in the LAT that begins a 3-in-a-row trigger? SIGNAL ALBEDO SIDE ALBEDO Tune’s Spike Layer 0 is at the TOP of the LAT - Layer 15 is the last GLAST Super Layer Note that Layer 0 always has more - why? SIGNAL - ACD conversions (ACD ~ 4% rad. lens.) ALBEDO - upward moving events leaving through the Layer 0 & ACD SIDE ALBEDO - ACD Conversions which move downwards through Tracker

20. GLAST LAT Project S. Ritz 20 Tower Rates Which Towers are hit most - Corners, Sides, or Core? The following plots require at least a valid 3-in-a-row & that the Start Layer = 15 (Last GLAST Super Layer) SIGNAL ALBEDO SIDE ALBEDO Expectation: Signal would be strongest in Corners and Sides Conclusion: SIGNAL is ~ FLAT ALBEDO is peaked in Corners / lesser in Sides SIDE ALBEDO is ~ FLAT

21. GLAST LAT Project S. Ritz 21 Calorimeter Energy Distributions Expectation: Events will be typified by low energy deposited in the CAL Data cut on a valid 3-in-a-row which starts in Layer 15 SIGNALALBEDOSIDE ALBEDO Straw-man Cut: Require > 20 MeV in CAL for Layer 15 Conversions SIGNAL: ~ 75% remains ALBEDO: < 10% remains SIDE ALBEDO: < 10% remains

22. GLAST LAT Project S. Ritz 22 Negative Impacts? (I) Many years have been spent on detailed and systematic background rejection studies. Detector modifications that would make the analysis easier have always been welcome provided they do not add significant complexity or cause other problems. Indeed, Atwood’s original design was shaped by these studies. The main concern with adding material in the TKR is the impact on the background rejection analysis. Rerunning the 10 7 background events required to do a quantitative study in the proposed configuration would derail the ongoing software effort on GLEAM, so instead we rely on our experience.

23. GLAST LAT Project S. Ritz 23 Negative Impacts? (II) There are two categories of background that will likely be worsened with additional tracker material: –horizontal primary particles (not tracked) that interact in the additional material creating secondaries that either look like, or are, gammas. The lowest row of ACD tiles were added to help reject these in the last layers of the TKR, however the efficiency requirement on these tiles is less strict since no candidate gammas come from this region. Additional converter (which does not add effective area for science) will be a target for background generation. –a major advance of GLAST over EGRET is the lack of a TOF system, enabling a much larger FOV. It is necessary for the instrument to distinguish upward from downward-going energy by other means. One method of removing upward gammas from primary interactions in the CAL is requiring a found track to be somewhere close to the CAL. The additional material will convert ~6% more upward-going photons closer to the CAL, removing this useful distinction. The additional converter in the TKR will make the problem of upward-going event rejection worse. There are also concerns about good gamma reconstruction. Both our beam tests (1997 and 1999-2000), which provided our detailed experimental check of the simulation, had no converter in the last tracking layers. We have no operational experience without the blank layers. Assessing these concerns quantitatively would require a detailed study.

24. GLAST LAT Project S. Ritz 24 Summary The requirement for the additional material has not been demonstrated. There does not appear to be an “anomaly”: events of this type have been included in the simulations. Means for removing the events of concern on orbit have been identified. These will be reviewed at the collaboration meeting in October. Concerns with adding additional material in the blank tracker layers have been identified. A detailed study would be required to assess the actual impacts quantitatively. An easy additional, optional filter: A CAL Energy requirement for each Start Layer - Would give us a natural and flexible throttle for downlink - If backgrounds are even close to those anticipated, only last Super layers (if that) would have non-zero cuts.

25. GLAST LAT Project S. Ritz 25 Ongoing Work Improve angular distributions of the background flux implementations. Finish flight software filter implementation (the results presented were based on the strawman algorithms that are the starting point for the flight software filter design). Include the flight algorithms in reconstruction/analysis packages to study the effects in detail.