1. HERD Tracker Layout and Photon performance Studies Xin Wu University of Geneva 3 rd HERD Workshop 19-20 January, 2016, Xi’an

2. Evolution of the HERD Layout 1 st HERD workshop in October 2012 – Si-PIN + 4 X 0 Shower Tracker (W + fibers) + ECAL + 3D HCAL – Proposed to replace shower tracker with Silicon-Tungsten Tracker (STK) to improve photon and tracking performance 5 converter layers, thickness 3x1 mm + 2x2 mm = 2 X 0 2 nd HERD workshop in December 2013 – Proposed to reduce W plate thickness to improve PSF for GeV and below photon 3 rd HERD workshop in January 2016 – Propose to add a PANGU-like tracker to the top for best sub-GeV PSF High energy part remains, 5-sides, DAMPE-like, with W converters – 4 layers W of 1 mm thick (1.14 X 0 ) to reach comparable PSF with Fermi in 1 -100 GeV range Low energy part, PANGU-like, no W converters – 40 layers of 320 µ thick silicon (0.14 X 0 ) – Improve PSF by x6-x3 better for 100 MeV – 10 GeV 100 MeV 2Xin WuHERD Workshop, Beijing, 19-20/1/2016

3. HERD conceptual detector design 17/10/123 PWO W+ CsI(Na) + Fiber + ICCD Charge detector: Si-PIN (1cm×1cm×500  m) – Top: 2x(1mx1m), 4 Sides: 2x(1mx40cm) Shower Tracker –W: 10x3.5mm + 2x17.5mm + 2x35mm (4X 0 = 1.6 ) –Scin. Fibers: 14 X-Y double layers, 1x1mm 2, 1m long Nucleon Tracker with Scin. Fibers ECAL: 16X 0 = 0.7 –PWO bar: 2.5x2.5x70cm 3 –6 layers alternate in X-Y HCAL: 30 layers of W plates + CsI cells –W: 30x3.5mm, 3X 0 = 1.2 –CsI cell:2.5x2.5cm 2 x0.2cm Neutron detector: B-doped plastic scintillator with delayed signals Geneva proposes to replace the scintillating fiber shower tracker with a Silicon tracker-converter to improve  and tracking performance October 2012

4. Baseline design of HERD 4/23 December 2013

5. HERD Design ： 3D Calo & 5-Side Sensitive n10X acceptance than others, but weight2.3 T ~1/3 AMS STK(W+SSD) Charge gamma-ray direction CR back scatter 3D CALO e/G/CR energy e/p discrimination STK(W+SSD) 5/43 S-N Zhang, May 2015

6. Characteristics of all components 6/43 typesizeX0,λunitmain functions tracker (top) Si strips 70 cm × 70 cm 2 X07 x-y (W foils) Charge Early shower Tracks tracker 4 sides Si strips 65 cm × 50 cm 2 X07 x-y (W foils) Charge Early shower Tracks CALO~10K LYSO cubes 63 cm × 63 cm 55 X0 3 λ 3 cm × 3 cm e/γ energy nucleon energy e/p separation S-N Zhang, May 2015

7. Now to wrap a beautiful gift … 7Xin WuHERD Workshop, Beijing, 19-20/1/2016

8. Now add PANGU to HERD … First try to fit the envelop: 1510×1480×1580 (overall) and 880×834×729 (calorimeter) – Very challenging to fit services for a 5-sides outward sensitive detector – Simple approach first: 5 identical sides (“DAMPE”) + a light top (“PANGU”) 8Xin WuHERD Workshop, Beijing, 19-20/1/2016 Silicon-Tungsten Tracker Calorimeter Silicon Tracker (“PANGU”) Anti-Coincidence Detector

9. “Economical” Layout High energy (STK): use DAMPE SSD (320 µm, 95×95, 121 µm pitch) – Long ladder: 7 SSD (~67 cm), readout electronics on one side to save space Higher noise all strips readout, no floating strips (uniform S/N) – 7 double-layers of 320 µm Si, 4 with 1 mm W, ~1.14X 0 (~Fermi), Support tray thickness ~25(W)/20(no W), total height ~20 cm 9Xin WuHERD Workshop, Beijing, 19-20/1/2016 “PANGU”: 20 double-layers of SSD, no W (total 0.14X 0 ) – Default same SSD as STK (alternative 150 µm) – New support structure, as transparent as possible – Total height ~ 30 cm ACD: ~6 mm thick, segmented, SiPM readout Alternative technology: scintillating fiber tracker with SiPM (for the STK part) – Advantage: flexible geometry, no wire bonds, less fragile – Disadvantages: low TRL, dark current noise, energy resolution(?) To be demonstrated!

10. HERD Tracker Layout HE (“DAMPE”) More robust tacking with 3 X/Y hits after the last W layer First X/Y hit serve as link to the low energy part on top 10Xin Wu HERD Workshop, Beijing, 19-20/1/2016 TRAY W 4 4 4 4 4 4 208 29 24 26.5 25 20 1 5 4 3 2 1 0 7 TRAY W 6 29 4 66 cm (active Si) W Si X-view Si Y- view

11. HERD Tracker Layout LE (“PANGU”) New light support structure 11Xin Wu HERD Workshop, Beijing, 19-20/1/2016 3 3 3 3 270 13 10 4 11 10 9 8 28 27 3 66 cm (active Si) Si X-view Si Y- view ……

12. Some very rough estimates … Silicon – STK: 7x7x2x5 = 490 ladders, 3430 SSD, ~31 m 2 – “PANGU”: 7x20x2 = 280 ladders, 1960 SSD,, ~18 m 2 Weight – Tungsten: 7x7x4x5 = 980 plates (same size as SSD), 8844.5 cm3, 170 kg – 50 kg support for each STK + 8 kg each ACD – 20 kg total for “PANGU” – Total : 170 + 58x5 + 20 = 480 kg + 25% margin = ~600 kg Readout channels and power consumption – STK: 768x490 = 376320 channels, assume 1 mW/cha (DAMPE) 376 W Can be reduced by going to 0.18µm ASICs (VA) – PANGU: 768x280 = 215040 channels, assume 0.3 mW/cha (TAA1) 65 W First 6 layers should use VA instead for charge measurement : +15 W Trigger and readout: 20 W (PANGU) – ACD: 10 W – Total: 376 + 80 + 15 = 471 W + 25% margin = ~600 W 12Xin WuHERD Workshop, Beijing, 19-20/1/2016

13. 13Xin Wu Some numbers of dimension It would be useful to know if the CSS is blocking some angular ranges Need larger top STK to increase coverage (8x8 or 10x10 SSD? )

14. 14Xin Wu Need larger side STK to increase coverage (8 SSD ladder?)

15. Available Space before mounting the ACD 15Xin Wu How to increase angular coverage? Obvious choice: increase the size of the top STK by 3 SSD (PANGU part unchanged) – 5-SSD ladders, number of ladders 98 280 Add 182 ladders, 140k channels, 140 kW – Recover ~4x5° coverage – Routing of the electronics would be complicated Additional: make 2 of the side STK larger – 5-SSD ladders on x, 7-SSD ladders on y Add 182 ladders, 140k channels, 140 kW – Recover ~4x5° coverage – Very difficult to route the electronics For large size tracker, scintillating fiber has a big advantage in cost and power (but charge measurement could be a problem!) Intermediate solution: DAMPE 4-SSD ladders with charge sharing, 8x8 SSD/layer

16. Performance Studies Detector simulation with Geant4-10.1.2 Only top tracker (“DAMPE” part and “PANGU” part) and calorimeter are simulated – ACD not implemented Tracker uses the DAMPE SSD geometry, including guard ring, inter-SSD distance, etc. – A ladder is made of 7 SSD – Readout pitch is 121 µm, no floating strip charge sharing, analog readout – Tracker layers placed as described in page 8 and 9 Only silicon and tungsten are implemented, support structure material ignored Calorimeter implemented as BGO bars – Only the total amount of energy deposited in calorimeter is used in the analysis Distance between sensitive surfaces of calorimeter and track is 5 cm – Also simulated 12 cm for comparison 16Xin WuHERD Workshop, Beijing, 19-20/1/2016

17. Event generation and detection Photons generated from a flat surface with 3 angular ranges – Normal incidence: cos  > 0.975 (  <12.84°) –  = 30° –  = 50° Filtering: only interacting events are selected Further selection – Photon converted in the tracker – Both electron and positron have at least 6 matching silicon clusters – Both electron and positron tracks are found by the Kalman filter with perfect pattern recognition (only matched clusters are fed to the filter) – At least 70% of the photon energy is deposited in the detector (tracker + calorimeter) Effective area and Point Spread Function are compared – Two different method of photon direction reconstruction Leading track Vector sum of electron and positron tracks weighted by Gaussian smeared energy (30%) 17Xin WuHERD Workshop, Beijing, 19-20/1/2016

18. Pair Opening Angle Leading track gives a as good PSF as the energy weighted measurement above a few GeV 18Xin WuHERD Workshop, Beijing, 19-20/1/2016 gamma ray electronpositron

19. PSF: Converted in DAMPE ~ 0.15° @ 10 GeV, ~0.8° @ 1 GeV, ~7° @ 100 MeV – Fermi: ~ 0.15° @ 10 GeV, 0.7° @ 1 GeV, 5° @ 100 MeV 19Xin Wu PSF Comparable to Fermi!

20. PSF: Converted in PANGU PSF improves x3, x4, x6 at low 10 GeV, 1 GeV, 100 MeV – ~ 0.05° @ 10 GeV, ~0.2° @ 1 GeV, ~1° @ 100 MeV 20Xin Wu

21. PSF: Converted in PANGU 21Xin Wu Measure the energy of each track to ~30% improves PSF at low energy by 20% – Using 150 µm SSD improves the PSF by 25% at low energy (but ~½ of Eff. Area) HERD Workshop, Beijing, 19-20/1/2016

22. Effective Area for different selections 22Xin Wu “PANGU” ≈ 1/6 “DAMPE Good for diff. or trans. At normal incidence, above 10 GeV, energy dependence is weak – Sharp decline below 1 GeV (larger opening angle and energy absorbed by W) Some acceptance probably can be recovered with reduced PSF

23. Effective Area, linear scale 70% containment > 10 GeV: ~1900 cm2 for “DAMPE”, ~300 cm2 for “PANGU” – > 1 GeV: ~1900 cm 2 for “DAMPE”, ~300 cm 2 for “PANGU” – > 400 MeV: “PANGU” is similar to ESA-CAS PANGU of same SSD thickness 23Xin Wu Fermi: 8000, 7000, 2400 cm 2 for 10 GeV, 1 GeV, 100 MeV Comparable to Fermi if x4 (+ 4 sides)

24. Off-axis Effective Area, converted in “DAMPE” Effective area decreases with incident angle because of the calorimeter is smaller – 100 GeV: 1900 cm 2 on-axis, 1400 cm 2 at 45° 24Xin WuHERD Workshop, Beijing, 19-20/1/2016

25. Off-axis Effective Area, converted in “PANGU” Even bigger drop for “PANGU” 25Xin WuHERD Workshop, Beijing, 19-20/1/2016

26. Off-axis PSF: Converted in DAMPE Small angle dependence > 10 GeV, larger (~25%) at lower energy – 1 GeV: 0.8° on-axis, 0.9° at 30°, 1.0° at 45° – 100 MeV: 6.8° on-axis, 7.4° at 30°, 8.2° at 45° 26Xin Wu

27. Off-axis PSF: Converted in PANGU Less sensitive (+16% at 45°) to angles than DAMPE because of less material – 1 GeV: 0.18° on-axis, 0.19° at 30°, 0.21° at 45° – 100 MeV: 1.12° on-axis, 1.17° at 30°, 1.30° at 45° 27Xin Wu

28. Effective Area vs. Calo-STK distance, DAMPE Loss of Effective Area if Calo-STK distance is large, in particular at large angle – Up to ~17% at 45° 28Xin WuHERD Workshop, Beijing, 19-20/1/2016

29. Effective Area vs. Calo-STK distance, PANGU Situation is worse for PANGU – Up to ~28% at 45° 29Xin WuHERD Workshop, Beijing, 19-20/1/2016 Should try to reduce as much as possible the distance calo-STK!

30. Conclusions We have implemented a layout of HERD into CAD including a low energy part (“PANGU”) on the top to check the envelops – Very challenging to cover all solid angles – “Economical” solution with 7-SDD ladders, same STK on all five sides Propose use 4x 1mm tungsten plates to have a PSF comparable to Fermi – Effective Area is also comparable with the “economical” layout The PANGU part has similar performance as the ESA-CAS PANGU above 400 MeV – Very limited sensitivity below 200 MeV because of large opening angle Hard question: What is the optimal balance between PSF and Effective Area, for both high and low energy, from the science point of view? – DM search vs.  -ray astronomy? – Probably should not optimized too much for “PANGU” given that HERD cannot point, and has small FOV? Would be very useful to simulate the sky coverage of HERD on CSS 30Xin WuHERD Workshop, Beijing, 19-20/1/2016 Long ladder noise performance to be demonstrated!

31. THANK YOU!

32. Effective Area vs. W thickness, DAMPE Loss of Effective Area at high energy because of conversion probability – Up to ~40% at 100 GeV 32Xin WuHERD Workshop, Beijing, 19-20/1/2016

33. Effective Area vs. W thickness, PANGU Better Effective Area below 1 GeV at low energy with 0.5 mm, ~x2 at 100 MeV! – Because less energy loss in W so more passed the energy containment cut 33Xin WuHERD Workshop, Beijing, 19-20/1/2016

34. PSF vs. W thickness, DAMPE 30-40% improvement of PSF between 100 MeV and 50 MeV 34Xin WuHERD Workshop, Beijing, 19-20/1/2016