1. Fast timing for a future upgrade TORCH is a detector concept intended for fast timing of charged particles over large areas: currently an R&D project, not yet officially part of LHCb upgrade(s) Discussions in progress to adapt it for generalized fast timing, including photons, in conjunction with the calorimeter upgrade Theatre of Dreams: LHCb beyond the phase 1 upgrade, 6 April 2016 Roger Forty (CERN) 1.Progress of the TORCH R&D project (Neville) 2.Ideas about broadening its use (this talk) 3.First results from simulation studies (Maarten)

2. Fast timing Fast timing is currently a hot topic for the upgrades of all LHC experiments Some slides extracted from recent presentation at ACES 2016 - Fifth Common ATLAS CMS Electronics Workshop for LHC UpgradesACES 2016 - Fifth Common ATLAS CMS Electronics Workshop for LHC Upgrades Roger FortyFast timing for a future upgrade2 https://indico.cern.ch/event/468486/session/5/contribution/21/ attachments/1240096/1826154/ACES_LindseyGray_08032016.pdf

3. Roger FortyFast timing for a future upgrade3

4. Time slicing of order 25 ps proposed, to reduce effective occupancy Roger FortyFast timing for a future upgrade4

5. Similarly extensive plans also discussed for ATLAS Roger FortyFast timing for a future upgrade5

6. Four candidate detector technologies considered Roger FortyFast timing for a future upgrade6

7. Interest in LHCb Luminosity will be lower (probably), but for the values being discussed (> 10 33 cm -2 s -1 ) it will still be important to resolve pile-up Fast timing first proposed by Sheldon for photon timing, following detector technology [3] on previous slide Large-Area Picosecond Photo-Detector (LAPPD) collaboration in US has been aiming to develop 8”x 8” (20 x 20 cm 2 ) MCP detectors Problems experienced sealing such a large tube, collaboration ended in 2012 [arXiv:1603.01843] Now being industrialized by Incom: first working prototypes still some years away Roger FortyFast timing for a future upgrade7

8. LAPPD-style solution Roger FortyFast timing for a future upgrade8 [M. Wetstein] Current design is read out with strip-line anodes (30/tube) which could be chained together Track impact position determined by timing difference at two ends (and centroiding in other projection) – could reach few mm resolution Suitable for sparse applications such as neutrino detectors, but issue of occupancy for busy environment of LHC upgrades? Interesting R&D, which should be followed, but not ready in near future Covering full LHCb acceptance at the calorimeter (z = 12 m) would require over 1000 such large tubes… To reduce the photon detector count, idea raised of adapting TORCH i.e. use a DIRC-like technique to bring photons to edge of acceptance → around 200 2”-tubes, total photocathode area 0.5 m 2 (i.e. ÷ 100)

9. Isolated tracks TORCH principle DIRC-style detector concept, with a ~1 cm thick quartz radiator plate: Cherenkov light produced in the plate propagates to the edge by TIR and is focused via a cylindrical lens onto fast photon detectors Reconstruction of the Cherenkov angle at emission allows the propagation time in the radiator plate to be corrected for dispersion Requires precise track information (~ 1 mrad) to achieve timing resolution of ~ 70 ps/photon → 15 ps/track by combining ~ 30 detected p.e./track Originally conceived for low-momentum PID (replacing aerogel in LHCb) Over 10 m TOF would give > 3σ K/π separation up to 10 GeV Roger FortyFast timing for a future upgrade9 (schematic)

10. Event display Typical event in TORCH detector in simulation of sterile neutrinos (low multiplicity) Photons colour- coded to match their parent track Roger FortyFast timing for a future upgrade10    Track impact points on quartz plate Photon impact points on detectors along each edge (  z vs. x) without dispersion

11. Reconstruction Effect of smearing from dispersion: Use timing information as well as spatial information from detector to separate signals from each track Calculate time of propagation of all photons relative to the blue track (  ) Hits from that track peak at true time Hits from other tracks spread out (but peak in time distribution when it is calculated relative to that track) Roger FortyFast timing for a future upgrade11 Refractive index vs. E 

12. LHCb event Roger FortyFast timing for a future upgrade12 Zoom on vertex region Track impact points on TORCH Typical LHCb event, at luminosity of 10 33 cm -2 s -1 (only photons reaching the upper edge shown) High multiplicity! >100 tracks/event Tracks from vertex region colour- coded according to the vertex they come from (rest are secondaries) K

13. Modular design Roger FortyFast timing for a future upgrade13 For the application in LHCb, transverse dimension of plane to be instrumented is ~ 5  6 m 2 (at z = 9.5 m) + central hole for beam pipe Unrealistic to cover with a single quartz plate  developed modular layout: 18 identical modules each 250  66  1 cm 3  ~ 300 litres of quartz in total (less than BaBar) Reflective lower edge  photon detectors only needed on upper edge 18  11 = 198 detectors Each with 1024 pixels  200k channels total

14. Roger FortyFast timing for a future upgrade14 Effect of modules Illustrate the effect of introducing modules, using the same LHCb event Far fewer hit-track combinations, but reflections from sides give ambiguities Module considered Without dispersion or reflection off lower edge Including dispersion and reflection off lower edge

15. Performance As for the previous application, signals can be isolated using the time info Calculate time for all hits in module with respect to the kaon track: Expected performance for low-momentum particle ID from full simulation: good, and robust to increasing luminosity (shown dashed) Study shown was made at the time of the upgrade LoI, for a single-plate, should be updated with the modular layout Roger FortyFast timing for a future upgrade15 LHCb simulation: efficiency vs. p K→KK→K →K→K

16. Full-scale prototype Final goal of the ERC-funded R&D project is to prototype a TORCH module which accommodates 10 MCP-PMTs of 128 × 8 effective granularity As discussed by Neville, R&D is on track to produce suitable MCP-PMTs Size of the quartz plate for the final prototype will be 60 × 120 × 1 cm 3 Test for radiator plate of full-scale prototype module in progress in industry (although order not yet sent…) Roger FortyFast timing for a future upgrade16

17. Conceptual layout in LHCb We have been offered to re-use the BaBar DIRC quartz radiator: up to 8 bar-boxes could be available for use in LHCb if the prototype is successful Assume located at z = 9.5 m → bar-boxes just fit acceptance Their use would reduce the area to be instrumented with new quartz by ~ 2/3 Roger FortyFast timing for a future upgrade17 New radiator (in high occupancy region) Could have different module layout, or even be glued together to form plane BaBar DIRC

18. Alternative layouts Nominal position of TORCH at z = 9.5 m chosen to be as close as possible to tracker and to reduce area to be covered (~ 30 m 2 ) TORCH radiator ~ 1 cm thick quartz (8% X 0 ) However, will need to be supported and enclosed in light-tight box, so total thickness will be greater (RICH-2 might need to be moved) Moving full detector to z ≈ 12 m would require increased area of 48 m 2, and 30% more MCPs (but TOF increases too…) Would make BaBar bar re-use less attractive, as no longer fit acceptance Using timing to suppress pileup → require timing resolution to separate PVs Resolution needed is O(5 mm), i.e. O(15 ps), well matched to TORCH goal High momentum particles are mostly at small angles, could imagine a hybrid arrangement if only central part of calorimeter is upgraded Roger FortyFast timing for a future upgrade18

19. Use for timing photons Roger FortyFast timing for a future upgrade19 Since TORCH detects Cherenkov light from charged particles, detection of photons requires conversion to e + e – Assume layer of converter material O(1 X 0 ) placed in front of TORCH –Too thin and efficiency of conversion will be low –Too thick, and shower will develop, degrading angular precision Even if only single pair produced, tracks will undergo multiple scattering Calorimeter cluster from pair should give sufficient angular precision for TORCH reconstruction, assuming photon originated at IP (to be checked with simulation)

20. Charged particles too Interactions (marked with ) at same z but separated by 300 ps → Essential that charged particles are timed too (but TORCH would do that) Roger FortyFast timing for a future upgrade20 z z = 0 t = -300 ps Vertex information alone is not sufficient: particles produced at same position can have very different production times Consider two beam bunches crossing at the interaction point: z t = 0 z t = +300 ps

21. First studies To avoid the two Cherenkov cones of e + e – interfering destructively, need to be separated by O(λ C /2): OK if converter and radiator > few mm apart For high energy photons (> few GeV), angular correlation of e + e – with photon direction is expected to be good enough: see next talk For low momentum tracks of interest for K/π separation (< 10 GeV) multiple scattering from extra material degrades performance: Roger FortyFast timing for a future upgrade21 Distance of track impact from beamline –For 1 X 0, at 10 GeV expect ~ 1.4 mrad of angular straggling –In a simplified simulation, 4 σ K/π separation is reduced to 1 – 3 σ depending on distance of propagation in radiator –Difficult to recover angular resolution after scattering using calorimeter, but using redundant information from TORCH itself might help… [Klaus Föhl]

22. Conclusions Using a TORCH-like detector for fast timing of photons as well as charged particles may be interesting for a future upgrade of LHCb –How important is pile-up suppression for the physics? A single TORCH detector providing both K/π separation at low momentum and timing photons is challenging, due to effect of the converter material –Might require separate systems, or perhaps a hybrid The idea of re-optimizing TORCH for pileup suppression at high-luminosity i.e. for timing both photons and charged tracks at higher momentum (with K/π separation as a secondary goal) is worth further exploration Various issues need to be addressed with detailed simulation –Performance for photon timing in full events –Optimization of converter thickness –Occupancy vs. luminosity, etc. First results from these studies shown in the following presentation Roger FortyFast timing for a future upgrade22