TORCH: a novel detector combining TOF and RICH

TORCH (Time Of internally Reflected CHerenkov light)is a possible solution for low-momentum particle ID under study for the upgrade of the LHCb experiment

Closely related concept to the TOP of Belle II [Toru Iijima] — a new generation of PID devices profiting from fast photodetectorsTORCH is at an earlier stage, but is aiming for higher resolution

Roger Forty (CERN) on behalf of the LHCb RICH group

Int. Workshop on Probing Strangeness in Hard Processes, Frascati, 18–21 October 2010

1. The LHCb upgrade

2. TORCH concept

3. TORCH R&D

Roger Forty The TORCH detector concept 2

1. The LHCb Upgrade• LHCb is one of the four major experiments at the LHC, dedicated to the

search for new physics in CP violation and rare decays of heavy flavours

• It is a forward spectrometer (10–300 mrad) operating in pp collider modeParticle identification provided by two RICH detectors [Clara Matteuzzi]Currently with three radiators: silica aerogel, C4F10 and CF4 gas

RICH-1 RICH-2

Roger Forty The TORCH detector concept 3

LHC luminosity• LHCb was commissioned ready for the LHC startup in 2008

After some teething trouble the LHC is now performing excellently

• Peak luminosity 1032 cm-2s-1 achievedIntegrated L ~ 20 pb-1 recorded so farTarget for next year: 1 fb-1

• The nominal luminosity for LHCb is only few 1032 cm-2s-1, to maximize events with single pp interactions

• First phase of LHCb will run for next ~ five years, integrating 5–10 fb-1

• Plan to then upgrade the experimentas doubling time would get too longAim to increase luminosity by 10(will already be available from machine)

Exponential increase 1.5 per week!

Roger Forty The TORCH detector concept 4

Current performance• Detector is performing superbly: clean b-hadron signals accumulating rapidly

B → hh data 3 pb-1 (without PID cut)

B → hh signal Monte Carlo

Applying particle ID cuts with the RICH system, can select a cleansignal for Bs → KK

(first mass peak for this mode)

→ Excellent K- separation at high p

Roger Forty The TORCH detector concept 5

Low-momentum PID• Flavour tagging (distinguishing B from B) is one of the primary

requirements for low-momentum particle ID in LHCb (2–10 GeV) currently provided by aerogel

• Can now be studied in data using B0–B0 oscillations

• Monte Carlo studies of high-luminosity running indicate that aerogel will be less effective, due to its low photon yield (< 10 p.e./saturated track) and the high occupancy environment

1st Phase Upgrade

Roger Forty The TORCH detector concept 6

Upgrade plan • Need to prepare for upgrade even though experiment has only just begun to

accumulate data, due to long lead time (R&D + construction + installation)

• Aim for installation of upgrade in 2016, during a planned LHC shutdown

• Main focus is on trigger, which must be upgraded to handle higher luminosity Current bottleneck is hardware level that reduces 40 MHz bunch crossing rate to 1 MHz for readout into HLT → read out complete experiment at 40 MHz into the CPU farm, fully software trigger

• RICH system will be kept for PID with photodetectors replaced Propose to replace the aerogel with time-of-flight based detector

• First muon station will be removed → space available for new device

Roger Forty The TORCH detector concept 7

2. TORCH concept• Want positive identification of kaons in region below their threshold for

producing light in the C4F10 gas of RICH-1, i.e. p < 10 GeV

• Difficult to achieve with a RICH system (aerogel was the best choice of radiator for this region), so possibility of time-of-flight investigated

TOF () = 35 ps at 10 GeVover a distance of ~ 10 m→ aim for 15 ps resolution per track

• Difficult to achieve with scintillator or other traditional TOF

• Cherenkov light production is prompt→ use quartz as source of fast signal

• Large-area fast photodetectors under development by the Picosecond timinggroup (http://psec.uchicago.edu/) but unlikely to be available in time for our application (and we would need ~ 30 m2!)

Roger Forty The TORCH detector concept 8

DIRC-like detector• Consider instead a first (naïve) design

based on quartz bars, à la DIRC of BaBar:Cherenkov photons produced in the quartz transported to the end of the bar by total internal reflection, where their arrival would be timed

• 1 cm thickness of quartz is enough to produce ~ 50 detected photons/track (assuming a reasonable quantum efficiency of the photon detector)

→ ~ 70 ps resolution required per detected photon

• However, spread of arrival times is much greater than this, due to different paths taken by photons in the bar

25 ns

3 m

Photon arrival time

Roger Forty The TORCH detector concept 9

Planar detector• Need to measure angles of photons, so their path length can be reconstructed:

~ 1 mrad precision required on the angles in both transverse planes

• This would be prohibitive for a set of quartz bars, but borrow nice idea from the end-cap DIRC of PANDA [Matthias Hoek]: use a plane of quartz → coarse segmentation (~ 1cm) is sufficient for the transverse direction (x)

~ 1 cm

Roger Forty The TORCH detector concept 10

Focusing system• To measure the angle in the longitudinal direction (z) we use a focusing

block, to convert angle of the photon into position on the photodetector

• Event display illustrated for photons from 3 different tracks hitting plane

Roger Forty The TORCH detector concept 11

Photon detection• Micro-channel plate (MCP) photodetectors are currently the best choice

for fast timing of single photons

• Anode pad structure can in principle be adjusted according to need

• Test result from K. Inami et al [RICH2010] (t) = 34.2 ± 0.4 ps

Faceplate

Photocathode

Dual MCP

Anode

Gain ~ 106

photoelectron V ~ 200V

V ~ 200V

V ~ 2000V

photon

Faceplate

Photocathode

Dual MCP

Anode

Gain ~ 106

photoelectron V ~ 200V

V ~ 200V

V ~ 2000V

photon

~10 m pores in the MCP

Roger Forty The TORCH detector concept 12

Modular design• For the application in LHCb, transverse dimension of plane to be

instrumented is ~ 5 6 m2 (at z = 10 m)

• Unrealistic to cover with a single quartz plate evolve to modular layout:

• 18 identical moduleseach 250 66 1 cm3

~ 300 litres of quartzin total (less than Babar)

• Reflective lower edge photon detectors only needed on upper edge

18 11 = 198 unitsEach with 1024 pads 200k channels total

Roger Forty The TORCH detector concept 13

Effect of edges• Reflection off the faces of plate is

not a problem, as the photon angle in that direction (z) is measured via the focusing system

• In the other coordinate (x) position is measured rather than angle → reflection off the sides of the plate gives ambiguities in the reconstructed photon path

• Only keep those solutions that give a physical Cherenkov angle → only ~ 2 ambiguities on average

• Effect of the remaining ambiguities is simply to add a ~ flat background to reconstructed time distribution

Roger Forty The TORCH detector concept 14

TORCH module• Focusing block in

quartz or plastic (should match refractive index)

• Cylindrical mirror

• Linear array ofphoton detectors

• Dimensions have been chosen to correspond tothe Planacon MCPfrom Photonis

• Plate thickness (~ 1 cm)to be optimized once p.e. yield known

Roger Forty The TORCH detector concept 15

Photon detector• Planacon XP85022 comes close

to matching photodetector requirements for TORCHCurrently availablewith 32 32 anode pads

• We require finer granularity in one direction than other, so assume an 8 128 anode pad layoutIn discussion with manufacturers to secure this development

• Lifetime of MCP may also be an issue for our application (depends on gain, and hence electronics)Following recent development of longer-lived MCPs [Hamamatsu] with great interest

Roger Forty The TORCH detector concept 16

Resolution• Smearing of photon propagation time

due to photodetector granularity ~ 40 ps

• Assuming an intrinsic arrival time measurement resolution per p.e. of 50 psthe total resolution per detected p.e. is 40 50 70 ps, as required

• For particle ID, need to correct for the strong chromatic dispersion of quartzAchieved by measuring the photon angles, and knowing path of track through quartz determine Cherenkov emission angle

cos C = 1/ nphase

t – t0 = L ngroup /c

Effectively the wavelength of the photon is determined by this construction

Roger Forty The TORCH detector concept 17

Performance• Different time-of-propagation for photons from or K, but this effect adds

to difference in time-of-flight increases the sensitivity

• To determine the time-of-flight, we also need a start time t0

This is achieved using the other tracks in the event, from the primary vertexMost of them are pions, so the reconstruction logic is reversed, and the start time is determined from their average assuming they are all (outliers from other particle types are removed)

• Full algorithm has been studied,including pattern recognition,using a simple simulation of theTORCH detector, interfacedto the full simulation of LHCb

• Excellent particle ID performanceachieved, up to 10 GeV as required

Roger Forty The TORCH detector concept 18

3. TORCH R&D • R&D has been launched on the following aspects:

1. PhotodetectorPerformance of existing MCP devices; Development of suitable anode pad structure; Lifetime; Cost

2. Readout electronicsSpeed; 40 MHz rate; Gain; Noise; Cross-talk

3. Quartz radiatorPolishing; Required quality for total internal reflection; Cost

4. Simulation Detailed simulation of TORCH; tagging performance in upgrade

• Two 64-channel Planacon MCPs procured from PhotonisCharacterisation with laser light source, starting with single channel electronics, then multichannel readout

• First results expected in time for Letter of Intent at the end of this year

Roger Forty The TORCH detector concept 19

Lab setup at CERN

Single channel electronics

Dark box

PlanaconMCP

Laser light source

Roger Forty The TORCH detector concept 20

Readout electronics• Under development by Oxford Univ. group

• Starting with 8-channel NINO chips and HPTDC, developed for the ALICE TOF

• Test-beam studies foreseen for next year

Spartan 3AN

Gigabit Ethernet PHY

SRAM / SDRAM

Optional

Clk Buf

SPI Flash

FPGA JTAG

HPTDC JTAG

HPTDC

HPTDC

Ext Clk

Hits

Hits

Shared data bus

Control bus NINO

NINO

MC

P Connection

Trigger

JTAG

Electronics board support for tests

Roger Forty The TORCH detector concept 21

Conclusions• TORCH is a novel detector concept

proposed for the upgrade of LHCbIt is intended to complement the high-momentum particle ID provided by the RICH system

• Based on time-of-flight, determined from Cherenkov light produced in quartz plateusing photon detectors at the periphery

• Assuming a per-photon resolution of 70 ps excellent K- separation achieved up to 10 GeV

• R&D is in progress, starting with the photodetector and readout electronicsImpact of the TORCH on tagging performance in the upgraded experiment is under study with detailed simulation

• Letter of Intent for the LHCb upgrade will be submitted at end of this year

Isolated tracks