CMS Electromagnetic Calorimeter and V-V fusion processes detection (general aspects of my research activity and status of my thesis) I.N.F.N. Turin

Summary (I) Electronics of Electromagnetic Calorimeter physics goals of CMS the CMS detector the CMS Electromagnetic Calorimeter (ECAL) ECAL read-out chain components motherboard testing VFE testing Weak bosons fusion for Higgs detection Higgs production in LHC Weak vectors fusion PHASE MonteCarlo simulation Signal extraction from PHASE generated sample 19/01/06

Higgs boson channels The experiment CMS and ATLAS at LHC have as main goal to detect new particles: Higgs boson(s), predicted by symmetry-breaking model supersymmetric companions of known particles, predicted by supersymmetric models LHC parameters: cross section [mb] 100 (60 inelastic) energy 2× 7.0 TeV particles/bunch 1.15·1011 bunches 2808 crossing rate 40 MHz beam size [cm] 5.3 luminosity [cm-2s-1] 2∙1033 ÷ 1034 19/01/06

Compact Muon Solenoid (CMS) An uniform magnetic field of 4 tesla includes Trackers and Calorimeters (Compact!) 19/01/06

Electromagnetic Calorimeter (I) Electromagnetic Calorimeter guide lines: small (it must be placed within magnetic solenoid together with hadronic calorimeter) homogeneous calorimeter of scintillating lead crystals to gather most of released energy high granularity and low Moliere radius to cope with pile-up, allow better spatial resolution, even of photons from π0 decay goal resolution (barrel; energy in GeV): 𝐸 𝐸 = 2.7% 𝐸 ⊕5.5⊕ 200𝑀𝑒𝑉 𝐸 19/01/06

Electromagnetic Calorimeter (II) Coverage: azimuth (φ) direction: full (2π) pseudo-rapidity: |η| ≤ 3 Granularity: ∆η × ∆φ = (0.0175)2 (barrel) Thickness: ~26 X0 Dynamic range: ~16 bit, 30 MeV ÷ 2 TeV Endcap (25 X0) (1.48 < |η| < 3) Preshower (3X0) (1.65 < |η| < 2.6) Barrel (25.8 X0) (|η| < 1.48) Barrel: 36 supermodules with 1700 crystals each, 20 along φ and 85 along η, grouped 4 modules and in 68 5×5 trigger towers η M1 M2 M3 M4 19/01/06

ECAL electronics overview ×5 ×5 ×5 LVR FE VFE MB 19/01/06

Lead crystals and photomultipliers active material PbWO4 average size 22×22×230 mm3 density 8.28 g/cm3 X0 0.89 cm λI 22.4 cm rM 2.19 cm spectrum centred on 420 nm (blue) decay times ~15 ns light yield 50~100 photons/MeV -2%/K Photomultiplier requirements: can operate under a 4T magnetic field radiation hard big gain Avalanche PhotoDiodes (APD): gain: 50 quantum efficiency: 75% but: small surface: 25 mm2 each very sensitive gain: 3.3%/V at 400 V, -2.2%/K 19/01/06

Motherboards Motherboards Connections test designed, managed and tested in Turin connect 5 photodiode pairs, temperature sensors, a Low Voltage Regulator and 5 VFEs distribute high and low voltage respectively from external Power Supply Units and Low Voltage Regulator boards flexible connections (kapton) are employed many capacitors provide power stability they arrive to Turin from manufacturer with no test behind Connections test design: custom board, PC (GPIB) driven checks all 240 connections in about 5 minutes 19/01/06

Motherboards: burn-in test Motherboards are left under 500V high voltage for at least 72 hours to verify the components resistance to stress. During the test motherboards are constantly monitored and disconnected on failure. Measured current (ideally null) decreases fast; part of this current is generated by impurities and it fades out in time; another residual part is made mainly of superficial currents, which can't be removed. Current [μA] in 18 MB in 5 days-run 19/01/06

Very Front-End boards: a close view Very Front-End boards digitise signals coming from APDs and temperatures. Each VFE handles 5 channels, each one made of: a capacitor a Multi-Gain Pre-Amplifier (MGPA), an analogue amplifier with three different gains (1×, 6×, 12×) following a common pre-amplifier; a custom AD41240 ADC: it has 4 channels, 12 bits each; it digitises all three amplified channels from MGPA, then returns the highest, non-saturated value (12 bit) and which that channel was (2 bits) a buffer of 256 words VFEs convert input signals continuously at 40 MHz frequency; each signal is 10 samples long. ADC MGPA buffer To FE To motherboard 19/01/06

Very Front-End boards: calibration Charge [pC] → ADC count 12× : σS/S = 1.24% 6× : σS/S = 1.14% 1× : σS/S = 1.25% 19/01/06

Very Front-End boards: database We have defined a “protocol” for VFE life-cycle; each passage is recorded in a MySQL-based database; there are currently three physical databases in Lyon, Turin and at CERN. 19/01/06

Statistics Motherboard (full process in Turin; 68/SuperModule): all 2650 MB will be ready in July 2006 1300 MB are ready now production and testing rate: 3 SM/month = 210 MB/month problems found: bad soldering (can be repaired in Turin laboratory) inverted kapton soldering bad kapton burn-in failure VFE (CERN, Lyon, Turin; 340/SuperModule): all 12880 barrel VFEs will be ready in Spring 2006 all VFEs are assembled, 7500 are calibrated, 7000 VFEs are ready now no burn-in failure so far (9000 VFEs) functionality test failed on 69 VFEs (0.5%) physical defects (bad assembled housing, broken connectors: 0.8%) barcodes repetition (35 VFEs) 19/01/06

Summary (II) Electronics of Electromagnetic Calorimeter physics goals of CMS the CMS detector the CMS Electromagnetic Calorimeter (ECAL) ECAL read-out chain components motherboard testing VFE testing Weak bosons fusion for Higgs detection Higgs production in LHC Weak vectors fusion PHASE MonteCarlo simulation Signal extraction from PHASE generated sample 19/01/06

Higgs production processes Due to the way Higgs boson couples with particles, the most probable channel is one-loop gluon-gluon fusion, with a top inner triangle, followed by direct production by W+W-/Z0Z0 fusion. 19/01/06

Weak vector bosons fusion Note: Higgs boson compensate a divergence in VL-VL scattering in S.M.; missing H0, a new way must occur to avoid unitariness violation Tags: weak boson decaying into leptons two jets in end-caps Examples of core reactions in V-V fusion processes: 19/01/06

PHASE MonteCarlo generator designed with V-V fusion in mind: qq → qq qq ℓνℓ exact computation of matrix element following S.M., currently at order αem6 thousand diagrams reduced to 16 groups made of 4 sets: 4W, 2Z2W, one b-quark pair and two b-quark pairs processes uses modularity to avoid computing again the same parts of diagrams Higgs boson can be “switched off” some approximations: massless fermions (except b and t quarks), no Higgs-light quark coupling (only H-b) customizable cuts are used; e.g.: electron: E > 20 GeV, pt > 10 GeV/c, || < 3 quarks (jets): E > 20 GeV, pt > 10 GeV/c, || < 6.5, mjj > 10 GeV/c2 forward/backward quarks required (|| >1) 19/01/06

PHASE: V-V fusion processes processes with Higgs boson: V-V fusion with Higgs boson in s or t channel processes without Higgs boson: electroweak V-V fusion (always present!) 19/01/06

PHASE: non-V-V fusion processes PHASE MonteCarlo generates events with an exact computation, which sums up all possible diagrams; it's not possible to neglect or forget unwanted (non V-V fusion) processes during generation, and they must be cut later. 19/01/06

Other background processes There are also some classes of processes that PHASE doesn't simulate at all, which can produce or mimic V-V fusion final state same final states: mimic V-V fusion final states:  𝑠 2  𝑤 4 :𝑔𝑔 𝑞 𝑞 𝑞 𝑞𝑉𝑊  𝑠 4  𝑤 2 :𝑔𝑔 𝑞 𝑞 𝑞 𝑞𝑗𝑗𝑊  𝑠 4  𝑤 2 : 𝑞 𝑞 𝑞 𝑞𝑉𝑊  𝑠 3  𝑤 2 : 𝑞 𝑞 𝑞 𝑞𝑗𝑊 19/01/06

PHASE data format PHASE output is a tagged text file; each event has 8 particles and 12 lines. Incoming quarks are extracted from 7 TeV/c protons, according to CTEQ5L PDF Particles ID, following PDG standard, (here: b b b b u d e νe) Particle direction (-1 = incoming, 1 = outgoing) IDUP -5 5 -5 5 2 -1 11 -12 ISTUP -1 -1 1 1 1 1 1 1 0.000000000E+00 0.000000000E+00 0.874366250E+03 0.874366250E+03 0.000000000E+00 0.000000000E+00 -0.457686024E+02 0.457686024E+02 0.527121540E+02 -0.491175506E+01 0.234792345E+03 0.240686813E+03 -0.106402908E+01 0.475255871E+02 0.253406976E+02 0.538698853E+02 -0.165698050E+02 -0.200266403E+02 0.101175515E+03 0.104461044E+03 0.547681974E+02 -0.652241468E+02 0.946830719E+02 0.127352381E+03 -0.496480074E+02 0.638221486E+02 0.261835690E+03 0.274036713E+03 -0.401985099E+02 -0.211851935E+02 0.110770327E+03 0.119728017E+03 ICOLUP 0 501 502 0 0 501 502 0 503 0 0 503 0 0 0 0 SCALUP 0.100315970E+03 AQEDUP 0.755509995E-02 AQCDUP -0.100000000E+01 Particles 4-momentum (cpx;cpy;cpz;E) [GeV] αem(Q2) αs(Q2) (unused) 𝑄 2 ≡ 𝑚 𝑊 2  1 6 𝑖=1 6 𝑐𝑝 ⊥,𝑖 2 Strong charge streams, passed to hadronisation 19/01/06

Tag quarks cut To cut away processes with incoming quarks annihilation in γ, W± or Z0 (or Higgs!), the first requirement is that event must be compatible with a event topology where each incoming quarks emits a W± or Z0, identifying in this way a pair of “tag quark lines”. In this process, each outgoing particle is assigned a role including “tag quark”, “central quark” (non-tag quarks, deriving from a weak boson) or lepton. Apart from the latter, all assignment suffer from potential ambiguities, which are resolved (arbitrary) choosing the pair closest to W/Z mass as central quarks and the minimum tag quarks scattering angle. Event statistics for: no-Higgs sample cross section (σ) 0.69 pb total PHASE events 400000 without tag quark lines 111079 combinations with same tag candidates 6811 multiple central quarks combinations 105677 fusion candidates 288921 initial quark/antiquark state 174986 19/01/06

PHASE: incoming quarks Distributions of incoming light quarks (d, u, s, c) follow Parton Distribution Functions of proton; bottom quarks presence, instead, is enhanced by the big cross section for events simulated by PHASE, mainly developing into a bb → tt, s- and t-channel processes. Incoming top quarks are simply not included. d t b c s u No-Higgs sample (288k events) Note: particles are indicated in purple, antiparticles in underlined orange. 19/01/06

PHASE: weak bosons “resolution” Invariant mass of “central” quarks (left) and leptons (right) shows that emitted weak bosons are usually on shell, with a width of about 2 GeV for both W± and Z0. mW ΓW mZ ΓZ 19/01/06

Three bosons cut Window: 10 GeV/c2 Cut away events whose tag quarks pair has an invariant mass less than 10 GeV/c2 away from W/Z. Cut results in 270053/288921 events last with ε = 96%; cutting only W has ε = 98% but purity < 99% (leaving ~2705 Z events) 19/01/06

Top quark cut (I) b b t t b b A “top candidate” is a bottom tag quark coupled with the correct W boson; such a couple is labelled “top” if its invariant mass is closer than 20 GeV/c2 to nominal top mass (175 GeV/c2) 19/01/06

Top quark cut (II) Within PHASE, no V-V fusion event can have a top quark: an event is discarded if it has even one top candidate; 100541/270053 events pass the cut, with ε* = 95%. These are my current signal (about 25% of starting sample). Two points per event! 19/01/06

A closer look: weak bosons Leptons (right) and “central” quarks (left) have η direction strongly correlated, meaning that their boson has a non-negligible boost. On last plot the PHASE cut on electrons (compatible with ECAL acceptance) can be seen. 19/01/06

A closer look: pseudorapidity qt,2 qt,1 Cuts have slightly “moved” tag quark closer to beam direction. before cuts (288k) after cuts (100k) qt,2 qt,1 qV 19/01/06

The show must go on the V-V fusion sample (100k events) must be hadronized hadronized products must pass through CMS detectors simulation a proper set of cuts must be defined to maximise the signal against the huge background quantities are to be identified which are sensitive enough to reveal the presence of resonances quantify the resolution on Higgs boson from this channel 19/01/06

(when you haven't enough) Backup slides (when you haven't enough)

Analogue signal path 19/01/06

(hey, you aren't supposed to see these!) Old slides (hey, you aren't supposed to see these!)

Large Hadron Collider (LHC) located in the old LEP tunnels at CERN four main experiments: ALICE, ATLAS, CMS, LHCb two beam pipes for two protons or 208Pb82+ nuclei, 8 cross points Beams data for protons nuclei cross section [mb] 100 (60 inelastic) energy 2× 7.0 TeV 2× 208× 2.76 TeV particles/bunch 1.15·1011 7·107 bunches 2808 ~600 crossing rate 40 MHz 8 kHz beam size [cm] 5.3 7.5 luminosity [cm-2s-1] 2∙1033 ÷ 1034 1.0·1027 19/01/06

Front-End boards Front-End boards (FE) collect data from 5 VFE and compute partial energy sums to be transmitted for trigger generation; on board: 7 FENIX chips, able to support three working modes: sums (5 chips): each chip provides sum of digitized and calibrated energy of 5 crystals (along φ direction) total sum: sum of 5 partial sums plus “fine grain” bit which describes whether energy deposit is very narrow data formatting for transmission a Communication and Control Unit (CCU) for slow control via a double-linked token ring two Gigabit OptoHybrid (GOH) data connections @800 Mb/s for data and trigger transmission To reduce data load, zero suppression is applied on each crystal first; a selective readout is directed by off-detector boards to read only interesting trigger towers. Trigger GOH CCU FENIX Readout 19/01/06

Lead crystals (PbWO4) active material PbWO4 average size 22×22×230 mm3 density 8.28 g/cm3 X0 0.89 cm λI 22.4 cm rM 2.19 cm n 2.3 (θγ > 25 are totally reflected) spectrum centred on 420 nm (blue) decay times 5 ns (39%), 15 ns (60%), 100 ns (1%) light yield 50~100 photons/MeV -2%/K 19/01/06

Avalanche PhotoDiodes Main requirements for ECAL barrel photomultiplier are: to be able to operate under a strong magnetic field radiation hardness big gain Avalanche PhotoDiodes (APD) fulfil these requirements with some drawbacks: based on 200 μm silicon gain: ×150 and over (50 was chosen with ~400V feed) quantum efficiency: 75% but: small surface: 25 mm2 each (two of them are used on each crystal) 4.5 photoelectrons/MeV very sensitive gain: 3.3%/V at 400 V, -2.2%/K As for crystals, temperature control and monitoring are needed. 19/01/06