DECAL: DMAPS S ENSORS FOR D IGITAL E LECTROMAGNETIC C ALORIMETRY S - PowerPoint PPT Presentation

DECAL: DMAPS S ENSORS FOR D IGITAL E LECTROMAGNETIC C ALORIMETRY S TEVE W ORM A , P. A LLPORT A , R. B OSLEY A , L. C HEN B , J. D OPKE B , S. F LYNN A , L. G ONELLA A , I. K OPSALIS A , K. N IKOLOPOULOS A , P. P HILLIPS B , T. P RICE A , I. S EDGWICK B , G. V ILLANI B , N. W ATSON A , M. W ARREN C , F. W ILSON B , A. W INTER A , Z. Z HANG B A U NIVERSITY OF B IRMINGHAM B STFC R UTHERFORD A PPLETON C U NIVERSITY C OLLEGE L ONDON M AY 29, 2018

O UTLINE AND O VERVIEW • IntroducIon: Calorimetry with silicon readout – Analogue (e.g. CALICE, CMS HGCAL) – Digital (e.g. SPiDER, ALICE FoCal) • SimulaIon results: Single parIcle resoluIons • Sensor R&D: Reconfigurable radiaIon-hard DMAPS • Conclusions Analogue: Sum energy deposited in a cell Digital: Sum the number of particles in a cell 2

D IGITAL C ALORIMETRY : T EST BEAMS [Rocco et al, ICHEP 2016] • SPiDeR: pixel counIng gives correct shower shape – CERN testbeam using EU Telescope (Mimosa-26) – Tracks in first 3 layers, shower in W, measure core in 4th layer 15 cm from W • Digital Calo Example: ALICE FoCal – High granularity to separate showers – DESY & SPS test beams: 24 plane prototype; four Mimosa-23 (30 μm Pitch) with W 3

A NALOGUE R EADOUT • Detector ConfiguraIon – Concentric octagonal geometry Linearity – 50 layers of 0.6 X 0 W – 5x5 mm 2 pixels – 300 μm thick sensiIve region Resolution 5x5 mm 2 Pads • PandoraPFA* gives jet energy resoluIon of <4% for jets up to 250 GeV Excellent linearity, resolu=on possible [*hlps://arxiv.org/abs/0907.3577] 4

D IGITAL R EADOUT • Detector ConfiguraIons – Concentric cylinder geometry – 50x50 μm 2 pixels – 18 μm thick epitaxial region ▼ 50 layers of 0.6 X 0 W ○ 30 layers of 1.0 X 0 W ☐ 50 layers of 0.6 X 0 Pb � 30 layers of 1.0 X 0 Pb • MIP MPV ~1400 e- with threshold of 480 • For 50 layers of Tungsten, energy linear up to ~300 GeV 5

D IGITAL R EADOUT • Octagonal barrel geometry – 50 layers of 0.6 X 0 W Linearity – 50x50 μm 2 digital pixels – 5x5 mm 2 counIng pads – 18 μm thick epitaxial region – 3 mm air for PCB, services, etc Resolution 50x50 µ m 2 Pixels Good performance for realis=c geometry 6

N ON -L INEARITY AND MVA • Can you recover linearity in soqware? – MVA package: LinearDiscriminant, MulI-Layer Perceptron (ANN) – Simulated DECAL: energy range 10 - 1000 GeV, 50 layers 2.1 mm W – Input: number of pixel hits per layer, grouped in 5 layers (1-5, 6-10, etc) Linearity recovered over full range (to 1 TeV) 7

DECAL P ILEUP S TUDIES • Simulated pile-up using Pythia min bias events – ⟨ µ ⟩ =1000 yields 3.5M pixels above threshold in DECAL barrel – Energy in first layer vs eta follows same trend as FCC-hh LAr baseline studies – Pixels above threshold vs eta also have similar trend decal_ParticlesPerEta_layer1 decal_ParticlesPerEta_layer1 • 500 GeV e - ; 50k hit pixels swamped by pileup Edep / 0.1 eta / event Entries Entries 290 290 0.16 Mean Mean 1.015 1.015 0.16 Mean y Mean y 0.04408 0.04408 E / 0.1eta / event Std Dev Std Dev 0.5857 0.5857 0.14 Std Dev y 0.03913 Std Dev y 0.03913 – First alempts at reducing hits very promising 0.12 0.12 0.1 – ALL PIXELS, MAX_STAVE, 10k ROI, TRUTH ⟨ µ ⟩ =0 0.08 0.08 0.06 – Simple cuts reduce BG by orders of magnitude 0.04 0.04 0.02 0 • Next: careful study with clustering, energy resoluIon 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.4 0.8 1.2 1.6 2.0 Eta Eta decal_ParticlesPerEta_layer1 decal_ParticlesPerEta_layer1 Pixels / 0.1 eta / event Entries Entries 290 290 pixels / 0.1eta / event 8000 Mean Mean 1.015 1.015 8000 10 5 Mean y Mean y 2632 2632 Std Dev Std Dev 0.5857 0.5857 7000 Std Dev y Std Dev y 2029 2029 pixels / layer 10 4 6000 6000 5000 10 3 4000 4000 10 2 3000 2000 2000 10 1000 0 1 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 10 2 0 30 40 50 0 0.4 0.8 1.2 1.6 2.0 Eta Layer Eta 8

DECAL T EST S ENSOR • DECAL monolithic acIve pixel test sensor features – Reconfigurable: strip mode for parIcle tracking, or pad mode (counts above threshold) for digital calorimetry – TowerJazz 180 nm, 18 μm epi, standard (opIcal) process – 5 mm 2 pixel matrix of 64×64 pixels with pitch of 55×55 μm 2 • Sensor matrix now under test – Four collecIon electrodes/pixel – Trim, preamp, shaper, discriminator – 25 ns digital output – Analog and digital tests ongoing Sensor shows analogue response (works!) 9

DECAL A RRAY : S TRIP M ODE • Strip Mode (1 x 64 pixel array) – Counts above threshold: 0, 1, 2 or >3 per column – Data rate: 320 Mbit/s x 16 = 4.8 Gbit/s Test Data Shift Register Calibration Shift Register 64 strip array Bias Mapping (64 x 64 pixels) Collated Column Outputs PLL Strip Logic Pad Logic 10 LVDS Output

DECAL A RRAY : P AD M ODE • Pad Mode (16 x 64 pixel array) – Up to 15 hits in each of 16 columns (240 total) – Lower rate; about 1/4 the LVDS lines Test Data Shift Register Calibration Shift Register 64 x 64 4 pad array pixel array Bias Mapping (each 16 columns wide) Collated Column Outputs PLL Strip Logic Pad Logic 11 LVDS Output

DECAL P IXEL 2V 4V 6V 8V 10V Single pixel field: DC Vbias 2 to 10V, without back contact • TowerJazz 180 nm standard (opIcal) process; 18 μm epi – Four collecIon nodes, low capacitance, expect good signal/noise – OperaIonal with 1-2 volts bias; higher voltage for faster collecIon • TowerJazz “modified process” also under invesIgaIon – Test structures submiled in modified TowerJazz process [NIM A V871, 90-96, 2017] – Full depleIon, low capacitance and good signal/noise demonstrated – Rad hard to few 10 15 neutron equiv. [JINST 12 P06008, 2017] Poster: H. Pernegger Talk: K. Moustakas 12

DECAL D ATA A CQUISITION AND S OFTWARE • DECAL Data AcquisiIon via NEXYS Video board (Digilent) • Soqware is based on ATLAS ITSDAQ • NEXYS board is programmed using Adept 13

DECAL: S TRIP M ODE D IGITAL L OGIC • OperaIng the chip in strip mode: – Inject 3 (binary 11), and shiq along the strips – Inject 3 (11), then 2 (10), then 1 (01), shiq along the strips Digital logic for strip mode works as expected Inject number 3 (11) Inject numbers 3, 2, 1 in the 1 st strip in the 1 st , 2 nd and 3 rd strip 14

DECAL: P AD M ODE D IGITAL L OGIC • OperaIng the chip in pad mode: – Inject the numbers 3, 2, 1 and shiqing along the strips – Compare the injected numbers 3, 2, 1 to the total sum of each output block Digital logic for pad mode also works as expected Inject numbers 3, 2, 1 Inject numbers 3, 2, 1 in the 1 st , 2 nd and 3 rd strip and total sum 15

A NALOGUE P IXEL T EST • Laser illuminaIon for analogue pixel in the top leq corner (2 nd row 2 nd column) • Laser-inject charge, observe the outputs of the test pixel • Preamplifier, shaper signals & laser trigger measured 16

A NALOGUE P IXEL T EST • Analogue tesIng of single pixels ge~ng started – Tests vs bias (e.g. 1 or 2 V, below), linearity, etc – Laser scan across matrix – Source tesIng & TCT planned 17

C ONCLUSIONS DECAL silicon calorimeter designs are progressing: • High granularity should allow excellent PFA • Pixel counIng gives good energy resoluIon at intermediate energies • Problems from pileup and mulIple hits can be miIgated by MVA, clustering • Prototype reconfigurable CMOS DMAPS for ECAL, pre-shower, and outer tracking fabricated at TowerJazz and under evaluaIon • Digital logic works (configurable), analogue now under test 18