Gregor Kramberger Jo ef Stefan Institute, Ljubljana on behalf of - - PowerPoint PPT Presentation

▶
gregor kramberger jo ef stefan institute ljubljana
SMART_READER_LITE
LIVE PREVIEW

Gregor Kramberger Jo ef Stefan Institute, Ljubljana on behalf of - - PowerPoint PPT Presentation

Dec 08, 2023 211 likes •473 views

Gregor Kramberger Jo ef Stefan Institute, Ljubljana on behalf of ATLAS HGTD group Motivation Imporatance of HGTD for offline analysis Luminosity meter/Beam monitor HGTD design Location and rates Radiation environment

slide-1
SLIDE 1

Gregor Kramberger

Jožef Stefan Institute, Ljubljana

  • n behalf of ATLAS HGTD group
slide-2
SLIDE 2

 Motivation

  • Imporatance of HGTD for offline analysis
  • Luminosity meter/Beam monitor

 HGTD design

  • Location and rates
  • Radiation environment

 Sensors and electronics

  • Problem of timing measurements
  • Prototype results (source and test beam measurments)
  • Radiation damage
  • Electronics

 Mechanical construction (modules and staves)  Conclusions

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

2

slide-3
SLIDE 3

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

3

8 c countrie ntries, 22 institutions tutions, >120 0 people involved ved

slide-4
SLIDE 4

 HL-LHC upgrade Phase II (2026->)

  • number of pileup collisions will increase to 140-200
  • huge task to assign reconstructed particles to individual collisions and to

extract interesting collisions

  • Is there a way to separate vertices also not only in space but also in t

time?

At LHC the vertices are distributed (Gaussian) with: sz=5 cm & st=180 ps Tracking detectors (ITk-pixel+strip) provide resolution of primary vertices in forward region typically >1 mm, which leads to merging of up to 7 collision vertices It is a task of the HGTD to provide timing resolution of around of 30 ps for minimum ionizing particles (60 ps/mip/layer).

  • n average

ge 1.6-2.35 35 verti tices ces per mm

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

4

ITk HGT GTD

slide-5
SLIDE 5

A goal for the future – to have also tracking in 4D Timing “layer” inserted between tracking detectors and EM calorimeter

HGTD TD

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

5

slide-6
SLIDE 6

A goal for the future – to have also tracking in 4D (superb “Vertexing”)

Timing “layer” inserted between tracking detectors and EM calorimeter

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

6

slide-7
SLIDE 7

Offline e capabilities es - full simulation:

cleaning up the pile up contamination (track fraction) in jets (at 1.6 collisions/mm from 14% to 3%)

up to 15 % improvement of lepton isolation in high pileup environment

20% improvement in forward PU jet suppression signal efficiency – larger fraction of Hard Scatter jets

Resolving primary vertex -> Potential large gains in b-tagging (up to factor of 2), pileup tracks contamination

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

7

slide-8
SLIDE 8

Luminosity meter – measuring the

total number of hits in HGTD:

bunch per bunch measurement (online)

no afterglow problems

easier to spot drifts

Tasks to explore:

amount of data

robust algorithms:

  • acceptance selection
  • linearity

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

8

Initial plans –> include HGTD in L0 trigger ger to separate HS jets from pileup ones (a HS jet is collimated both in time and space) Participation in L0 trigger decision (VBF jet trigger) are not considered dered in the baseline, except using “luminosity mode” for triggering minimum bias events Beam conditi tion

  • n monitor
  • ring:

Timing distribution of hits can be exploited to monitor the cavities performance

t0 re-synchronization : monitor expected timing with measured one for each BCiD (drift) <10-4 stat. uncertainty @ m=200

slide-9
SLIDE 9

Initial studies included variety of sensors, but due to required compactness the choice was limited to Silicon.

A desired timing resolution of the detector is around 30 30 ps ps – a single layer resolution of 60 ps for mip particles –> “making HL-LHC pileup LHC like”

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

9

HGTD-Si: 4 la layers of Silicon pixel/p /pad ad detec tectors tors

(studied also option with W-layers/preshower not considered as baseline HGTD-SiW)

Track isolati tion

  • n efficienc

ency for electrons rons

cleaning pile up contamination

slide-10
SLIDE 10

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

10

5 cm BPE moderator required to shield ITk/HGTD from neutrons – lots

  • f effort in simulation and design

(“optimized” design shown)

slide-11
SLIDE 11

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

11

https://twiki.cern.ch/twiki/pub/AtlasPublic/LArHGTDPublicPlots/Radiation_IDR_last.pdf

Note that ratio od charged hadrons/neutrons in NIEL varies with radii: NIEL p/NIEL n~1 @ R=12 cm NIEL p/NIEL n~0.2 @ R=40 cm Safety factors: x1.5 for fluence and x2.25 for TID (electronics)

Feq,max=6∙1015 cm-2 TIDmax=4 MGy

slide-12
SLIDE 12

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

12

Hit rates determine the cell size (# channels/electronics) –

  • ccupancy < 10% required

1.3x1.3 mm2 pixels are compromise between:

number of channels

pixel capacitance – noise/jitter performance

inefficient area around each LGAD cell

slide-13
SLIDE 13

Several detector options (all Si) investigated: HVCMOS, PIN diodes and LGADs. We focused our efforts to LGADs.

Low Gain Avalanche Detectors: seem to fulfil the requirements (radiation hardness?) and is the only technology having the required maturity and interest.

Gain depends on doping of the multiplication layer.

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

13

Not only LGADs – what is required are thin LGADs (50 mm):

  • We want to have fast rise time -> small jitter

er

  • High gain with proper electronics design also leads to high S/N -> small

l jitter

  • Thin detectors minimize time walk (short drift and saturated drift velocity ~ 1 ns) – smaller effect of

“Landau” fluctuations

  • Thinner detector improve radiation hardness (less leakage, better electric field profile)

Landau fluctuations

mip

thickness Gain I N S t

h e rise jitter TW TDC jitter elec Landau elec t

     

, 2 2 2 2 2 2 2

/ ~ s s s s s s s s

slide-14
SLIDE 14

diffe ffere rent dopin ing g doses es

LGADs have so far been produced by CNM (several runs, RD50), FBK and HPK all proving to work well and can provide the quantities needed for HGTD in time. Most studies performed on CNM and HPK LGADs

  • CNM several runs (R9088 most studied, 3 different doses, 3 different structures)
  • HPK ECX20840 run (4 different doses, 2 different structures)

45 mm thick

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

14

1.3x1.3 mm2 C=2pF

CNM HPK

50,80 mm thick

https://indico.cern.ch/event/637212/contributions/2608660/attachments/1471120/2276430/Kramberger-HPK.pdf

slide-15
SLIDE 15

Timing resolution of these sensors (R9088) was shown to be very good – 26 ps/layer for mip particles!

wideband amplifiers

350 MHz (UCSC) used Quartz (Č photons) +SiPM

DUTs

20 ps resolution Gain=40

Constant Fraction Discrimination

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

15

2 ns

  • N. Cartiglia et al., NIM A850 (2007) 83.

https://doi.org/10.1016/j.nima.2017.01.021

slide-16
SLIDE 16

For high gain Landau fluctuations related time walk dominates the resolution – noise jitter is not so important

Besides high gain also velocity has to be close to saturated <E>=3 V/mm to make signal as short as possible and minimize the effect of Landau fluctuations

Landau fluctuations (not correctable) contribute ~25 ps for 50 mm detector

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

16

TEST BEAM – Sing ngle e Pads ds TEST BEAM – 2x2 array ay (CNM NM – R9088) 8) Mini nimization ation of dead ad area ea @ same me HV performa mance nce is cruci cial al

90 90Sr UCSC setup

up Sing ngle e Pads ds

http://arxiv.org/abs/1707.04961

slide-17
SLIDE 17

Gain degrades with fluence, due to loss of doping in multiplication layer.

“Breakdown” of the device is shifting to higher bias voltages with irradiations

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

17 medium dopin ing g od p+ layer (1.9e13 cm-2)

Reduction of gain after charged hadron is larger (at the same NIEL) than for neutron irradiations.

https://indico.cern.ch/event/587631/contributions/2471705/

Radia iati tion

  • n damage

mage and gain in:

Feq<1015 cm-2 : reduction of the gain due to acceptor removal in multiplication layer (smaller slope in Q-V)

1015 cm-2 < Feq<2∙1015 cm-2 : substantial multiplication visible only at highest voltages

Feq>2∙1015 cm-2 : no difference between LGAD and PIN – bulk multiplication seen in both

slide-18
SLIDE 18

With irradiations the required bias voltage for gain increases (above <E>=3 V/mm the velocity is almost saturated) -> the gain becomes directly related to time resolution (providing the noise level doesn’t change)

At highest fluences trapping and charge multiplication in bulk lead to faster signal, hence better timing resolution at the same gain as for the highest fluences.

20 20 ke ke/50 mm

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

18

TEST BEAM – Sing ngle e Pads ds

90 90Sr UCSC setup

up - Sing ngle e Pads ds

Opti timi miza zation ion of interl terlinked: nked: timing ming reso solut ution

  • n,

, detec tecti tion n effi fici cienc ency, y, nois ise occu ccupanc ncy is very ry challe lengi nging ng for r HGTD TD.

http://arxiv.org/abs/1707.04961

  • J. Lange et al., JINST 12 (2017), P05003
slide-19
SLIDE 19

Ways to increase radiation hardness are investigated:

  • Galli

lium um implantation instead of B – Ga can be more difficult to displace

  • Carbo

rbon n implantation-spray (multiplication layer) or diffusion (bulk and multiplication layer) - suppression of B removal mechanism

Improved HV design for very high voltage operation

  • Junction Termination Extension added to each cell

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

19

CNM Run 10478

5∙1014 cm-2 T=-10oC

shown at 30th RD50 Workshop in Krakow, June 2017

slide-20
SLIDE 20

Atlas LgadTiming Integrated ReadOutChip (ALTIROC)

Designed in TSMC 130 nm

four channels dedicated to «2pF-channel (1 mm x 1 mm sensors)» and four to «8 pF (2 mm x 2 mm sensors) /18 pF-channel(3 mm x 3 mm sensors)»

channel area (200 μm x 100 μm) = Preamp+ TOT and CFD, no TDC

area = 3.4 x 3.4 mm2, thickness=300 μm: large chip to fit four 1 x 1 mm2 sensors(for test beam)

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

2

Omega design

For expected leakage current increase with irradiation, the shot noise contribution should be small enough not to increase the noise and by that the jitter

Currently extensive lab tests:

  • Bump bonding tests (large pixels-less demanding)
  • Calibration signal tests

ENC 650 e 1150 e

slide-21
SLIDE 21

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

21

Full size (2x2 cm2) ) - 225 channe nnels of f 15x15 cells s of 1.3 x 1.3 mm mm2

2

Output put:

  • 24 bits transmitted at 1 MHz (L1 trigger) for timing
  • 8

8 bits s for cell hit position:

  • 7 bits

s for ToA and 9 bits s for TOT 1 discriminator (ToT architecture) OR 2 discriminators (CFD architecture)

  • Hits summed on ASIC level at 40 MHz for Lumi/BM : 7 bits for hits, 5 bits

for out of time window (BC+2ns hits), 4 bits header

Trans nsmi missio ssion n spee eeds/Ba s/Band ndwi width th:

  • At max 10% occupancy <n> = 30 -> 30x24 bits x 1MHz = 720 Mb/s

/s – per r chip « Average FIFO » to average the rates and match the LpGBT inputs

  • Luminosity Link @ 40 MHZ- elink to LpGBT -> 40MHz x 16 bits= 640 Mb/s

/s – per chip

  • Small clock jitter (<20 ps) and clock stability is crucial and demanding

Omega/SLAC design

slide-22
SLIDE 22

Conventional Hybrid-Pixel approach

Easier bump-bonding – large pixels (SnAg,SnPb)

Extremely limited space – full layer only 14 mm

Flex: 20-25 lines (HV-1kV,Power, Slow control, Data, Clocks)

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

22

Modules will be mounted on bots side of the cooling disks with overlap to prevent gaps in acceptance:

  • 13952

13952 modules required (3000-4000 per site in 2y)

  • 6979 modul

dules/ es/en endcap cap, 5.64 m2/e /endcap ndcap, , ~11.2 m2 Si in total al

  • 220

220 mW/cm2 power consumption (200 ASIC + 20 Si)

  • 10 kW per endcap, 20 kW in total (CO2 cooling)
  • T=-30oC

For R<300 mm the modules could be replaced at half of the HL-LHC lifetime (mounted on thin metal sheet and screwed to cooling plate)

2x4 cm2 detector 1.3x1.3 mm2 pads

slide-23
SLIDE 23

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

23

HGTD quadrant view Note different staves The longest stave- hosting 15 modules on each side Cross-section

  • f the HGTD
slide-24
SLIDE 24

 Significant improvement in physics potential can be reached  ATLAS HGTD is a very challenging project:

  • Superb timing resolution in harsh environment
  • Very demanding in terms of constrained space

 LGAD – novel silicon detector technology with gain used

  • Timing resolution of 26 ps/silicon layer reached before irradiation

and after 3∙1015 cm-2 close to 60 ps

  • Dedicated ASIC prototypes already produced

 How will we proceed?

  • C-spay/Ga doping -> under investigation
  • First ASIC with analogue part + sensor bump bonded -> Summer 2017
  • Complete pixel read-out channel ASIC limited to 5 mm2 (for MPW) with 16
  • r 25 channels bump-bonded to sensor (including a module flex)-

>Summer 2018

  • First Module with one (at least) ASIC bump-bonded to 2 cm2 -> end 2019

9/14/2017

  • G. Kramberger, ATLAS-HGTD, 26th Vertex Workshop, Las Caldas

24

IDR review September 2017, if approved, TDR in late 2018.