THE CMS PIXEL DETECTOR Danek Kotlinski Paul Scherrer Institut, - - PDF document

the cms pixel detector
SMART_READER_LITE
LIVE PREVIEW

THE CMS PIXEL DETECTOR Danek Kotlinski Paul Scherrer Institut, - - PDF document

Pixel2000, Genova, 5 th June 2000 THE CMS PIXEL DETECTOR Danek Kotlinski Paul Scherrer Institut, Switzerland OUTLINE : Detector design, mechanics. Data rates, readout architecture. Simulation of the readout efficiency. Sensor


slide-1
SLIDE 1

THE CMS PIXEL DETECTOR

Danek Kotlinski Paul Scherrer Institut, Switzerland

Pixel2000, Genova, 5th June 2000 OUTLINE :

  • Detector design, mechanics.
  • Data rates, readout architecture.
  • Simulation of the readout efficiency.
  • Sensor development.
  • Performance, possible pixel 2nd/3rd level

triggers.

  • Summary.
slide-2
SLIDE 2
  • 3D - tracking points
  • σ

σ σ σ= = = =( ( ( (z) ~ σ σ σ σ= = = =( ( ( (rϕ ϕ ϕ ϕ) ~ 15µ µ µ µm for precise impact parameter in rϕ ϕ ϕ ϕ= = = =& & & &= = = =z

  • LAYERS: r= 4.3cm 7.2cm 11.0cm
  • replace layers after 6 x 1014/cm2
  • all 3 layers compatible

Low Lumi high Lumi

CMS PIXEL SYSTEM

slide-3
SLIDE 3
  • Mech. & Cooling

0.6% X0

CMS PIXEL BARREL

separate for insertion

  • Layer 1 & 2 & 3 mech. compatible
slide-4
SLIDE 4

BARREL PIXEL MODULE

  • Silicon sensor 16 x 64mm2, 250 µ

µ µ µm thick, 150µ µ µ µm square pixels

  • 16 Readout chips (52 x 53 pixel) ---> ~44k pixel ---> 2.4 Watt
  • ~ 1 % XO

butting chips !

slide-5
SLIDE 5 T able 1: P arameters
  • f
CMS Pixel Barrel Conguration Radius F aces full/half Chips Pixels Area [mm] () Mo dules [m 2 ] La y er 1 lo w lumi 41
  • 45
18 128/32 2304 6:35
  • 10
6 0.15 La y er 2 lo w & high lumi 70
  • 74
30 224/32 3840 10:6
  • 10
6 0.25 La y er 3 high lumi 107
  • 112
46 352/32 5888 16:2
  • 10
6 0.38 () Tw
  • half
faces are coun ted as
  • ne
face T able 2: P arameters
  • f
CMS Pixel End Disks z Radius Blades Sensor Chips Pixels Area cm mm Mo dules m 2
  • 32.5
60
  • 150
24 7 1080 3:0
  • 10
6 0.07
  • 46.5
60
  • 150
24 7 1080 3:0
  • 10
6 0.07

BARREL & FORWARD PIXEL

slide-6
SLIDE 6

CMS PIXEL SENSOR

undepleted E=∼=0 depleted E > 0

ionizing particle track p+ - implant ( - 300 V)

n+ - pixel implants

holes electrons

B - Field ( 4 T ) Silicon

(p-type)

  • Charge sharing by Lorentz effect ---> position interpolation
  • n+- pixel on n-silicon (initial)
  • Operate partial depleted up to 6 x 1014/cm2
  • natural rϕ=- pixel size ~ 150 µ

µ µ µm

  • position interpolation by analog readout ---> σ

σ σ σ= = = =( ( ( (rϕ ϕ ϕ ϕ) ~ 10µ µ µ µm

  • need ~ 20000 µm2 pixel area for pixel circuit ( #transistors)
  • defines z- pixel size ~ 150 µ

µ µ µm ---> good σ σ σ σ= = = =( ( ( (z) ~ 17µ µ µ µm

slide-7
SLIDE 7

PIXEL DATA RATES

LHC high luminosity

  • 1. Tracks :

800 ( η≤2 ), ≤100 (pT > 1 GeV).

  • 2. Pixels :

4.6 + 3.3 + 2.5 + 1. + 1. = 13 k pixels/event;

e.g. @ 7 cm

  • ccupancy ≈ 3.3 × 10-4 ;

pixels/module ≈ 15 (1 from noise); pixels/ROC ≈ 1; single pixel rate ≈ 10 kHz.

  • 3. Data volume :

13 k pixels * 3 Bytes ≈ 39 kB/event; 13 k pixels * 100 kHz ≈ 1.3 G pixels/s * 3(10) Bytes ≈ 3.9(13) GB/s

  • 4. Double column
  • ccupancy ≈ 1.5% ;

0.6 MHz column rate; 2 pixels/column.

  • 5. Optical readout links ≈ 3000.
slide-8
SLIDE 8

FES CCU FED FEC Monitor WS Crate Control RU DCS TTS TTCvi

VME 64x Clock, T1, Reset Clock,T1 Clock &T1 Data Data

Local Network

Frontend 9U VME Pixel Readout 4/2/00

slide-9
SLIDE 9

READOUT OF TRIGGERED DATA

Readout Time

Token Bit Manager Chip

  • Zero-suppression ----> data dependent readout time

Radius /cm/ 4 7 11 ROCs/link 8 16 16 <pixels>/link 16±11 15±9 8±6 <readout-time> /µs/ 4 5 4

  • max. time /µs/

14 13 10 <wait-time> /µs/ 1.4 2.4 1.1 (High luminosity, 40MHz link, 100kHz L1T rate)

slide-10
SLIDE 10

10.45 mm 8.00 mm

Time-stamp & Readout Bus

Control & Interface Block

I C - DAC's

2 Power supply & Clock Pads Readout Amplifier Tranmsmission Line Driver

Column Periphery

PIXEL READOUT CHIP

Data flow of pixel hits in Column Drain Architecture

slide-11
SLIDE 11

A B

Time Stamps Data Buffer A=B

Timestamp w/r Contr read out r w Data Buffer write/read Control 8 8 8 3 3 DAC DAC Chip Token OUT Chip Readout Control Signal Out Analog Multiplexer Column Readout Control clear Pixel hit

  • Col. OR

Data A0 A8 Data valid

  • Col. Token

Token return Clock

Search BC Write BC T1

Chip Token IN Trigger #

COLUMN PERIPHERY

  • Dual-ported hit buffer ( Column Drain <--> Chip Readout)
  • Time-stamp verification with T1 trigger
  • Data formatting and readout of T1-verified hits with analog coded

pixel address header

  • Prototype circuit : DM_PSI35 (DMILL) Dec. 98
slide-12
SLIDE 12

READOUT TIME

r=7cm, 100kHz, 40MHz, 16-chips, hi/low-lumi

Number of pixels per data packet and the packet readout time for the 7cm pixel layer at high and low LHC luminosities. Number of ROCs/link = 16, link speed = 40MHz, L1T rate = 100kHz.

slide-13
SLIDE 13

TBM Chip Stack Counter

The TBM stack counter occupancy. The solid line is for 100kHz 1LT rate, the dashed line is for 30kHz. The Token-Bit-Manager (TBM) chip has to queue the incoming 1st level triggers. Trigger stack for the pixel barrel at 4cm and 7cm, LHC high luminosity:

slide-14
SLIDE 14

Barrel at 7cm

0.01 0.02 0.03 0.04 0.05 0.06 4 8 16 ROCs/link Data loass 100/20/hl 100/40/hl 30/20/hl 30/40/hl 100/20/ll 100/40/ll 30/20/ll 30/40/ll

Data loss for the pixel barrel at 7 cm

BLUE : high-lumi, 1LT 100kHz, 40/20MHz link. RED : high-lumi, 1LT 30kHz, 40/20MHz link. GREEN : low-lumi, 1LT 100kHz, 40/20MHz. BLACK : low-lumi, 1LT 30kHz, 40/20MHz.

slide-15
SLIDE 15

READOUT DATA LOSS

Sources of data loss :

  • 1. Column readout losses e.g. pixels overwritten,

column busy, time-stamp and data buffers full.

  • 2. Readout data buffer overflow, a column can

store only one 1LT trigger confirmed time stamp.

  • 3. After readout the column is reset, all time

stamps are lost.

  • 4. Dead time of 1 clock (25ns) to setup the column

readout mechanism after this column was hit. Pixel Barrel Layer at 7cm, full LHC luminosity

R=7cm, tt+25mb

0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 20 40 60 80 100 120 Trigger rate in kHz Data Loss Column rd. 2clock int. 1-Buffer Reset/Block

slide-16
SLIDE 16

SEONSOR DEVELOPMENT

Recent progress, future plans :

  • Test new sensors received from CSEM

and SINTEF, low and high resistivity wafers, standard and oxygen enriched.

  • Variations in:

p-stop rings, n implant size, # of p-stop rings, guard-rings, ….

  • Irradiation (April 2000 at CERN).
  • Beam tests with bump-bonded detectors

(fall 2000), overall optimization program.

  • Summer-Fall 2001 selection of the final

design. Status at the time of the CMS Tracker TDR :

  • Si sensors irradiated up to 6 * 1014 pion/cm2 ,
  • After some annealing time (~1 year) can be depleted

to 200 µm at -300V bias voltage (14 ke signal).

slide-17
SLIDE 17

MIP Signal [electrons] at -300 V

5000 10000 15000 20000 80 160 240 320 400 days after irradiation

Depletion Depth [µm] at -300 V

100 150 200 250 80 160 240 320 400 days after irradiation

SILICON SENSOR PERFORMANCE

after 6 x 1014/cm2 300 MeV/c pions at PSI

  • expect at least 14000 electrons signal up to 6x1014/cm3
  • need 2000 - 3000 e pixel threshold for analog interpolation

Goal:

  • run with 2500 e pixel threshold

σ σ σ noise threshold ----> noise < 500 e

also after irradiation !

slide-18
SLIDE 18

SENSOR TESTS

50 100 150 200

  • 2.0
  • 1.0

0.0 1.0 2.0 Vbias [V] Ipix [µA] pixel @ -0.2V pixel @ GND Bump Pad n -pixel

+

p -stops

+

150µm

Inter-pixel current as a function of bias voltage.

slide-19
SLIDE 19

SOME COMMENTS ABOUT THE PERFORMANCE

How could the pixel detector be used in triggering?

  • 3rd level CMS trigger involves full event

reconstruction and the pixel detector is fully included;

  • Can one do something at the 2nd level

with a "standalone" pixel detector information? Despite the somewhat different designs both detectors Atlas and CMS show very similar simulated performance. → See the LHC physics yellow report, e.g. Bs → J/Ψ(µµ) Φ(KK) ATLAS CMS Events 300000 300000 Time resolution 0.063 fs 0.063 fs Background 15% 10%

slide-20
SLIDE 20

PIXEL HIT RESOLUTION

Pixel Barrel Layer at 7cm Hits from 100 GeV µ µ µ µ tracks.

rϕ ϕ ϕ ϕ direction

5 10 15 0.5 1 1.5 2 2.5 Rapidity Resolution [µ [µ [µ [µm] 200-125 300-150 300-125 150-125 250-150 200-150 250-125

z direction

10 20 30 40 0.5 1 1.5 2 2.5 Rapidity Resolution [ µ µ µ µm] 200-150 200-125 300-150 300-125 150-125 250-150 250-125

slide-21
SLIDE 21

PIXEL HIGH LEVEL TRIGGERS

THE ALGORITHM

  • Correlate hits in two pixel layers which

point in ϕ and z to the interaction region.

  • Match hits in ϕ and z in the 3rd pixel layer.
  • Find the PV (histogramming the z impact

parameter) and eliminate track candidates which do not point to it.

slide-22
SLIDE 22

PIXEL TRACK FINDER

h500 2tau-jets, Pixel Track Efficiency

Efficiency for track finding using the 3 pixel layers only.

slide-23
SLIDE 23

PV Finding using Pixel Hits

qcd 60, high/low lumi, 3/2 layers

The difference (in cm) between the z position of the Monte Carlo primary vertex and the reconstructed primary vertex. Only pixel hits are used in the vertex reconstruction. The upper plot is for qcd-jet events at high luminosity with the 3 layer pixel barrel. The lower plot is for same events at low luminosity and with a 2 layer pixel barrel. The dashed line shows all found vertices and the solid line is the main "signal" primary vertex as found by the algorithm.

slide-24
SLIDE 24

SUMMARY

CMS pixel collaboration :

  • Readout chip : PSI, RWTH Aachen, ETHZ,
  • Sensors : PSI, Uni. Zuerich,

Purdue Uni., Johns Hopkins Uni.,

  • TBM chip : Rutgers Uni.,
  • Bump bonding : PSI, ETHZ, UC Davis,
  • Trigger : PSI, Uni. Basel,
  • Control : PSI, Uni. Basel, Fremilab,
  • Mechanics, cooling : PSI, Uni. Zuerich,

Northwestern Uni., Uni. of Mississippi,

  • VME readout : HEPHY Wien,