the atlas ibl project – 5 th trento workshop g. darbo – infn / genova manchester, 24-26 february...

The ATLAS IBL Project – 5th Trento WorkshopG. Darbo – INFN / Genova Manchester, 24-26 February 2010o

The ATLAS IBL Project

The 5th "Trento" Workshop on Advanced Silicon Radiation Detectors

Manchester, 24-26 February 2010G. Darbo - INFN / Genova

Conference Site: • http://www.hep.man.ac.uk/Radiation-Detectors2010/agenda.html

http://www.hep.man.ac.uk/Radiation-Detectors2010/agenda.html

The ATLAS IBL Project – 5th Trento WorkshopG. Darbo – INFN / Genova Manchester, 24-26 February 20102

ATLAS Pixel DetectorATLAS Pixel Detector• 3 Barrel + 3 Forward/Backward disks• 112 staves and 48 sectors• 1744 modules• 80 million channels


The ATLAS Pixel ModuleATLAS Pixel Module• 16-frontend chips (FE-I3)

modules with a module controller chip (MCC)

• 47232 pixels (46080 R/O channels), 50 x 400 µm2 (50 x 600 µm2 for edge pixel columns between neighbour FE-I3 chips)

• Planar n-on-n DOFZ silicon sensors, 250 µm thick

• Designed for 1 x 1015 1MeV fluence and 50 Mrad

• Opto link R/O: 40÷80 Mb/link


Pixel Integration and Installation


Mission impossible… fit an additional layer in between Pixel and beam-pipe:

• Reduce beam-pipe by 4 mm in radius… and make it possible!

IBL: Project HistoryThe ATLAS Pixel B-Layer initially designed for replacement• September 2007 B-Layer replacement Workshop outcome: replacement not possible

in 1 year shutdown.• January 2008: ATLAS Task Force (A. Clark & G. Mornacchi) Report July 2008 (Bern):

preferred (only) option Insertable B-Layer (IBL)• February 2009: project approved by ATLAS May 2009 IBL management organization

in place.• Now: design fast advancing, IBL Technical Design Report (TDR) draft, interim

Memorandum of Understanding (i-MoU) in discussion.

Existing B-Layer

IBL

Iourii Gusakov


Motivation for IBL

The existing B-Layer cannot be replaced in a long LHC shutdown (8-months):• This was a major finding of the B-Layer task force. Many reasons make it very difficult:

• Extraction, moving to surface and opening the whole Pixel Detector package.• Work on an activated material.• Risk of damage (many last moment operations made the process “irreversible” in

the final phase of the detector integration).

Reasons for an IBL to back-up existing B-Layer:Radiation damage• Sensor and electronics degradation of the existing B-Layer reduce detector efficiency

after 300÷400 fb-1 (last forecast of LHC integrated luminosity move it more far away)

Insurance for hard failures in the Pixel B-Layer• The Pixel Detector cannot be repaired in case of cooling, opto-links, module hard

failure. Inefficiencies of the B-layer have high impact on many Physics channels.

Improve existing B-Layer Physics performance• A low mass detector (~50% of existing B-Layer) improves Physics performance.


IBL LayoutBeam-pipe reduction:• Inner R: 29 25 mm

Very tight clearance:• “Hermetic” to straight

tracks in Φ (1.8º overlap)• No overlap in Z: minimize

gap between sensor active area.

Layout parameters:• IBL envelope: 9 mm in R• 14 staves.• <R> = 33 mm.• Z = 60 cm (active length).• η = 2.5 coverage.


BP Extraction & IBL+BP Insertion

Material from Raphael/NealThe present 7m long section of the beam-pipe will be cut (flange too big to pass inside the existing pixel) and extracted in situ:The new beam-pipe with the IBL will be inserted at its place:

• A carbon tube (IST) is inserted before the IBL: to support the new detector and to simplify the insertion procedure.

ISTIBL Support Tube

StaveStave

InsertTo fix to support, survey reference

Alignment wirers

Sealing service ring

PP1 Collar

Ref.: Y.Gusakov, N.Hartman, R.Vuillermet


B-Layer ScenariosPhysics performance studies ongoing for the IBL TDR (ATHENA/GEANT4).Preliminary studies (ATLSIM/GEANT3) show improved performance with the addition of IBL (see low mass Higgs b-jet tagging plot on the right).Performance improvement due to low mass and smaller radius:

• Aggressive reduction of material budget is a must!

ATLAS

b-inserted as 4-layer R=3.5 cm

b-replaced

SV1 εb=70%

2-layers R=3.5

cm and 8 cm

2-old layers

SV1 εb=60%

Component % X0 (*)

beam-pipe 0.6

New BL @ R=3.5 cm 1.5

Old BL @ R=5 cm 2.7

L1 @ R=8 cm 2.7

L2 + Serv. @ R=12 cm 3.5

Total 11.0

WH(120 Gev)

Light jets rejection

Ref.: A. Rozanov(*) Material budget used in the simulation


• Fit made for 2 < r < 20 cm for L=550fb-1

• Gives for IBL @ 3.2 cm (550 fb-1):

Φ1MeV=3.3x1015 neq/cm2 (1.6 MGy)

• Safety factors not included in the computation (σpp event generator: 30%, damage factor for 1 MeV fluences: 50%)

Requirements for Sensors/Electronics

Requirements for IBL• IBL design peak luminosity = 3x1034 cm-2s-1 FE-I4 architecture & R/O bandwidth:

must be understood after Chamonix• Integrated luminosity seen by IBL = 550 fb-1 Survive to sLHC phase II• Design sensor/electronics for total dose:

• NIEL dose = 3.3 x 1015 ± (“safety factors”) ≥ 5 x 1015 neq/cm2

• Ionizing dose ≥ 250 Mrad

€

Φ(r) =2.9

r2 +0.14

r

⎛

⎝ ⎜

⎞

⎠ ⎟×1016

Ref.: Ian Dawson


FE-I4 Architecture: Obvious Solution to Bottleneck

>99% or hits will not leave the chip (not triggered)• So don’t move them around

inside the chip! (this will also save digital power!)

This requires local storage and processing in the pixel array• Possible with smaller feature

size technology (130 nm)

Large chip - design methodology:• Custom digital layout

substituted by automatic place & route of synthesized design.

• Chip verification is the challenge: analog/digital and mix-mode test bench simulation.

Ref.

: M

. B

arb

ero

et

al.

FE-I3 @ R = 5 cm


FE-I3 FE-I4FE-I4 Collaboration:Bonn: D. Arutinov, M. Barbero, T. Hemperek, A. Kruth, M. Karagounis.CPPM: D. Fougeron, M. Menouni. Genova: R. Beccherle, G. Darbo.LBNL: S. Dube, D. Elledge, M. Garcia- Sciveres, D. Gnani, A. Mekkaoui.Nikhef: V. Gromov, R. Kluit, J.D. Schipper

The first version of full FE-I4 chip will be submitted by end of March 2010

~70 million transistors, 0.13 µm CMOS technology6 Cu and 2 Al routing layers.

7.6mm

8mm active

2.8mm

FE-I3 74%

20.2mm

active16.8mm

~2mm

~200μm

FE-I4 ~89%

Ch

art

ere

d r

eti

cu

le (

24 x

32)

IBM

reti

cu

le

~19 mm

FE-I3 FE-I4

Pixel size [µm2] 50x400 50x250

Pixel array 18x160 80x336

Chip size [mm2] 7.6x10.8 20.2x19.0

Active fraction 74% 89%

Analog current [µA/pix] 26 10

Digital current [µA/pix] 17 10

Analog Voltage [V] 1.6 1.5

Digital Voltage [V] 2.0 1.2

Pseudo-LVDS out [Mb/s] 40 160


FE-I4-P1

LDORegulator

ChargePump

CurrentReference

DACs

Co

ntr

ol

Blo

ck

Ca

pa

cit

an

ce

Me

as

ure

me

nt

3mm

4mm

61x14 array

SEU test IC

4-LVDS Rx/Tx

ShuLDO+trist LVDS/LDO/10b-DAC

CLKGEN proto: PLL core + PRBS + 8b10b coder + LVDS driv

low power discriminator

The Way to FE-I4: Test Chips


FE-I4

periphery

digital 4-pixel region

analog 1-pixel

4-pixel region

pixel array336×80 pixels


FE-I4: Sensor Related SpecsSpecifications Value Unit Conditions/comments

Pixel size 50 x 250 µm2

Bump pad opening 12 µm diameter

Input -Q DC coupled

Maximum charge 100,000 e

DC leakage current tolerance 100 nA

Pixel array size 80 x 320 Col x Row

Last bump to physical edge ≤ 100 µm

Normal pixel input capacitance range

100÷500 fF

Edge pixel input capacitance 150÷700 fF Sides for long pixels and top for ganged

Radiation tolerance 250 Mrad Specs met at this dose

In-time discriminator threshold with 20 ns gate and 400 fF load

≤5000 e Region can still assign small hits below in-time threshold to correct time bin


FE-I4: Discriminator & R/O Specs

Specifications Value Unit Conditions/comments

Hit-trigger association resolution 25 ns

Same pixel two-hit discrimination 400 ns At 5000 e in-time threshold and when both hits are 20 ke

Single channel ENC sigma <300 e 400 fF load, nominal current

Tuned threshold dispersion <100 e sigma

Charge resolution 4 bits

ADC method ToT

Average hit rate with 1% data loss

400 MHz/cm2 3.2 µs trigger latency, 100 kHz trigger rate

Max number consecutive triggers 16

Trigger latency (max) 6.5 µs

Maximum sustained trigger 200 kHz

Serial command/clock input 40 Mb/s - MHz

1 + 1 input per chip

Serial data output 160 Mb/s 1 output per chip

Output data encoding 8b/10b

I/O signals ~LVDS Current balanced differential


FE-I4: The Pixel Cell

2-stage architecture optimized for low power, low noise, fast rise time.• regular cascode preamp.

NMOS input.• folded cascode 2nd stage PMOS

input.

• Additional gain, Cc/Cf2 ~6.• 2nd stage decoupled from

leakage related DC voltage shift.

• Cf1 ~17 fF (~4 MIPs dynamic range).150 µm

13-bit memory/pixel: 4 FDAC, 5 TDAC, 2 cap, 1 HitEN, 1 HitOR

Ref.: A. Mekkaoui


Noise and Radiation Resultsa) ENC on “Collaboration Proto 1”

before and after irradiation (200 Mrad)

b) Measured ENC for pixels with and without Cload

c) Simulated ENC and time-walk @ 10 µA/pixel (preamp + amp2 + comparator)

ENC @ Low Current (10µA)

(loaded ~400 fF)

<ENC> ~ 65 e<ENC> ~ 90 e

(10 µA)

200Mrad, Cload~400fF

a)

b)

c)

20 ns timewalk for 2 ke- < Qin < 52 ke- & threshold @ 1.5 ke-

ENC=160e- @ Cd=0.4pF & IL=100nAENC[e-]

Cd[F]

Qin[C]

tLE[s]

100f 200f 300f

60

100

150

10k 20k 30k 40k

IL = 0 nA

20n

10n

0

IL=100 nA


Module Design: Sensor Technology Independent

Decision on sensors after TDR• Need module prototypes with FE-I4 (second half 2010)

Common sensor baseline for engineering and system purposes• 3D / Diamond sensors – single chip modules / Planar sensors – 2

chip modules

Sensor/module prototypes for ~10% of the detector in 2010• Stave prototype tested with modules and cooling

Credits: M.Garcia-Sciveres – F. Hügging

Single chip module:Edge < 325 µm

Double chip module:Edge < 450 µm


Sensors3 sensor technologies considered for IBL• Planar, 3D and Diamonds• Full scale prototypes with FE-I4 – Decision on spring 2010

Some specifications agreed:• Max fluence > 5 x 1015 1MeV neutrons / cm2

• Max power after full life dose < 200 mW/cm2

• Low dead area in Z: slim or active edge• Maximum bias voltage (system issue) : 1000 V

Sensor R&D and prototype work for IBL are presented in many talks in the Workshop…


Bump BondingLarge volume bump-bonding experience from Pixel Detector (see table):• PbSn and Indium bumps: PbSn AgSn

Program to qualify for the larger FE-I4 and different sensor technologies.• Setting up with mechanical/electrical

dummies, but finally real parts needed: thermo-mechanical process strongly dependent on actual metal layers of electronic chip and sensor.

• Goal to go below 190 µm of the Pixel Detector: target to 90 µm.

“dummy – sensor”(monitor wafer)

ATLAS Pixel bump-bonding production – Ref: Jinst 3 P07007 (2008)

Prototype test of advanced AgSn bumping with 90µm FE-I4 size dummies.


Thermal Figure of Merit and Thermal Run-away

Thermal runaway happens in sensors if not adequately cooled• Leakage current shows exponential

behavior.

Stave thermal figure of merit (Γ = [ΔT•cm2/W]) main parameter for thermal performance.

Power design requirements for IBL:Sensor Power 200 mW/cm2 @ -15 CFE power 400 mW/cm2

Stave prototype qualification program:Titanium / carbon fiber pipes (D = 2÷3 mm)

Cooling CO2 and C3F8

Carbon foam density: 025÷0.5 g/cm3

Radiation length: 0.36÷0.66 %X/X0

Pipe + stave structure + coolant

-45.00

-40.00

-35.00

-30.00

-25.00

-20.00

-15.00

-10.00

-5.00

0.0 500.0 1000.0 1500.0 2000.0 2500.0 3000.0

Sens

or Te

mp

[C]

Integrated Luminosity [fb-1]

Thermal run-away Plot-40CF MonopipeWorst Thermal Figure of MeritTi Monopipe

C Evaporation TemperatureEvaporation T = -40 ºC = 30.0 C•cm2/W = 18.5 C•cm2/W

= 3.2 C•cm2/W

Thermal Runaway Plot

Ref.: D Giugni, H. Pernegger, M. Gilchriese

IBL including safety


Stave StructureStave structure made of carbon foam + cooling pipe (carbon fiber or titanium boiling channel)

The stiffness is provided by a carbon fiber laminate:

Fiber YS-80A; resin EX-1515; lay-up (0/60/-60)S2

Carbon foam diffuses the heat from the module to the cooling pipePoco Foam r=0.55g/cm3; K=135/45 W/mK OR Kopers KFOAM L1-250 r=0.245g/cm3; K=30 W/mK

0

10

20

30

40

50

60

70

80

90

100

0 0.1 0.2 0.3 0.4 0.5

K(W

/m-K

)

Density(g/cc)

K = 1070(( - 0.045)/2.2)1/0.67

Module (sensor + bumps + FE-I4)

Carbon foam

Omega CF laminate

Ti or CF pipe


Stave Prototype OptionsSTAVE CARACTERISTICS SIMULATION RESULTS

Pipe ID/OD[mm]

Omega Thickness

[µm]

Foam Density[g/cm3]

Coolant X/X0 [%] Thermal Figure of Merit (Γ)

[ºC•cm2/W]Bare

Stave with Coolant

Full layer (+ Module

+ Flex)

CF pipe, heavy foam 2.4 / 3.0 150 0.55 C3F8 0.48 1.056 17.25

CF pipe, light foam 2.4 / 3.0 150 0.25 CO2 0.36 0.956 18.56

Ti 3mm pipe, light foam 2.8 / 3.0 300 0.25 C3F8 0.66 1.276 2.79

Ti 2mm pipe, light foam 2.0 / 2.2 300 0.25 CO2 0.57 1.166 3.22

Module parameters• Sensor thickness = 250 µm• FE-I4 thickness = 90 µm• Flex Hybrid (η = 0) = 0.18 % of X0

Additional technical requirements (prototype work)• Max pressure of cooling pipe: 100 bar.• Develop pipe joints and fittings.• Gravitational / thermal deformation < 150 µm.• Isolation of the carbon foam from sensor high

voltage.• Mock-up for thermal measurements.

Carbon Foam 0.25g/cm3


When IBL in ATLAS?IBL plans to be ready for installation by end of 2014.

• Cannot be much before without compromising performance• A shut down of the machine of 8 month needed (4 to open/close ATLAS)

LHC plans after Chamonix are not clear: how peak and integrated luminosity increase and when machine shutdown will be scheduled:• Only plans up to 2012 are known.• Many LHC upgrades need shutdowns:

• Linac4, Collimators phase II, new interaction region quadrupole triplets, etc.

• Probably in one year from now we will know next 5 years plans.

Chamonix Agenda: • http://indico.cern.ch/conferenceDisplay.py?confId=67839

Summary of the Chamonix Workshop at Cern:• http://indico.cern.ch/conferenceDisplay.py?confId=83135

IBL

http://indico.cern.ch/conferenceDisplay.py?confId=67839

http://indico.cern.ch/conferenceDisplay.py?confId=83135


ConclusionsIBL will improve physics performance of ATLAS and it is a “safety insurance” for present B-Layer

TDR and MoU in progress – project cost evaluated • Motivated groups and institutes support

Challenging project:• Tight envelopes, material budget reduction, radiation dose and R/O

bandwidth requirements

New technologies in advanced prototype phase:• FE-I4, light supports, cooling, but mainly…

SENSORS !!!


BACKUP SLIDES

The ATLAS IBL Project – 5th Trento WorkshopG. Darbo – INFN / Genova Manchester, 24-26 February 201028R. Vuillermet

• The support carbon tube is fixed in 2 point of PP0 and on PP1 walls on side C and A.

• The structural pipe with a support system is moved out from the support carbon tube.

. The new beam pipe (in any configuration with OD up to 82,5 mm) is inserted from A-side. It has 2 supports at PP0 area and 2 floating wall at PP1 on side A and C.

Two global support / installation scenarios: IBL support tube (1) / no tube (2):An IBL support tube would have advantage on stiffness and simplicity/safety for IBL installation, but drawback are envelope needs (~1÷1.5 mm) and increase of radiation length

Procedure studied on mock-up at bld.180 - procedure (1) animation:• The beam pipe flange on A-side is to close to the B-layer envelope - Need to be cut on the

aluminum section• A structural pipe is inserted inside the Beam Pipe and supported at both sides. • The support collar at PP0 A-side is disassembled and extracted with wires at PP1.• Beam pipe is extracted from the C-side and it pulls the wire at PP1• New cable supports are inserted inside PST at PP0.• A support carbon tube is pushed inside the PST along the structural pipe.

Started to setup a 1:1 mock-up of Pixel/beampipe/PP1 in Bat 180

A-side C-side

Installation Scenarios


IBL Organisation Structure

Module WG(2 coordinators)• FE-I4• Sensors• Bump-Bonding• Modules• Test & QC• Irradiation

Stave WG(1 Phys + 1

Eng.)• Staves• Cooling Design

& Stave Thermal Management

• HDI• Internal

Services• Loaded Stave• Test & QC

IBL Integr.-Install.(2 Eng.)

• Stave Integration

• Global Sup.• Beam Pipe (BP)• Ext.services

inst.• IBL+BP

Installation• Cooling Plant• Test & QC

Off-detector(1 Phys + 1 E.Eng.)• Power• DCS• ROD• Opto-link• Ext.serv.design/

proc.• Test Beam• System Test

IBL Management BoardMembership:• IBL PL + IBL TC• 2 coordinators from each WG• Plus “extra” members

MembershipIBL Project Leader: G. DarboIBL Technical Coordinator: H. Pernegger“Module” WG (2 Physicists): F. Hügging & M. Garcia-Sciveres“Stave” WG (1 Phy. + 1 M.E.): O. Rohne + D. Giugni“IBL Assembly & Installation” WG (2 M.E. initially, a Phy. Later): N. Hartman + R. Vuillermet“Off-detector” WG (1 Phy. + 1 E.E.): T. Flick + S. Débieux“Extra” members:Ex officio: Upgrade Coordinator (N. Hessey), PO Chair (M. Nessi), Pixel PL (B. Di Girolamo), ID PL (P. Wells), Pixel Chair (C. Gößling)Offline “liaison” Pixel Off-line coordinator: A. AndreazzaTDR editor (temporary): K. Einsweiler

Whole project divided into 4 working groups• IBL Management Board has 10 members, plus

“extra” and ex-officio members.• Frequent meetings (every ~14 days) in this phase

of the project.

the atlas ibl project – 5 th trento workshop g. darbo – infn / genova manchester, 24-26 february...

Documents

atlas ibl projectthe

othe atlas ibl project

darbo infn genova manchester

backup existing blayer

iblthe existing blayer

trento workshopg

blayer task force

pixel integration