The LHCb VELO Upgrade Jianchun Wang Syracuse University For the LHCb VELO Group VERTEX 2009 Workshop,Veluwe, Netherlands, Sept 13-18, 2009

The LHCb VELO Upgrade

Jianchun Wang

Syracuse University

For the LHCb VELO Group

VERTEX 2009 Workshop,Veluwe, Netherlands, Sept 13-18, 2009

Sept 13-18, 2009 Jianchun Wang, Vertex 2009 2

The LHCb Environment

Choose to run at <L>~21032 cm–2s–1 (2 fb–1/year), ~1012 bb pairs produced per year.

Maximize probability of single interaction per crossing.

Clean environment (average interactions per crossing <n>~ 0.5 ), easier to reconstruct.

Less radiation damage ( the closest tip ~7 mm away from beam).

Designed to maximize B-acceptance within cost and space constraints.

Forward spectrometer ( 1.9 < || < 4.9 ), where b are maximally boosted, benefits proper time measurement.

One arm is OK since bb directions are correlated. Production bb~500 b, Min. bias: inel ~ 80 mb.

A dedicated b-physics experiment

b

b

b

b

doesn’t occur b

b


VErtex LOcator

The LHCb Detector

VErtex LOcator• primary vertex• impact parameter• displaced vertex

Tracking StationsTrigger Tracker

(IP) ~ 14 m + 35 m / Pt

Primary vertex ~10 m in X/Y, ~50 m in Z.

B proper time resolution ~ 40 fs.

Ready

to g

o !


Limitation of Current Trigger System To improve the physics sensitivity LHCb plan to

increase the luminosity by a factor of 10.

Current trigger system has two stages: L0 hardware trigger, reduce to 1 MHz. HLT software trigger, reduce to 2 kHz.

L0 trigger limit is driven by 1 MHz maximum readout rate.

,

3.5

2.5

1

hadronT

eT

T

E GeV

E GeV

P GeV

With increased L, the L0 criteria has to be tighten to stay below 1 MHz. There is not much net gain, especially in hadron channel.

Increasing L to 1033 will not introduce too much complexity to event. Most events have 1 or 2 interactions. Increasing to 2x1033 the average number of interactions increases to ~ 4,

At L=2x1033, unacceptable increase in CPU time due to large combinatorics.

hadron trigger

muon trigger

Ra

te


LHCb Upgrade Strategy

Run at 2x1033 cm-2s-1 for 5 years (~100 fb-1). This plan is consistent with, but independent of the sLHC upgrade program.

LHCb will readout all sub-detectors at 40 MHz replace FE electronics of most detectors.

The trigger decision will be performed entirely on CPU farm. Removing L0 trigger, thus no limit on the data rate. Software trigger allows sophisticated triggering scheme with lower Pt cuts and

use of IP to maximize the signal yields. Preliminary studies show that at current L the hadronic channel efficiency

improves by ~2 (the goal), and the yield increases proportionally to L. Increases in statistics by ~20 for hadronic channels and ~10 for leptonic

channels. More complicated events significantly increase trigger processing time,

increasing granularity is mandatory.

Improve or maintain efficiency and resolution.


pile-up veto (R-sensors)

RF foil

interaction point

3cm separation

~ 1 m

Current Vertex Locator Silicon micro-strip, n+ in n-bulk sensors. Detector halves retractable (by 30mm) for

injection. 21 tracking stations per side. R-Φ geometry, 40–100μm pitch, 300m thickness. Optimized for

tracking of particles originating from beam-beam interactions.

fast online 2D (R-z) tracking. fast offline 3D tracking in two steps (R-z then ).

r = 8 mm( ~7 mm)

r = 42 mm

2048strips


Current VELO Detector for Upgrade?

Radius (cm)

Par

ticle

Hits

/ E

vent

/ c

m2

L(x1032cm-2s-1) 2 5 10 20

a

b

At L=2x1033 the occupancy will be high, especially at the inner region (particle/hit occupancy ~2%, strip occupancy ~4%). Higher granularity and lower noise are needed. Pixel architecture meets the need.

Analog readout of VELO data at 40MHz is unrealistic. Signals need to be digitized and sparsified at FE.

The data rates are enormous. Clustering reduces the data rates out of FE.

This triggered a dedicated R&D on radiation hard FPGAs.

Pixel architecture reduces pattern recognition difficulty.

Inn

er

Ra

diu

s

Start upgrade studies from the current VELO


Pixel Detector

Two main thrusts of R&D required for the upgrade are: Front-End Electronics Sensor

We had studied different types of sensors and readout schemes. We decide to focus on an upgrade solution based on pixel architecture.

Electronics


Front-end Electronics

Must provide digitized data to trigger processor in real time: digitization, data sparsification, and pushing data to storage buffer.

Withstand same radiation environment as sensor TID: ~400 MRad.

Modern fabrication technologies may allow larger radiation tolerance – 130 nm, 90 nm.

Issues to be investigated Analog electronics optimization: noise, peaking time, stability of

performance with irradiation, leakage current compensation Digitization: speed, effective number of bits. Zero suppression: effective threshold, time walk, discriminator stability Digital section: data rate capability, stability of performance with

irradiation.


Timepix Readout Chip

Pixel readout chip based on Medipix2

256 x 256 pixels 55 m square, and chip is 3 side buttable. Single Layer Modules Possible!! (X0)

By using TSV (through silicon vias) dead side can be reduced to 0.8 mm in Medipix3 (out of 1.5 mm)

Analogue power consumption 6 W per pixel. TOT provides better than 6 bit equivalent ADC resolution

Upgrade being considered to 90 nm technology (power consumption, density of logic and radiation hardness benefits)

Specifications for Timepix adaptation to VELO upgrade needs are under study.

14100

14

10

0

800Medipix3Chip dimensions

See Richard Plackett’s presentation for more details

Q


Conceptual Design

Silicon (1-3 pieces)55x55 m pixels+ 800 m pixels inareas under chip periphery

10 Tiimepix chips(periphery indicated in white)

Ground plane(aluminised directly onto diamond)

Diamond thermal plane withcutouts immediately above TSV regions

Power strips andsignal routing area

Cooling channel

Cross section cutout region

Advantage: Single layer, less material, very important for IP.

Challenges: power tape + TSV, Thinned electronics / sensor.

Beam position


Timepix In Real Beam

A telescope was constructed with 6 double rotated (9o) around both axes. 4 Timepix and 2 Medipix sensors were used.

The DUT (in this case another Timepix chip) position and angle are controlled by a stepper motor to reduce the number of interventions

6 plane Timepix/Medipix telescope

DUT

Track reconstructed


Quick Look at TimePix Testbeam Data

More to Be Studied Time walk, which could potentially affect physics. Non-linear gain curve. Investigate methods of moving data off chip at

required speeds (started, digital processing within pixel array may be necessary).

Testbench features of baseline module.

N=1

N=2

N=3 N=4

Angle (Degree)

(Npi

xel=

N)

/ All

Angle (Degree)

Unb

iase

d R

esid

ual (

m)

Normal Incidence

All

N=1 N=2

N=3 N=4

Indiv.pixel

Total Charge


Medipix3 X-ray Irradiation

Total ionizing dose measurement on a single chip up to 400Mrad (X-ray, continuously over 4 days)

Confirmation of previous single transistor studies on 130nm CMOS

Chip readout DACs, LVDS etc remained operational for full dose.

This needs to be followed up by hadron irradiation.


Current VELO

Pixel X/Y

Pixel X/Y

Pixel Y

~ 1 m

5 stations6 modules per station4 chips per moduleTotal 120 FPIX2 chipsAperture 35 x 35 mm2

Number of Rows

Num

ber

of H

its (

arb

Uni

t)

Residual (mm)

=11.3 m75.8%


Electronics Development

FPIX2

Features (selected): Derived from BTeV Columnar architecture (50m x 400 m) Flash ADC (3-bit) Data driven readout. Sensor and electronics successfully

thinned down to 200 m. Separate X/Y precision measurements.

Large digital periphery requires overlap material.

Need to Study (selected): Modern fabrication technology Radiation hardness study. Stability of front end electronics Readout speed, with designed for 132ns. Cooling.

Focus resources on a single chip development (still useful for sensor R&D).

TIMEPIX

Features (selected): Derived from MediPix Square pixel (55m x 55m) Time over threshold Single layer module possible, edgeless

possible

Need to Study (selected): Modern technology: 130nm or 90nm. Radiation hardness study. Optimization of analog front end Time walk Digitization precision and readout scheme Data flow and data rate TSV technology Detector and sensor thinning Cooling

Development continues.

Two main avenues of investigation in LHCb Upgrade

Sensors


Sensor

Sensor requirements Radiation hardness Low leakage current, heat dissipation High granularity Minimize material

Sensors investigated: Sensors of current type: n-type, strips Sensors with small modifications: p-type Sensors with large modifications: pixels Sensors using new technology: 3D Alternative material: diamond


Test of Irradiated Sensors

N type sensor

Preliminary VELO sensors of n-type and p-type

were differentially irradiated.

They were tested in the beam at Fermilab using FPIX2 pixel system for tracking (in collaboration with Dave Christian).

Irradiation particle flux ~ 0.86x1015 neq/cm2 (~6 years running of current L at inner radius).

More results on irradiated sensors are coming soon.

N-type

~ 25% drop

Measured in 120 GeV proton beam @ -10C

Preliminary

1 MIP Qmp


Double-sided 3D Sensor

Passivation

p+ doped

55m pitch

50m

300m

n+ doped

10m

Oxide

n+ doped

Metal

Poly 3m

Oxide

Metal

50m

TEOS oxide 2m

UBM/bump

n-type Si

Passivation

p+ doped

55m pitch

50m

300m

n+ doped

10m

Oxide

n+ doped

Metal

Poly 3m

Oxide

Metal

50m

TEOS oxide 2m

UBM/bump

n-type Si

Optimisation for SLHC: Nucl. Instr. Meth. A 592 (2008) 16 Glasgow / CNMTest beam results: Nucl. Instr. Meth. A 607 (2009) 89

• novel double sided structure• n-bulk and p-bulk detectors produced & tested

SEM after polysilicon deposition and etching

Pixel on Medipix detector

m


Double-sided 3D Sensor

Glasgow / CNM

P+

N+

1/Capacitance, Pad detector

0.0E+00

5.0E+08

1.0E+09

1.5E+09

2.0E+09

2.5E+09

3.0E+09

3.5E+09

4.0E+09

4.5E+09

5.0E+09

0.0 5.0 10.0 15.0 20.0

Bias (V)

1/C

(F

-1)

2.3V lateral depletion

~9V back surface depletion

Tesbeams at Diamond Light Source (X-rays), CERN (MIPs)

3D Planar

0 2 4 6 8 10 12 14 16 18 200

5

10

15x 10

6 3D - Signal spectrum vs Energy - 15keV

Energy (keV)

Ra

te o

f ch

an

ge

in c

ou

nts

3D, 15keV

Charge sharing

Noise Signal

Measurements of:• Charge loss in holes• Charge sharing• IrradiationsTwo presentations at IEEE NSS with full results


Sensor R&D

Diamond sensor Ultra radiation hard, Very low leakage current (~nA instead of A) Needs no guard ring and can be edgeless. CVD diamond can also be used as heat conductive

spine. Joined RD42 collaboration for investigate this option. Planning test beam studies of different solutions.

2551-6FZ ; 312 micron thick ; n on p

Bulk currents for 3 sensors

0.E+00

1.E-08

2.E-08

3.E-08

4.E-08

5.E-08

6.E-08

7.E-08

8.E-08

9.E-08

0 200 400 600 800 1000 1200

Voltage (V)

Cu

rre

nt

(A)

S1

S2

L P-type silicon sensor “BTeV style” single chip pixel devices.

Fabricated by Micron Semiconductor. Depletion voltage 20-80V before

irradition. Started examining performance of

irradiated detectors

Syracuse/RD50

Syracuse/RD42

Other Issues


POSSIBLE NEW RF FOIL for UPGRADE MAIN REQUIREMENTS

Separates accelerator and detector vacua (must be ultra-high vacuum compatible)

Should have the smallest radiation length possible, esp. before first measured point

Must shield against RF EMI pick-up effects Must carry beam image charge Must allow for sensor geometry and overlap Must withstand high radiation levels

CURRENT DESIGN Foil is 300 um AlMg3, coated with insulator and

getter Foil shape set by overlapping sensors, beam

clearance and beam effects Wakefield suppressors to adapt beam pipe

geometry Was a huge engineering effort (NIKHEF)

UPGRADE DESIGN Replace AlMg3 by Carbon Fiber composite

Use large-modulus fibers (stiff, low density) Resin with high rad tolerance, low outgassing

(space-qualified), micro-crack resistant Produce foil + box + flange as a single

integrated unit Avoids sealing problems

Can reduce mass thickness to ~50% current Similar material mechanically stable to above

500 MRad (CERN 98-01) Currently in development with industrial partner

CMA (Composite Mirror Applications, Inc.) Prototyping planned to start by End 2009,

testing to start in early 2010

CURRENTDESIGN

FOIL~200 mm x 1 m

Dominant contribution to the average X0 of particle traversing VELO at 2<<4.2


Conclusion

The LHCb detector is ready to take data. Initial plan is to collect 10 fb-1 in the first 5 years.

LHCb upgrade forseen for 2015/16, with luminosity increases to 2x1033 cm-2s-1, expected to collect 100 fb -1 in 5 years.

Key aspects of the upgrade are: Readout of the full detector at 40 MHz fully software-based trigger

flexibility. Improved granularity in sub-detectors needed to cope with larger

occupancies, provide better background suppressing and reduce CPU time/event.

Goal is to increase sample sizes by a factor of 10-20 with comparable or better S/B to current detector.

Many R&D projects associated with VELO upgrade are in progress: readout electronics, sensor, hybrid, mechanics, cooling cabling etc.

The VELO upgrade is feasible.

Backup Slides


Backup Mini Strip Plan

200 m Diamond heat planerouting out Beetle signals

20 Beetle40 chips

Inner radius at 7.5mm, 25-30 m pitch(edgeless sensors would bring 10% improvement)

200 m thin silicon sensor

Radiation length 0.6%.

Double sided module

Cooling channel

p-stop or p-spray.

Need to study more on data readout scheme.


Trigger Time in HLT Di-hadron

33 32

33 32

10 2 10

2 10 2 10

2.2

42x

x

t t

t t

L = 2 x 1032 cm-2s-1

L = 2 x 1033 cm-2s-1

L = 1 x 1033 cm-2s-1

At L=2 x 1032 cm-2s-1, the events are more complex. With current VELO-like detector, the pattern

recognition is very slow. And trigger can not be decided within 2.5 m latency.


Irradiation Issue

Operating up to ~120 fb-1

Flux: 0.8x1014 neqcm-2 per fb-1

TID (Electronics): 3.7 MRad per fb-1at tip~ 7mm

500

50

5

Radius (cm)

Dose after 100 fb-1

n eqc

m-2

x 1

016

TID

(M

Rad

)

•After this dose @ 900V we expect• 102 uA / cm-2 at -25o C • CCE of ~ 8.5 ke-

Thermal runaway at the tip is the issue

tip of current VELO

T. AffolderTIPP 09


FPIX2 Readout Chip

FPIX2 Chip

22x400m

128x

50

m

10.3

mm

9.0 mm

I/O & Control pads from Chip to HDI

6.4

mm

0.7 mm

Documents

The LHCb VELO Upgrade Jianchun Wang Syracuse University For the LHCb VELO Group VERTEX 2009 Workshop,Veluwe, Netherlands, Sept 13-18, 2009