VELO Decisions Thomas Ruf LHCb week February 2001 Interference with LHC machine LEMIC, January 2001: Presentation of the VELO Vacuum Chamber design

VELO Decisions Thomas Ruf LHCb week February 2001

Interference with LHC machine Interference with LHC machine LEMIC, January 2001: Presentation of the VELO Vacuum Chamber

design was well received, see also conclusion from A.Rossi. The subsequent discussion in the LHC technical board was also

quite positive (unofficial). BUT, BUT, possible request from the LHC machine:

Downtime of LHC in case of a worst case accident should be limited to 2 weeks.

This means for LHCb: Replace VELO Vacuum Chamber with spare beam pipe RICH1, inner tracker ? continue LHC running without LHCb

Detailed planning needs to be worked out how this could happen.Started by M.Ferro-Luzzi in context of risk analysis.

Need a spare beam pipe. Fast switch to spare beam pipe needs to be taken into account in design for Station1 and Rich1.


ConclusionsConclusionspresented by A.Rossi (LHC-VAC)

The LHC/VAC group accepts:The LHC/VAC group accepts: Foil not withstanding atmospheric pressure - compromise between

safety and physics performances. 2 phases CO2 cooling system in secondary vacuum.

Provided that:Provided that: Risk assessment complies with LHC standards. Design developments in close collaboration/supervision with CERN. Prototype and testing prove principles. Replacement of the vacuum chamber in the case of a major accident.

Yearly design reviews, 2 day workshops: April 2001 February 2002 February 2003


VELO DecisionsVELO Decisions

Sensor choiceSensor choice Position of Off-Detector ElectronicsPosition of Off-Detector Electronics


The VELO Sensor ChoiceThe VELO Sensor Choice

Unanimous decision at the 3rd VELO workshop, 26 January, to adopt n-strip detectors as the baseline for the TDR.

Driving arguments for a technology choice are: Resolution Signal / Noise Radiation hardness Availability

Problem:Problem: Cannot get all the Cannot get all the good things at the same time.good things at the same time.

Find a solution of which we Find a solution of which we are convinced it will work.are convinced it will work.

Improvements are still possible Improvements are still possible at a later time.at a later time.

p-strip OR n-strip detectors ?

TP: 150m thin n-strip detectors with 40m minimal strip pitchsince then: trying to prove feasibility and to improve


What do we have tested?What do we have tested?n-on-n detectors from HAMAMATSUn-on-n detectors from HAMAMATSUThickness: 300 m, Smallest strip pitch: 40 m

p-on-n detectors from MICRON:p-on-n detectors from MICRON:Thickness: 200 m (150 m and 300 m exist but not tested

Smallest strip pitch: 32.5 m for the r-detector and 24.4 m for the phi-detector

p-on-n detectors, DELPHI module, double sided readout:p-on-n detectors, DELPHI module, double sided readout:Thickness: 310 m, Smallest strip pitch: 42 m n-side

25 m p-side (only every second strip read out

+ measurements from ATLAS, CMS, ROSE, …But LHCb is special: Small strip pitch !But LHCb is special: Small strip pitch !


Effect of RadiationEffect of Radiationfrom Rose collaboration based on diodes NOT directly comparable to strip detectors

VELO: ~ 1 x 1014 neq cm-2 / year at r=8mm

Increase of depletion voltage after irradiation.

Lifetime of the VELO is limited by the maximum voltage which the detectors can stand.Because of small strip length and low temperature, the VELO is not limited by the increasing current.Detector will slowly die from inside to outside: The region at r>11mm will last twice as long as the region at r=8mm.

depletion voltage ~ d2

use thinner detectors

BUT signal ~ d

or run detector not depleted

n-type p-type radiation


What is the difference of a n-strip and What is the difference of a n-strip and p-strip detector after irradiation ?p-strip detector after irradiation ?

Charge collection efficiency and resolution of an irradiated double sided silicon microstrip detector operated at cryogenic temperatures.

Nucl.Instrum.Meth.A440:17-37,2000

Measurements done using a double sided detector

If the detector is not fully depleted, charge is spread over a large area on the p-side.

Effect only seen for small strip pitches (<100m)

Confirmed by Y2k testbeam results and laser tests.Conclusion: p-strip detectors need to be fully depleted !

double pitch

acts as insulating layer

n-type p-type radiation


One word about oxygenated siliconOne word about oxygenated silicon

A higher CCE is reached for a lower bias voltage with oxygenated silicon compared to standard silicon.

HOWEVER, the voltage, where maximum CCE is reached, is not so much different between oxygenated and standard silicon.

Oxygen could help in case of n-strip detectors.

Liverpool, G.Casse


What can p-strip detectors offer ?What can p-strip detectors offer ?Or why consider p-strip detectors after all ?

p-strip detectors can have smaller strip pitchesn-strips need to be isolated. Done by using p-stops, p-spray. Limits minimal strip pitch.

p-strip detectors can be made thinnerHamamatsu produces n-strip detectors with 300m thickness only. BUT, MICRON accepted order for 200m thin detectors.

p-strip detectors are cheaperexpected saving ~30% (ATLAS). But money is not an issue. Sensor cost is only 10-15% of total VELO cost.

Physics Study


Optimization StudyOptimization StudyGeneral Constraints

Alignment, handling of detectors: use 1800 detectors instead of 3600 detectors as in the TP reduce outer radius from 6cm to 4.5cm for fitting on 6-inch wafer reduce distance between stations and increase number of stations 25(17)

RF shield, wakefield guide, secondary vacuum: use RF-box with corrugated structure which acts as RF-shield and as

wakefield guide have to use 250m thick Aluminum, because

of maximum pressure and RF penetration


Optimization StudyOptimization StudySensor Geometry

Two solutions were proposed: Conservative detector, 300 m thin

r-detector, strip pitch: inner region: 40m medium region: 60m

outer region: 80m

phi detector: inner region: 40-114m outer region: 40- 72m 50 stereo angle

(strip geometry similar to TP) Ultimate detector, 220 m thin

half the strip pitch to above

NOTE: Too many channels ! For realistic design, change strip pitch linearly as function of radius and use floating strips.

From testbeam (SCTA chip):S/N20 with 300m thin detector expect S/N 10 for 150m expect S/N 15 for 220m

Available as p-on-n and n-on-n

p-on-n: 32.5m and 24.4m existn-on-n: 32.5m and 24.4m ordered

The performance of many different designs was studied.A clear improvement was seen by going closer to the beam, rmin=8 mm (10 mm)


Optimization StudyOptimization StudyPerformance

Multipl. scattering: ~1/perror on impact parameter ~ ldistance to first material: l~ 1/sinerror on impact parameter ~ 1/pt

(moving closer to beam axis helps)

Average decay length errors Conservative detector

B176m BKsJ/236m Ultimate detector

B138m BKsJ/176m

ultimate detector 20% betterrealistic detector 10% better ?(existing: p-strip, 200m, ~30m pitch)

10 5.6 3.2 1.78 1.0 pt (GeV/c)

No error on primary vertex !

TP design: B: 211m BKsJ/ : 280m


How realistic is the simulation ?


ThicknessThickness[220m - 300m]

depletion voltage ~ddepletion voltage ~d22 signal ~dsignal ~d

multiple scattering ~multiple scattering ~ddoptimization study: d=300m: acceptabled=500m: effect seen

lifetime of inner detector region ~ 1/d

RF + exit window shield 11.8% X0

Hybrid/support 1.7% X0

Silicon 4.0 - 5.4% X0

Radiation lengthRadiation length

nn: HAMAMATSU 300m tested new MICRON design aims for 200m 200m tested, (~15% less signal as expected,

to be understood) 150m available

pn:pn:

Availability Availability

Average = 17.4% - 18.9% X0


ThicknessThicknessRF Shield / Sensors

Total Aluminum seenTotal Aluminum seen Total Silicon seenTotal Silicon seen

Average = 1.05 cmincludes 2mm of exit window

Average = 0.51 cm


Radiation lengthRadiation length

Mean without beam pipe: 18.5% X0

75% of the particles see less than 20% X0.

5% see more than 40% X0.


Material seenMaterial seen

Particle trajectory: x=y=z=0, =0.1, 00

X0 Si+hybrid = 142300m = 0.84cm 9%Al(450) = 2250m = 0.04cm 0%Al(900) + exit window = 2.5mm 2.8%

11.8%

Particle trajectory: x=y=z=0, =0.1, 900

X0 Si+hybrid = 1221.5300m = 1.08cm 11%Al(450) = 14222250m = 2.0cm 22%Al( 00) = 250m/0.016 = 1.6cm 18%Al(900) + exit window = 0.25cm 2.8%

54%


RF box, support and cooling framesRF box, support and cooling frames


ConclusionsConclusions The safest solution is to use n-strip detectors.The safest solution is to use n-strip detectors. A possible improved cluster resolution doesn’t A possible improved cluster resolution doesn’t

justify the use of p-strip detectors. justify the use of p-strip detectors. 200200m thin detectors seem not to be mandatory.m thin detectors seem not to be mandatory.

Starting with 300Starting with 300m thick detectors and a m thick detectors and a minimum strip pitch of 40minimum strip pitch of 40m seems acceptable.m seems acceptable.

R&D will continue to find out if n-strip detectors R&D will continue to find out if n-strip detectors with smaller strip pitch and/or smaller thickness with smaller strip pitch and/or smaller thickness can be made to work. can be made to work.


PlanningPlanning

Next steps: n-on-n detectors from MICRON: same design as p-on-n detectors, 200m thick

Status: order placed November’00, Mask finished, first detector expected in AprilAim: evaluate performance for comparison with HAMAMATSU

n-on-n detectors from HAMAMATSU:final design, phi-detector with large stereo angle, floating stripsContact HAMAMATSU now !

Contact other companies July 2001 - June 2002: Evaluation of prototypes July 2002: Review of sensor design, start tendering process December 2002: Order final sensors

Need to have prototype detectors with final design in test beam before submitting the final order.


Position of Off-Detector ElectronicsPosition of Off-Detector ElectronicsAnalog transmission over 60mAnalog transmission over 60m

Response is flat up to 10MHzAt 40MHz the loss is 2.3 [dB]

Lausanne, R.Frei


Measurements with SCTAMeasurements with SCTA

Lausanne, G.Gagliardi

Results Signal/Noise: 10% less but cable is

not the best on the market Cross talk similar to 8m cable

without line equalizer, CF=0.037

Conclusion Twisted pair cables are a cheap

alternative to analog optical links 10m, twisted pair: 48kCHF + connectors 60m, optical: 960kCHF + fibres 60m, twisted pair: 192kCHF + connectors

SCTA Header SCTA Data

Output of line driver

after 60m

1 Mip signals

Pickup noise comes from RB2

LHCb note to come

Documents

VELO Decisions Thomas Ruf LHCb week February 2001 Interference with LHC machine LEMIC, January 2001: Presentation of the VELO Vacuum Chamber design