J.R. Carter SCT Barrel Module PAR - Overview 1 14th May 2003 ATLAS SCT BARREL MODULE PAR COMPONENTS AND PRODUCTION OVERVIEW Janet Carter, Cambridge TOPICS:

SCT Barrel Module PAR - Overview 114th May 2003 J.R. Carter

ATLAS SCT BARREL MODULE PARCOMPONENTS AND PRODUCTION

OVERVIEW

Janet Carter, Cambridge

TOPICS:1. Barrel Module Requirements2. Module Component Production3. Module Assembly Clusters4. Module Specifications and Categories5. Module QA Steps and Common Issues6. Production Schedule7. Summary


The Barrel Module 2112 Identical Barrel Modules required for SCT

mounted on 4 Barrels (B3, B4, B5, B6)

Bridged wrap-around hybrid – copper-polyimide flex glued on carbon-carbon substrate

4 single-sided p-in-n ac-coupled silicon microstrip sensors, 80 µm pitch,mounted back-to-back, 40 mradstereo rotation angle

12 128-channel ABCD3TAbinary readout ASICs

Thermo-mechanical baseboard -encapsulated thermalised pyroliticgraphite with fused BeO facings


Module Requirements

Detailed in Barrel Module FDR, May 2001: SCT-BM-FDR-4 (http://atlas.web.cern.ch/Atlas/GROUPS/INNER_DETECTOR/SCT/module/SCTbarrelmod.html)

Summary of Principal Module Requirements: MIP Detection efficiency >99% with noise occupancy ≤ 5×10-4

Intrinsic r-φ point resolution per single-side measurement of 23 µm (80 µm pitch sensor strip)

Giving 17 µm precision in r-φ co-ordinate and 500 µm in z from back-to-back sensor pair with stereo rotation angle of 40 mrad

Precise mechanical assembly to simplify digitisation and alignment in ATLAS

Low mass - <1.2% X0 per module, averaged over sensor area 1.17% X0 achieved

Electrical performance maintained up to radiation levels of 2×1014

neqcm-2 (barrel 3 with safety factor) Verified through proton (CERN PS) and neutron (Llublyana)

irradiation and lab and test beam studies (sensors, ASICs, modules)

http://atlas.web.cern.ch/Atlas/GROUPS/INNER_DETECTOR/SCT/module/SCTbarrelmod.html


Summary of Principal Module Requirements (continued):

Cold operation in ATLAS (cooling pipes ~ -20oC) – module must routinely withstand thermal cycling between -25oC and +30oC

Tested by thermal cycling in module QA Module safe against thermal runaway of sensors after

irradiation in ATLAS Thermal designs of baseboard, hybrid and cooling

block – thermal simulations and tests with irradiated modules


Module Component Production

Hybrids – in production in Japanese industry – talk by S. Terada Baseboards – in production by collaboration at CERN – talk by A.

Carter Glues – purchased centrally and distributed to module assembly

sites Silicon Sensors and ASICs – status briefly summarised here:

All SCT barrel sensors manufactured by Hamamatsu Photonics, Japan Identical detectors for all modules: 64mm×63.6mm×285 µm thick Single-sided, ac-coupled, 768 readout strips at 80 µm pitch

Milestones – all achieved, on schedule: Sensor FDR: April 1999 Sensor PRR following evaluation of pre-series: August 2000 Start of series deliveries: January 2001 Delivery (including purchase options) completed: May 2003

QA completed by collaboration (pre-irradiation and sampling after 3×1014 p.cm-2 24 GeV/c proton irradiation at CERN PS)

Sensor quality is excellent

Silicon Microstrip Sensors


Series Deliveries (Barrel + Endcap)Purchase options

added Contra

ct

Ordered (includin

g purchase options)

Delivered

Japan 6000 5913Norway 1950 1950

UK 2750 2750Total 10,700 10,613

Barrel 99% Complete on 1st May 2003

Pre-Irradiation:All quantities within specificationVery low leakage currents Typically <200 nA at 500V bias at 20oC>99.9% good readout strips

Initialcurrent at350V bias (20oC)

Number ofStrip defects


Post-Irradiation: Within specification for operation at ~400V bias after 10 years of

LHC

Points to note for module construction concerning high bias voltages:

The edges of the sensors are at the back-plane voltage Great care must be taken to avoid conducting debris shorting

to grounded areas – eg bond wires, openings in sensor passivation in guard rings, bond pads etc

Module production and assembly to barrels must be a clean process

On Barrels 3 and 4, modules must be tested to 500V bias during assembly and commissioning at CERN to ensure there are no HV shorts that would prevent final post-irradiation HV operation

So I-V of modules are tested up to 500V in assembly QA (even though initial modules will operate at <200V bias)


Pre-irradiation sensor ‘microdischarge’ A small fraction ( ≤ 2%) of sensors show ‘microdischarge’

(impact ionisation) pre-irradiation between 350V and 500V bias Rapid current rise with bias voltage, but falls to normal levels

with short time-constant (~30 mins) if bias maintained Not a problem for module operation in ATLAS:

Disappears after irradiation and type-inversion - field configuration changed:

But will not use ‘mirodischarge’ modules for the inner barrels 3 and 4 as they complicate initial HV tests up to 500V during macro-assembly and commissioning

Example of smooth, normal I-V curves of ‘microdischarge’ detectors up to

500 V bias after irradiation to 3×1014 p cm-2


Front-End ASICs

ABCD3TA ASIC DMILL technology, fabricated by

Atmel Single-strip threshold binary readout Threshold trimming for each

channel

Performance meets requirements both pre- and post-irradiation, with ~1fC

binary threshold Series production released in July 2001

following PRR ~87% of SCT requirement for perfect

ASICs now delivered and die identified after wafer testing

But Atmel may not deliver any more before the DMILL line is stopped

The SCT will use an ASIC with 1 bad channel on each module as necessary to make up the shortfall

Should have a negligible effect on barrel performance

Deliveriesstopped

ASIC Statistics

Lots delivered 50Accepted wafers 795Accepted wafer yield 25.9%Tested wafers rejected 24.0%Accepted perfect chips

52,695

% of total requirement 87.5%Wafers left to test 0


Module Assembly Clusters

4 Barrel Clusters for module assembly: Japan, Scandinavia, UK, US – reports from each on progress

to-date Clusters have developed different jigging and assembly

techniques Each builds to the same module specification Site qualification process following FDR/PRR in May 2001

with documented requirements and module exchange before series module assembly can get fully underway

Cluster Site Qualification

Date

Number of modules to deliver

to macro-assembly sites (Oxford, KEK)

Japan December 2001

800

Scandinavia Not yet 400UK June 2002 550US April 2003 480


Summary of Principal Module Build Specifications: Detailed in Barrel Module FDR, May 2001: SCT-BM-FDR-

7 (http://atlas.web.cern.ch/Atlas/GROUPS/INNER_DETECTOR/SCT/module/SCTbarrelmod.html)

Electrical Performance <1% bad readout strips at 1fC binary threshold (noisy, dead,

part-bonded, untrimmed, pipeline errors etc). Verified through a Characterisation Sequence test using

custom SCT VME readout, prototype SCT voltage supplies and standardised DAQ and analysis code (SCTDAQ)

Module sensor leakage current < (sum of individual sensors + 4 µA) up to 500V bias at 20oC

Stable performance during 24 hour cold operation (hybrid at ~0oC)

All Modules classed for use in ATLAS satisfy the full electrical specification

Module Specifications and Categories




Mechanical Specifications Principal in-plane parameters (further parameters define the hybrid position and angle) (y is perpendicular to the sensor strips to within half the

stereo angle)

In-Plane ParameterDesign Value

‘Good’Module

Tolerance

‘Pass’Module

Tolerance

Baseboard dowel mounting hole in x, mhx [µm] -6500 ±30 ±40Baseboard dowel mounting hole in y, mhy [µm] -37500 ±30 ±40Baseboard dowel mounting slot in x, msx [µm] 38500 ±100 ±140Baseboard dowel mounting slot in y, msy [µm] -37500 ±30 ±40Mid-point of pair of front sensors in x, midxf [µm] 0 ±10 ±10Mid-point of pair of back sensors in y, midyf [µm] 0 ±5 ±8Separation of centres of sensors; front pair, sepf; back pair sepb [µm]

64090 ±10 ±20

Rotation angles of the 4 sensors, a1, a3, a3, a4 [mrad] 0 ±0.13 ±0.13Half stereo angle between the front and back sensor pairs, stereo [mrad]

-20 ±0.13 ±0.13

The category ‘Good’ satisfies the agreed specification The category ‘Pass’ is an extension to cover measurement error and the tails of the observed distributions Both ‘Good’ and ‘Pass’ modules will be used in ATLAS


Out-of-Plane ParameterDesign Value

‘Good’Module

Tolerance

‘Pass’Module

Tolerance

Maximum deviation of lower sensor from module plane, maxZlower [mm]

0 -0.2 -0.2

Maximum deviation of upper sensor from module plane, maxZupper [mm]

0 0.2 0.2

Module thickness [mm] 1.15 ±0.1 ±0.1Maximum deviation of lower sensor from common module profile, optimalmaxZerrorlower [mm]

0 0.05 0.07

Maximum deviation of upper sensor from common module profile, optimalmaxZerrorupper [mm]

0 0.05 0.07

RMS deviation of lower sensor from common module profile, optimalRMSZerrorlower [mm]

0 0.025 0.025

RMS deviation of upper sensor from common module profile, optimalRMSZerrorupper [mm]

0 0.025 0.025

Lower BeO cooling facing angle perpendicular to mounting line, b [mrad] (module cooling contact issue)

0 ±3 ±5

Lower BeO cooling facing angle parallel to mounting line, a [mrad]

0 ±0.5 ±0.5

Lower BeO cooling facing concavity along the mounting line, loCoolingFacingConcavity [mm]

0 ±0.03 ±0.03

Maximum thickness of module at surface of large capacitors on hybrid,capMaxThickness [mm] (clearance issue)

5.78 6.44 6.44

Principal out-of-plane Parameters Detector surfaces compared with a standard shape ‘Good’ and ‘Pass’ categories again defined


Barrel Module Categories Each produced module is assigned to one of the following

categories:

From component availability and schedule requirements, ~90% of modules started in assembly need to be suitable for use in ATLAS

Category DescriptionGood Satisfies all electrical and ‘good’ mechanical

specifications.Pass In ‘pass’ grade for 1 or more mechanical

parameters.Satisfies all electrical specifications.

Hold Outside ‘pass’ grade for 1 or more mechanical parametersand/or does not satisfy all electrical specifications.Such a module is stopped in production and stored.

Fail Could never be used in ATLAS (damage, gross errors).

Rework Rework is needed before the module could be usable.


Module QA Steps and Common Issues

Basic steps in module assembly (with small variations between Clusters):

Step QA after step1. Load 12 ASICs on hybrid

Electrical tests warm and cold

2. Glue 4 sensors to baseboard

In-plane metrology

3. Wrap hybrid around and glue to baseboard-sensor sandwich4. Wire-bond module Metrology (in-plane and out-of-

plane) following thermal cyclingElectrical characterisation warmLong-term (24 hr) electrical test coldVisual inspection


A Common Electrical Issue: s-curves

Noise-occupancy s-curves are part of module electrical characterisation

These are occupancy vs threshold for no injected charge Normally they are smooth curves:

picture here


But coherent effects can build up between ASICs when digital activity is high and produce s-curve distortions at thresholds << 1fC

A severe example:

picture here


Barrel Clusters have recently been studying all their available data because a high fraction (of the small number) of series modules completed by Scandinavia have severely distorted s-curves

Has caused a delay in Scand site qualification, in case of any association with the hybrid mounting technique used by Scand – they are now proceeding cautiously with a new method

But s-curve distortions are seen in varying degrees by all Clusters

An on-going study to look for correlations and differences between Clusters

Does it matter? We think not, provided the distortions are well below (<0.3 fC) the

operating threshold of 1fC – which they are Noise Occupancy of module at 1fC is not affected Effect reduces with irradiation Has not to-date caused problems in system test operation

The properties after irradiation and system operation will be further tested this summer using modules with the most severely distorted s-curves


Production Schedule

SCT Schedule requirements for macro-assembly imply:

******I have made up these dates, which are later than the schedule*******

There are further constraints on selecting the modules with the most robust high bias voltage characteristics for barrels 3 and 4

Nevertheless, sufficient modules should be available in time for the required completion of the first barrel - B3

Barrel Number of Modules to

deliver

From Clusters

Delivery to be completed

byB3 404 All December

2003B6 706 All March 2004

B5 605 All May 2004B4 504 Japan June 2004


Summary

Sensors are in-hand and of very good quality Problems with the end of ASIC delivery, but 1-bad

channel chips can be used to make up the difference

Module assembly underway – now x% of total ‘Good’ + ‘ Pass’ requirement completed

Site qualification of the Scandinavian cluster is an urgent priority

Integrated assembly yield is still lower than the 90% target for all but the Japanese cluster (to be seen from the Cluster presentations)

Overall quality of modules is good – s-curves need careful monitoring and some more study

Documents

J.R. Carter SCT Barrel Module PAR - Overview 1 14th May 2003 ATLAS SCT BARREL MODULE PAR COMPONENTS AND PRODUCTION OVERVIEW Janet Carter, Cambridge TOPICS: