Speaker
Description
The replacement of the ATLAS Inner Detector with the new ATLAS Inner Tracker for the LHC High-Luminosity Upgrade will require the assembly of 17,000 silicon strip modules. In preparation for production, scheduled to take place in 30 institutes worldwide over 3.5 years, a pre-production programme was conducted, assembling 5% of the full production quantity. During pre-production, the discovery of several design flaws, such as high noise, premature death after active proton irradiation and manufacturing problems required design changes at a late stage of the project. This presentation provides overview of the discovered problems and lessons learned for future projects.
Summary (500 words)
The construction of the ATLAS Inner Tracker to replace the current Inner Detector as part of the ATLAS Phase-II Upgrade will require the assembly of 11,000 silicon strip sensor modules for the central tracker region (“barrel”) and 7,000 modules for the forward region (“end-cap”). After an extensive prototyping programme involving both barrel module geometries (“long strip” and “short strip” modules) and one end-cap module geometry (the innermost “R0” module), a pre-production phase was launched, with the aim to assemble 5% of the full module production quantities. Despite the previous prototyping phase, multiple technical issues were discovered in pre-production, requiring additional tests, material replacement, adjusted procedures and time-consuming component redesigns. This contribution aims to provide a comprehensive overview of the various encountered issues, how they were solved, why they were initially overlooked and how they might be avoidable in future projects of comparable scope.
Depending on the allocated time, this presentation aims to include:
• premature death of modules after proton irradiation (identified to be an SEU problem due to the overly harsh conditions at the irradiation facility rather than a design flaw)
• problems with the flex manufacturing process (surface quality and suitability for wirebonding; which had to be addressed by process changes at the manufacturing facility)
• problems with flex population (which required a revised design, population process and vendor change)
• Reliability studies for one of the involved ASICs (a linear regulator showing premature death when tested at currents close to the expected operation level)
• Discontinuation of the glue intended to be used for module assembly (requiring its replacement with a different epoxy glue in a time-consuming search)
• A brief update on the status of Cold Noise discovered on barrel modules and the current understanding of its mechanism
• the noise discovered on R4 and R5 end-cap powerboards requiring a re-design of the powerboard circuits
• permanent deformation of modules during thermal cycling due to the change to a glue with a
lower glass transition temperature in combination with overly high QC temperatures•
• Modules cracking on loaded structures when cooled down to detector operating temperature due
to a mismatch in coefficients of thermal expansion
• bPOL12V chips breaking due to a previously not exercised combination of irradiation, cooling
and thermal cycling
Challenges arising from other areas of the detector which were resolved by adapting the module
design (e.g. lost hits on the ASIC level resolved by adjustments to the hybrid wire bonding
layout, insufficient clearance between module powerboards and the global holding structure
requiring another end-cap powerboard redesign, low strip isolation on sensors requiring sensor
recovery measures at module assembly sites)
In addition to an analysis of the individual topics, this contribution will also present an overview of how the occurrence of multiple technical issues in parallel affected the programme by both slowing down the resolution of other challenges as well as the overall pre-production through the repeated need for re-tooling, design validation and resulting shortage of available components