Microchannel CoolingD. Bouit, J. Daguin, A. Jilg, L. Kottelat, A. Mapelli, J. Noel, P. Petagna, G. Romagnoli – CERN PH-DT

K. Howell – Georges Mason University / CERN PH-UFT

G. Nuessle – Université Catholique de Louvain / CERN PH-UFT

A. Pezous - CSEM

P. Renaud – EPFL-LMIS4

E. Da Riva, V. Singh Rao, P. Valente – CERN EN-CV

October, 28th 2011, CERNNA62-GTK special meeting on cooling options

Basic concept: couple to the heat production region a thin silicon layer structured with m-channels, where coolant is circulated

ADVANTAGES:

•Low material budget

•High thermal stability

•High thermal uniformity

•High efficiency (low fluid-sensor DT)

•Self-adapting to changing conditions

•Conventional single phase cooling plant

•M&O by proximity team (EN/CV for the plant and PH/DT for the assembly replacement)

1 Cooling System Requirements Absorption of nominal dissipated power ~100% Stabilization of detector temperature ~5 C Surface temperature uniformity ±3 C Material budget 0.15% Xo

2 Operation Aspects Meeting stable temperature Reaction time to power failure Reaction time for hydraulic failure3 Integration Aspects

Preliminary integration design Assembly operation sequence Test Plan

4 Cooling System Details Conceptual Design Cost Estimate Safety and reliability issues Infrastructure Requirements

Main conditions to be met by a candidate cooling system as defined by NA62 GTK

In the following presentations we will cover this list of conditions and prove that m-channel cooling is indeed a solid

and satisfactory option for application to the NA62 GTK

Sequence of talks:

•Overall design and production (Petagna)

•Experimental analysis of the thermal performance (Howell / Nuessle)

•Optimization of the geometrical configuration (Da Riva / Singh Rao)

•Electro-mechanical integration (Jilg / Romagnoli)

•Cooling plant (Valente / Battistin)

•Safety, reliability, failure prevention (Nuessle)

•Summary (Petagna)

Microfabrication

Microfabrication technologies are derived from the microelectronics industry. They comprise a wide variety of techniques used for manufacturing micro-devices. In the last decades, they have been widely used to manufacture micro-electro-mechanical systems (MEMS) and microfluidic devices.

These devices are typically made out of silicon wafers but other types of substrates are commonly used such as plastics and glasses.

Common steps in microfabrication comprise photolithography, thin films deposition and structuration, etching, thinning by grinding and chemical mechanical polishing (CMP), bonding, and dicing.

Microfabricated cooling plates are presently under investigation in our team for the upgrade programmes of the ALICE ITS and the LHCb VELO

Final cross section of the cooling plate

Acceptance

30 mm epoxy = 0.0008 % X0

10 mm Pyrex = 0.0008 % X0

(removed in final production)

70 mm C6F14 = 0.0037 % X0

Channels =200 x 70 mmWall thickness = 200 mmCover thickness = 30 mm

30 + 30 mm Silicon = 0.0064 % X0

(above and below channels)30 + 30 + 70 mm Silicon = 0.0139 % X0

(between channels)

Total material budget in the acceptance area = 0.013 X0%(min 0.011% - Max 0.015%)

Final top shape of the cooling plate

Optimized geometry for minimal pressure drop with double inlet / outletOperational pressure at nominal flow rate ~ 7 bars

(see presentation of E. Da Riva)

Estimated cost of the final cooling plate

About 1.5 kCHF / cooling plate for 20 plates with such offers…

Silicon Fusion Bonded cooling plates (All silicon devices with no Pyrex).

Cross section of the vehicle for thermo-hydraulic tests

380 µm

525 µm

Pyrex60 µm

Channels =100 x 100 mmWall thickness = 100 mm

Same structural critical point than the final device: 100 mm thick silicon cover over a 1.6 mm wide channel: operated without damage at 12 bars and tested to withstand > 18 bars

10 x Silicon thickness between channels and heat load with respect to the final device: all thermal test results are conservatives. Test device with “thinned area” hydraulically tested

Top shape of the test vehicle in use

Simplified geometry for ease of production and test with single inlet / outletOperational pressure at nominal flow rate > 12 bars

Thermal performance

•Power absorption

•Temperature stability

•Temperature uniformity

•Reaction to failures

Sample cooling plates following a first proposed design have been produced in-house (EPFL CMI) and tested for thermal performance. The production process is fully under control and the produced plates prove to be reliable. An extensive testing activity has been performed in realistic conditions:

Initially designed under the assumption of a uniform power distribution over the surface of the sensor assembly, the cooling plates have been successfully tested again more realistic conditions of uneven power distribution between the digital and the analog part of the chip. Due to inherent limitations of the cooling station available for the tests, the thermal performances have been analysed up to 80% the nominal flow rate. A simple extrapolation allows for an exact forecast of the performance under nominal conditions

Geometrical optimization

•Manifold geometry optimization (pressure drop and flow uniformity)

•Channel cross section optimization (pressure drop and material budget)

The tests performed on sample cooling plates produced with the “design 0” geometry showed their fully satisfactory behaviour with respect to the specifications on the thermal performance. However, there is space for further optimization towards material budget reduction and reliability enhancement (structural safety factor). This has been accomplished through a detailed CFD analysis.

The CFD optimization performed brings to an optimized geometry with double inlet / double outlet and minimized channel depth (i.e. enhanced performance, larger structural safety margin and reduced material budget). Cooling plates prototypes based on the new optimized design are presently under production.

Electro-mechanical integration

•Concept of integrated module

•Precision / repeatability issues during production

•Study of the jigs required for the production

•Preparation to the prototyping phase

The geometry produced by the CFD optimization has been integrated with the detailed understanding of the production cycle and led to a final cooling plate configuration. This configuration has been implemented in a full 3D CATIA model of the GTK module and a complete study of the electro-mechanical integration of the module has started.

The integration concept moves from the experience of the Totem Roman Pots, and LHC detector with several similarities to the GTK. The integration process proposed appears feasible and solid. The most critical issue, the gluing of the sensor assembly to the cooling plate, is being now directly tested in realistic conditions.

Cooling plant

•Local infrastructure

•Cooling plant specification

•Cooling plant draft P&I

•Cost estimates

The proposed microchannel cooling is based on a conventional single phase refrigeration plant. With a sound production, integration and operation process available, it is possible to correctly define the specifications for the cooling plant and therefore launch its definition.

The proposed cooling plant falls into the category of standard design and production already managed by CERN EN/CV-DC. All components and solutions are standardized on the basis of the LHC experimental cooling plants. The system is designed with full redundancy in order to minimize its downtime. EN/CV will be available to follow the production of the final cooling station as well as it maintenance and operation.

Safety, Reliability, Failure prevention

•Safety, Reliability and Failure prevention issues

The concept is robust against system failures with extremely limited control needs.The microfabrication procedure is now consistently producing highly reliable devices.A sound assembly concept is being developed to maximize the yield during integration.Preliminary studies performed by RP rule out any concern about fluid activation.Local vacuum monitoring interlocked to shut-off vacuum valves will be introduced for each GTK station for a timely action in case of highly unlikely accidental leaks.

The proposed cooling is highly reliable in operation and only requires extremely limited “slow controls”. No problem is expected by RP due to the irradiation levels. All details of the assembly procedures are being worked out and tested on realistic mock-ups in order to maximize the yield during the final module integration.

CONCLUSIONS•A fully reliable production technique for Si m-channel structured cooling plates suited for application in the NA62 GTK has been individuated

•A first design proposed for integration in the module (“design 0”) has been thoroughly tested for thermal performance in realistic conditions and proved to be fully compliant with respect to the specification proposed (temperature stability and uniformity, reaction to failures, etc)

•A CFD optimization study led to the definition of new geometry (“design 2”) enhancing the thermo-hydraulic performance, reducing the material budget and increasing the structural safety margin. Cooling plates based on the new design are presently in production.

•The optimized geometry (“design 2”) have been implemented in a full CATIA model of the GTK module and a full electro-mechanical integration study has been launched. Tests on ad-hoc produced mock-ups ongoing.

•The specifications for the cooling station have been defined and a preliminary design study has been performed by EN/CV-DC

SUMMARY: THE PROPOSED COOLING PLATEFlat coupling (direct adhesive gluing) to the sensor assembly

Dual inlet/outlet and manifolds (additional thickness) outside the sensitive area

PEEK connectors

Minimized material (double sided local thinning) on the sensitive area

130 m

m

All inclusive average material budget in the sensitive area:

X/X0 =0.13%(min 0.11%, Max 0.15%)

(Might finally go down to 0.11%)

SUMMARY: GLOBAL COST AND AVAILABLE TEAMFinalization of a few “design 2” prototypesCompletion of the integration studiesConstruction of assembly/integration jigsSmall items and contingency

Outsourced production of 30 “design 2” cold plates

Cooling plant(Design, production and commissioning: M&OIncluded in the M&O agreement with EN/CV)

…….…….50 kCHF

……………………………………………………………………180 kCHF

…..............…….20 kCHF

TOTAL ……………………..……………………………………………………250 kCHF

•PH/DT team (including technicians) available for module production and assembly / integration•EN/CV team available for cooling plant design production and commissioning

Availability to launch the activity: as from NA62 decision

OPTION: R&D ON “FRAME CONFIGURATION”R/O chips

Acceptance

No CTE b/t Si chips and Si frameAdhesive thermal interface

Silicon

(ALICE’s test cooling plates)

A detailed study about the “frame” configuration (similar to the one ongoing for ALICE ITS) can be launched as from January. However this would require additional resources:

•30 kCHF for cold plate prototyping / testing and to adapt the module integration procedure

•Ideally, procurement of a few “flip-chip” silicon thermal mock-ups (CSEM investigating this)

•CFD support from EN/CV team to be discussed