INTEGRATED COOLING CHANNELS IN POSITION-SENSITIVE SILICON DETECTORS

Transcription

1 20TH INTERNATIONAL WORKSHOP ON DEPFET DETECTORS AND APPLICATIONS INTEGRATED COOLING CHANNELS IN POSITION-SENSITIVE SILICON DETECTORS L. ANDRICEK, M. BORONAT, J. FUSTER, I. GARCÍA, P. GOMIS, C. MARIÑAS, J. NINKOVIC, M. PERELLÓ, M. A. VILLAREJO, M. VOS

2 CONTENTS 2 1. MCC case for physics detectors 2. MCC DEPFET-like module 3. Finite element simulation 4. Setup & results: 4.1. Thermal performance 4.2. Mechanical impact 5. Cooling a whole powered module 6. A more realistic approach: considering the bumps 7. Next steps & summary

3 MCC CASE FOR PHYSICS DETECTORS 3 PXD Cooling and support structure Belle II cooling structure would be too massive to higher acceptance detectors like ILC.

4 MCC DEPFET-LIKE MODULE: PRODUCTION An integrated cooling channel is designed for a DEPFET module, focusing in the EOS. The MCC production adds one extra step to the chain: etching the µchannel in the handle wafer. ntroduction n the sensor Handle before bonding Dummy X-ray 4

5 MCC DEPFET-LIKE MODULE: CONNECTORS 5 In order to feed the cooling circuit a number of connectors to interface with commercial fitting elements have been designed: H2O Past (0.81% X/X 0 ) Present (0.2% X/X 0 ) Future (0.05% X/X 0 ) Up to 183 bar 0.05% X/X 0 /5 cm Self aligning connector 3D-printed (15 µm precision) Glue sealed

6 FINITE ELEMENT SIMULATION (I) 6 TEMPERATURE = 25 ºC H2O 6W Low-cost mono-phase cooling liquid: H 2 O. Low volumetric flows (~1 l/h) and low pressure (< 1 bar) are enough to dissipate 6 W in the EOS. Possibility to use CO2 at high pressure, but not necessary at the power densities studied. /W] 2 T/Power density [K cm 6 FE Simulation H 2 O FE Simulation PWG Volumetric flow [l/h]

7 EXPERIMENTAL SETUP: SCHEME 7 ROOM TEMPERATURE ~25 ºC

8 EXPERIMENTAL SETUP: REALITY 8 50 KHZ INFRARED LASER FLOWMETER PURITY FILTER SHOCK ABSORBER PERISTALTIC PUMP WATER STORAGE AIR COOLING Air H2O TERMOMETERS CLAMPED-FREE MCC SI MODULE

9 RESULTS: THERMAL PERFORMANCE (I) 9 Errors: P: ±1% W T: ±1 ºC Flow: ±0.03 l/h MCC dummy cooled non-stop for a week with no leaks and no clogging. Good agreement with the FE simulation (within 10% error).

10 RESULTS: THERMAL PERFORMANCE (II) 10 H2O 0-22W Max. power supported for ΔT of 10 ºC as a function of the volumetric flow: Power capped at max. pump power ~3 l/h Low pressure measured: bar

11 RESULTS: MECHANICAL IMPACT (I) 11 Signal [mm] NO fluid circulation Signal [mm] Fluid circulation 50% (1.47 l/h) Signal [mm] Air flow (v=3m/s) No fluid circulation and no air flowing Peak to peak of the signal ~0.7 μm RMS ~0.3 μm Time [s] Fluid circulation (1.47 l/h) Peak to peak of the signal ~0.1 μm RMS ~0.4 μm Time [s] Air flowing (3 m/s) Peak to peak of the signal ~130 μm RMS ~57 μm Time [s] MCC has no significant impact on mechanical stability in the clamped-free configuration but air deformations are over 100 μm for v = 3 m/s.

12 WHOLE POWERED MODULE: HYBRID APPROACH 12 H2O 6W 1W 0.5W Air 0.5 m/s SENSOR HOTTEST POINT C] o T [ 60 Sensor: MCC 50 Sensor: MCC+air Volumetric flow [l/h] Cooling strategy: micro-channels running under the front end and gentle air flow on the sensor part. Big difference between MCC and MCC+air at the sensor area hottest point. Nearest regions to air input are efficiently cooled even with low air flow. MCC has less impact in away points as expected and great cooling locally.

13 WHOLE POWERED MODULE: MCC ALTERNATIVES 13 Standard MCC layout ΔT = 73 K Front end HOTTEST Standard MCC layout + channel below switchers ΔT = 15 K Standard MCC layout + channel below switchers + channel in the balcony ΔT = 5 K

14 6W H 2O T/Power density [K cm2/w] MORE REALISTIC APPROACH: BUMPS 14 9 FE Simulation H2O 8 7 FE Simulation H2O realistic design Front end HOTTEST POINT Thermal studies Volumetric flow [l/h] Realistic design 300 μm Si ASICS μm Bump-boundings thermal resistivity of 6 W/m K (a) Carlos Mariñas PhD Thesis In the realistic design the power dissipation is degraded (b)

15 NEXT STEPS 15 Test the radiation resistance of the 3D-printed connectors. Repeat the thermo-mechanical measurements for the new designs of the connectors. Reproduce the study for the more realistic approach, with bumped resistors instead of printed ones. Produce and test the thermo-mechanical properties of the whole powered modules of the new MCC alternative layouts.

16 SUMMARY 16 MCC shows very efficient local cooling, up to 25 W/cm 2 for ΔT ~ 10º C using low pressure mono-phase cooling liquid. The thermal measurements agree with the FE simulation. MCC has negligible impact on the module mechanical stability. Three in-plane connector concepts have been designed and manufactured, going towards less massive connectors. MCC modules have been successfully assembled (in 3/3), operated non-stop for a week, and supporting pressures up to 183 bars. These features qualify MCC as a real option for silicon detectors in physics.

17 THANKS FOR YOUR ATTENTION THIS STUDY IS SUPPORTED BY THE AIDA2020 THERMO-MECHANICAL PACKAGE MORE INFORMATION AVAILABLE AT ARXIV:

18 BACKUP

19 VIBRATION S SPECTRAL POWER DENSITY 19 PSD [mm 2 /Hz] 10-4 NO fluid circulation 10-6 Ladder eigenfrequency (~150 Hz) Fluid circulation 15% (0.45 l/h) Fluid circulation 50% (1.47 l/h) Air flow (v=3m/s) Frecuency [Hz]

20 VIBRATION AMPLITUDE VS. AIR SPEED 20 [µ m] RMS y Peak to peak χ 2 / ndf / 5 Prob p ± p ± p ± [µ m] RMS y Peak to peak 8 χ 2 / ndf / 5 Prob p ± p ± p ± Clamped-Free v[m/s] Clamped-Clamped Peak-to-peak amplitude is the change between peak (highest amplitude value) and trough (lowest amplitude value) v[m/s] RMS (PeaktoPeak/2) * (approximation) For v= 2.5 m/s the amplitude of vibration is: ~19 μm for clamped-free configuration ~2.8 μm for clamped-clamped configuration