1. Pixel Upgrade Local Supports Based on Thermally Conducting Carbon Foam E. Anderssen, M. Cepeda, S. Dardin, M. Garcia-Sciveres, M. Gilchriese, N. Hartman, T. Johnson, R. Post LBNL W. Miller iTi

2. Outline Concept Implementation Foam development status Prototypes tested and planned Thermal performance studies Material estimates Issues Conclusion 2

3. Concept Use carbon foam for heat conduction and structural support Couple to round tubes to allow high pressure operation Same basic concept for ALL pixel regions – outer(permanent) layers, inner(replaceable) layers and disks 3 Carbon fiber (or higher K material) Round tube with “glue” Thermally conducting carbon foam Heat flow

4. Pixel Upgrade Inner Layers - Examples Monolithic structures – R  4 cm only – Modules one side – Modules alternate sides Single-sided staves – R  4 cm – R  7-10 cm Single-chip modules(e.g. 3D) 4 Potential cable location

5. Coupled Layer Structures Similar to ALICE. Also being considered for STAR upgrade(at LBL) Thermal performance not estimated Might result in material reduction within active region. 5 BOX Section Beam, single cooling circuit. I Section Beam, single cooling circuit. Foam Molded carbon fiber structure

6. Outer Layer Staves 6

7. Outer Layers - Structure Meter-scale staves Modules on both sides Staggered for coverage Large area, simple, fast and cheaper for larger R pixels 7  250 W per stave About 4 cm wide, length can vary to fit layout No real work done yet on support structure, deflections

8. Disks Hermetic disk structures possible by staggering modules in R (by  1mm active) and in z to give offset in . Electrical connections(cable) complicated! Very early thermal FEA shows step/overhang OK Edge HV? Much more work needed! 8 Active both sides Active one side Module envelope Facing Foam Tube Module cable to side

9. Foam Development Carbon foam has been around for many years, some  40 years – see brief history herehere So why is R&D needed? Density  thermal conductivity (K) Readily available, good thermally conducting carbon foams have density (  )  0.5 g/cc. These are graphitic foams Readily available, low-density carbon foams (   0.05 g/cc) have very poor thermal conductivity. These are reticulated vitreous carbon (RVC) foams. R&D involves, primarily, lowering the density of the graphitic foams but keeping a reasonable K (and mechanical strength) and raising the K and the density (but not so much) of the RVC foams. This requires close collaboration with the manufacturers of these foams (currently working with three companies) but some techniques are proprietary. 9

10. Leading Candidate Foam Development work is being done by Allcomp, IncAllcomp, Inc The starting material is RVC foam of density  0.05 g/cc – Porosity is typically 100 ppi (pores per inch) although lower porosity also being done for other applications – Highly thermally conducting carbon is added to the ligaments of the RVC precursor through a combination of chemical vapor deposition (CVD) and heat treatment – First examples for pixel structure prototypes raise density by factor  4 but thermal conductivity by factor 100 – 200 – Strength also enhanced. Readily machined. Continuing work with graphitic foam vendors (POCO, Koppers) as backup 10 More info herehere

11. 11 Pixel Prototype Components Tube with CGL7018 Very thin YSH-70 or K13D2U facings glued to foam Tube in foam with CGL7018  6cm long  2.5 cm wide One prototype 20 cm long. The rest about 6 cm long  0.8 cm thick Tube OD  2.8mm Applicable to CO 2

12. 12 VG 12 Simple Thermal Test Set-Up Silicon heater Infrared image and water coolant at room temperature 20 cm prototype

13. Simple Summary of Prototypes Results are for single-sided heating. Same after thermal cycling from +25C to -30C (50 times).  T ave /W is averaged over dummy heater. 13 Foam  (g/cc) K(W/m-K)  T @ 0.63 W/cm 2  T ave /Watt Allcomp 1 (RVC) 0.18~ 6 Measured  10 ~ 1.2 Allcomp 2 (RVC) 0.21  8 Measured  8.5 ~ 1.0 Allcomp3 (RVC) 0.17  18 Measured  7  0.85 POCO 1 (Graphitic) 0.09 ~ 17(z) ~ 6(x-y) Vendor guess  11 ~ 1.3 POCO 2 (Graphitic) 0.13Not known yet  9  1.1 Koppers (Graphitic) 0.21 ~ 15? Extrapolation  8 ~ 1.0 20 cm

14. Next Prototypes Allcomp, Inc – Will fabricate short prototype(s) for practice – Make two long (  1m) 4cm wide prototypes of outer staves. One to be tested(thermally) at LBL and the other at Allcomp, Inc – Timescale by Jan-Feb. 2009 LBNL – Making dummy platinum-on-silicon heaters to simulate 4-chip modules for outer layers. – Fabricating now  20 cm long, 1.2 cm wide, 5.5mm thick simple prototype for 3D sensor-FEI3 module mounting tests – Additional short prototypes to help understand stress issues(see Issues at end), adhesives and effects of thermal cycling. – Inner layers prototypes. Box or I-beam structure? TBD 14

15. Thermal Performance Outer staves – 4 cm wide, modules both sides mounted on metal/kapton cable for power, HV and signal routing. 1 cooling tube 4 cm wide but 2 tubes. Look at single and doubled sided heating Simulates also 2 cm wide inner stave by symmetry 15

16. Many Static Cases Studied 16 Properties Fixed cooling tube temperature.

17. Thermal Runaway – Results 17 Innermost stave, 2cm wide, 3D modules,“cable” not under modules Outer staves, 4cm wide Modules on both sides Mounted on cable. 250W. CO2 cooling is assumed. -35C but take into account  T in fluid(3C), heat transfer coefficient

18. Tube Contribution Simple calculation of  T for 250W for 1.1m long, 4cm wide stave. 18

19. Implications of Thermal Studies Possible to meet all thermal requirements but to do so with minimum material requires much more measurement, prototype construction/test as well as design. Need to measure K of all major components{foam(now routine)}, cable underneath module for outer layers, carbon- fiber facings, adhesives….) Compare prototype results with FEA(good agreement so far on small prototypes) Work on increasing K of cable for outer layers (higher thermal conductivity base materials seem possible but have to make prototype) 19

20. Material Estimates Example below(near minimum, I guess) for outer stave, modules staggered both sides. Higher with stainless, two tubes, thicker wall => understanding tube important. 20

21. Issue Interface of metallic tube and foam, CTE mismatch – If very compliant (grease-like), reliability after many hundreds of thermal cycles + after irradiation? – If not very compliant, stresses are calculated to be significant. Does the foam break? See talk here for more information and background.here 21

22. Stresses Simplistic summary is that calculations indicate that stresses for rigid bonding can be close to edge. Not easy to model tube-adhesive-foam interface => measure foam properties, make prototypes and test – Stresses concentrated at ends (where tube exits) and are practically independent of length=> can use short prototypes to assess effects – Note that thermal cycling of all prototypes made so far shows no effect but all have been made with very compliant interface. Just starting prototypes with bonded joint between tube and foam. Substantial program of work(help!). Large overlap with same issue for strip staves that use POCO foam. See talks at MIWG meeting this workshop. 22

23. Conclusion Use of thermally conducting carbon foam for local supports for ALL regions of SLHC pixel detector continues to be promising Thermal requirements can be met for all regions but to minimize material requires extensive materials measurement and prototype program Most critical design issue: reliability of connection between tube(metallic) and foam after extensive thermal cycling. Generally reliability engineering must increase: radiation to 10 7 Gy,  1000 thermal cycles, corrosion possibilities,….. Adhesive engineering to meet reliability and other properties. Extensive measurement and prototype program underway but lots to do! 23