Dear the various "rphi like" module design proponents: We have received a note ("Evaluation of the various rphi modules at KEK") from KEK. In order to stimulate mutual understanding and collaboration, we distribute the note to the proponents (and CC to instiututes). We would like to hear your response to the note, (or to the various designs, or whatever else to be constructive). A plan is to clarify questions by the May Munich SCT meeting, to draw an agreed measurement plan by June, and hopefully a convergence by Summer, then, solve all the details of a design by the end of year. Unno and Carter Module convenors =================== Evaluation of the various rphi modules at KEK 1996/3/28 ATLAS-J/Silicon group of KEK "Baseboard rphi-Z" module [1] Cooling (1) The cooling method on the drawing seems to have a large thermal resistance from the cooling medium to the heat spreader (i.e., "pipe-(thermal grease)-cooling block- (not a wide contact area)-cooling finger(s)). The cooling pipe has to move at the contact area due to thermal contraction (assuming ZERO CTE stave), which means there must be a sliding joint (i.e., thermal grease?), then, the cooling is not very efficient due to the transitions. The contact length seems need to be elongated. Is there any spring-effect to make a good contact of the pipe and the heat spreader? Is it done by the skrewing force? (2) The length of the cooling point to the far end of the IC chips is 6 cm. The area will be hot, somewhere like 10 deg.C above the near end to the cooling point. When the modules are overlapped in z-direction, these chips are facing the detector itself (at least at one end; sandwiching the air of about 2 mm). The far end of the detector must be already a few degrees above the cooling point, and, adding the heat by radiation (and convection) from the IC chips will certainly heat up the far end of the detector; thus it is very likely lead to the thermal runaway. This is the most serious issue of the design and the thermal performance must be measured for the overlap. (detector) (hybrid) -----------------======== ++++ (chip) ++++ =========--------------- [2] Position accuracy and stability (1) The cantilever supporting is vulnerable to vibration/acceleration. The module is very heavy due to the baseboad etc. with respect to the strength of the cantilever structure and the material. (2) There is a concern on developing cracks (and eventually to breakage) in the BeO at the support due to stress concentration (such as static gravity, thermal distortion, vibration), where the BeO will be fixed by skrewing with a relatively strong force to have a good contact to the cooling block and is cantilevered. (3) The position of the farend of the module from the fixing point is sensitive to the accuracy of the plane of the support tube. The plane would be distorted by, e.g., a small amont of (grease) thickness variation due to skrewing force, creaping in time, and, thermal distortion of the support tube. (4) It seems rather difficult to keep the exact step and flatness of the three supporting points since the "baseboard" and the "cooling finger" are not a one-piece design. The same for the "mounting pads". When the module is fixed with too-strong force, it is very easy to generate distortion, and eventually leads to breakage. [3] Amount of material and space (1) There seems a lot of materials are involved in the design (baseboard, cooling finger, ...). There is material concentration in the stave area (stave itself, cooling block, cooling finger, cables,...). (2) The radial separation of the adjacent modules must be more than 2 mm (3 mm confortably?) to overlap the modules in the z-direction because there are readout hybrids (chips and wire-bonds) at the z-end of the modules (facing each other at one end). Also, to make the f(phi) overlap, the tilt angle must be larger due to the bigger radial separation of the z-overlap. This leads to a smaller effective module size and thus a higher overall-cost (slightly?), a more material. [4] Others (1) Carbon-fiber and its composite material have quite a large difference in the thermal conduction in directions. Relevant chracteristics must be measured before fixing the design when these materials are used in the critical area, such as cooling. (2) No material on the surface of the detector: This is the advantage of the design. However, the price to pay seems to be large: [1], [2], [3]. "T/rphi-Z" module [1] Cooling (1) Not a good cooling efficiency due to the transitions of "pipe-(thermal grease)- cooling block-(not a wide contact area)-cooling finger (as same as the "Baseboard" type) for the ZERO CTE stave and a separate cooling pipe. (2) Concern on the amount of material to have the temperature of the far-end chip to an acceptable level which is constrained by the thermal runaway at the far-edge corner of the detector (some 6 cm away from the far-end chip/heat spreader). [2] Position accuracy and stability (1) The cantilever supporting is vulnerable to vibration/acceleration. The module is reasonably heavy with respect to the strength of the cantilever structure and the material. (2) There is a concern on developing cracks (and eventually to breakage) in the cooling finger at the support due to stress concentration (such as static gravity, thermal distortion, vibration), where the finger(s) has to be fixed by skrewing with a relatively strong force to have a good contact to the cooling block and is cantilevered. (3) The position of the farend of the module from the fixing point is sensitive to the accuracy of the plane of the support tube, e.g., a small amont of (grease) thickness variation due to skrewing force, creaping in time, and, severly due to the thermal distortion of the support tube. [3] Amount of material and space (1) It looks like more materials are in the design (BeO or Carbon-brick). There is material concentration in the stave area (stave itself, cooling block, cooling finger, cables,...). [4] Others (1) Carbon-fiber and its composite material have quite a large difference in the thermal conduction in directions. Relevant chracteristics must be measured before fixing the design when these materials are used in the critical area, such as cooling. (2) Glue/adhisive on the surface of the module. -> See the [4]-(1) of the "classic rphi" module. "Oxford Spine" module (1) "No sliding joint": Is there a realistic material which statisfies the CTE of null to that of Silicon, and simultaneously having a high thermal conduction coefficient, K? (2) "Straight cooling pipe": What is the material of the cooling pipe? (3) "Direct cooling of the chips": The design which has the readout hybrid at the end of one module and extra-heating of the detectors by radiation (and convection) where the adjacent modules are overlapped in the z-direction. Need confirmaition that this is not a problem. Also, in this design, due to the z-overlap, the readout hybrid must be cooled from one direction; there is a detector underneath the hybrid. (4) "Simple z-alignment": Need a detailed drawing to understand how this is acomplished. The design which has the readout hybrid at the farend of a module requires a larger radial separation for z-overlap, say 2 to 3 mm. A larger tilt angle in phi direction; smaller effective module size; higher cost. (5) "Easier assembly": It seems very difficult to repair, due to glueing, sandwiching by cooling pipes,... Essentially this design implies no repair for a stave. Very severe drawback. (6) "Direct glueing of support and cooling": This is a big concern. Silicon, cooling pipe, etc. are hard enough. Any distortion (e.g., thermal) will lead to a damage on the surface of the detector, e.g., breakage of the passivation and readout couping, etc. "Classic rphi" module (e.g., KEK type D (with zig-zag cooling tube) [1] Cooling (1) The design (D) has an efficient cooling since the cooling pipe is along the IC chips. An issue of the design is the handling of the zig-zag Al cooling tube, mitigating/escaping the thermal contraction of the Al tube. Two possibilities are pursued: to design the zig-zag Al tube in the local support underneath the module, or on top of the module. The latter is easier to design the local support but has the handling issue. The former is the best for assembling and repairing, however, requires a solution to mitigate the themal contraction. (2) The cooling is efficient because the Al pipe is brazed to the metal heat spreader (called "cooling plate"). The contact of the module's "shield plate" and the "cooling plate" is done with a thin thermally-conductive silicon-rubber (instead of a thermal grease) which is "dry" rather than "wet" (grease). The usage of "dry" type is possible because the thermal movement is small, limited within a length of 6cm of Al tupe (because of the Zig-zag). [2] Position accuracy and stability (1) A module is supported in a wide area, thus very stable. A small local distortion is averaged out, and no amplification of distortion in the far ends. An issue is the flatness matching of the two supports in one module. (2) A module is supported indirectly in wide area with a spring-force (e.g., no direct glueing/skrewing to a support). Little stress concentration due to extra force from outside. The module itself is very rigid with the help of two (i.e., top and bottom) metal shield/thermal plates. Also the local support plate can be designed to be strong enough not to introduce a large distortion to the module due to the spring force. (3) The module is very symmetric (top and bottom). No thermal distortion is introduced in the module itself (This is already demonstrated with the thermo- mechanical distortion measurement at Oxford using the ESPI system). (4) The Be shield-plate is a one-pieace design: good flatness, no step in a module. [3] Amount of material and space (1) The material (radiation length) of a module (excluding the cooling pipes and the local support) is very small, 0.85% Xo. The design has the least material among the designs. (2) The minimum radial space (e.g., less than 1 mm) can be acomplished in the z-overlap because the readout hybrid is placed at the center of the module and only the silicon detectros at the far ends. (3) Smaller material concentration due to spreaded local support and cooling. Type E design (offsetting the hybrids on top and bottom) will have more uniform/spreaded material distribution. However, this design is required to be confirmed for the thermal distortion because the mechanical symmetry ( top and bottom) is broken. [4] Others (1) Hybrid on the surface of the detector: If this is turned out to be prohibitive (with very fundamental reason), this design (or any design of this sort) has to be abandoned. Two issues: (a) glue/adhisive on the surface of the detector, (b) Metal (grounded) over the surface sandwiching the glue/adhisive. The issue of (a) has been shown to have no fundamental problem (shown in the 95 H8 binary modules, the Feb 96 binary modules for the beamtest at KEK). The hybrids are glued on the surface and no serious degradation of the performance was observed. (This has to be shown again and again for more samples, of cource.) The issue of (b) needs to be demonstrated. The 95 H8 and Feb 96 KEK modules have a ceramic substrate glued on the surface of the detectors. The distance of the ground metal and the surface of the detector is somewhere 500 microns. The current design (at KEK) is that the Be metal base/shield plate is 100 microns away from the detector surface sandwiching a Silicon RTV (dry-type) adhisive. If this distance is turned out too critical (adding too much capacitance), a fall-back is: enlarging the distance, say, to 200 microns, or to use a ceramic substrate instead of the Be metal base/shield plate (by paying a more material budget). (2) Material availability: In Japan, other than BeO are available. Note that Be is less toxic; BeO is the toxic material. Conclusion After the evaluation of the various rphi designs, KEK still marks the "classic rphi" desgin the best; other designs have more concerns (for KEK, at this stage of understanding). KEK would like to listen the explanations of the proponents on the correctness of the understandings.