Glue Programme Material for Berkeley (Tim)
Contents Why do we need to have a glue programme? Polling the Collaboration First thoughts on a Programme Outline 1/9/13Glue Programme Discussion2
Origins Came out of ‘decision’ to migrate from SE4445 to ‘something else’ for module mounting during stave production. Keep SE4445 during prototyping as there is ‘a chance’ damaged/failed modules can be removed General consensus is that ‘something more sticky’ should be used for production General realisation that glue choices across project are not ‘well defined’. – Likes and dislikes – Reasons for choice lost in history What’s Been Done ? – Attempt to elicit information by polling whole community 1/9/13Glue Programme Discussion3
Polling the Collaboration 1/9/13Glue Programme Discussion4
What Surfaces are Glued? 1/9/13Glue Programme Discussion5
How are Surfaces Prepared? Good surface preparation procedures are essential for reliable glue joints But, they are notoriously difficult to define and ensure compliance to… – Eg. “abrade surface and clean up with alcohol” What grit? How much pressure? How often should the paper be changed? How deep should the surface scratches be? Which alcohol? What sort of wipe? How should the part be dried & stored? For how long is the procedure effective? Is it ‘safe’ to repeat & if so how many times, etc, etc..? – Online surface preparation guides Hysol: – Distinguishes between ‘cleaning’ and ‘surface preparation’ – Extensive use of chemical processes Our parts… – Are quite delicate… Thin sections of CFRP (<0.2mm), cooling tubes (~ 0.1mm) Sensitive areas: silicon sensor, bond-pads, etc… – Are small / fragile making handling difficult End close-outs, stave mounting components, etc… – Have minimal ‘surface’ Thermally conducting foam The tops & bottoms of honeycomb walls 1/9/13Glue Programme Discussion6
Cleaning Processes Clean Surface Alcohol Abrasion Plasma & Corona Chemical Vopor Honing Ultrasonic Blow- off & Vacuum Web-search reveals many different techniques – see graphic…. All come with dis-claimers. – Need to be satisfied any technique fulfils ALL requirements (including collateral damage) – Some techniques may require access to specialist equipment Try to make progress by … – Deciding which techniques are suitable for which substrates – Initiating a programme of work to identify which glues offer best chance of working with the cleaning desired cleaning regime Talk to industry – Preparation of samples? 1/9/13Glue Programme Discussion7 CO2 Snow??
How Much Preparation can we Tolerate? Silicon Remove particulates with air duster Plasma surface activation Alcohol Bus Tapes Alcohol Plasma surface activation Mechanical abrasion ? CFRP Face sheets AlcoholPeel ply? Mechanical abrasion? Other? Titanium tube De-grease?Abrade? Chemical Treatment? Thermal Foam ? ….. ? Suggest starting with a web-search – Details of ATLAS construction (Can we find out what was actually done?) – similar projects (CMS, AMS), – Aerospace (NASA,ESA, …) – Electronics packaging & general industry (eg Hysol guide) Begins to open up assembly issues….eg…. – If the bus tape is aggressively cleaned before lamination, how do we keep the top surface clean during stave assembly OR if the top surface is cleaned after stave assembly we need an effective procedure which is physically and chemically compatible with the stave? 1/9/13Glue Programme Discussion8
Surface Characterisation Techniques Clean Surface Contact angle Dyne Test Pens ? Electron Microscopy Surface Roughness ? Need to … – Compile complete list of useful surface analysis tools and understand which are applicable to each substrate – Understand what facilities are accessible to the collaboration to characterise surfaces Equipment in labs Contacts with partner institutes / industry – Initiate a programme to evaluate the surface properties of different substrates cleaned & treated using the allowable procedures 1/9/13Glue Programme Discussion9
Measurements of Adhesive Properties Here I’m not considering ‘Application Testing’ – Hybrid-silicon peel test – Bus-tape CFRP peel test – ASIC-hybrid push-off test – Silicon-bus tape peel test Information should be available from suppliers! Many tests use ‘industry standard’ equipment – DSC, DMA, Universal Materials Testers Identify institutes with suitable equipment & effort to collate manufacturer’s data 1/9/13 Thermal Conductivity Tg Thermal Tensile modulus & strength Shear modulus & strength Mechanical Packaging Pot life & gel time Cure temperature & time Processing Radiation Temperature 10 years Humidity Environmental Glue Programme Discussion10
Application Testing Verification of ‘real life’ performance eg… – What is the (change in) force required to pull off a 1cm 2 baby silicon detector from a section of bus tape after 100 thermal cycles and 1.5xLHC dose? – What is the (change in) peel strength for the bus-tape from the CFRP face-sheet after 100 thermal cycles & 1.5xLHC dose? – What is the(change in) thermal impedance through a 1cm2 x 0.5cm thick bus tape/CFRP/foam/CFRP/bus-tape sandwich after 100 thermal cycles and 1.5x LHC dose? Such tests are ‘expensive’ – How many tests? Ideally would like 5-10 tests of identical samples Need to minimise (=1) process variations – Tests will involve many parts from different sources moving between different locations Full life-cycle tracking of ALL components needs to be in place for every step at every location Probably need to identify a few ‘simple’ tests – NB ‘Simple’ = well understood geometry, components, assembly processes, environmental conditioning & test protocols 1/9/13Glue Programme Discussion11
Programme Outline Application testing using prototype parts Verify basic glue properties against manufacturer’s data Review default / identify new glue candidates & compatibility with desired surface preparation process Poll community for advice on surface preparation & surface analysis equipment Document & circulate requirements I (Tim!) would be willing to coordinate the programme Before Upgrade Week:- – Document & circulate requirements – Poll community for info on surface preparation & analysis Hold a ½ day meeting during upgrade week – Review requirements – See who’s interested and what they can contribute – Begin to define a ‘plan’ Goal might be to have an informed choice for the SE4445 replacement by end of 2014? 1/9/13Glue Programme Discussion12
Backup The I sent to the groups (1 slide) The responses I got (3 slides) Incomplete Glue Matrix (1 slide) A RAL “designer epoxy”. 1/9/13Glue Programme Discussion13
Request to the Groups I'm trying to compile a survey of the different adhesives which are being considered for use at various stages of strip stave and petal assembly. The following list is a compilation of different gluing steps from hybrid assembly all the way through to stave core assembly. 1) ASIC to Hybrid 2) Hybrid to Silicon 3) Silicon to bus tape 4) Silicon to HV bias pad (if different from #1). 5) Honeycomb to Carbon-fibre face sheet 6) Cooling tube to thermally conducting foam and foam to face-sheet I'd be grateful if you could me any information on any gluing experience (both good and bad!) for any of the above steps. Also, I'd be especially interested to learn if the subject of adhesives is either one which interests you or in which you have expertise! Many thanks, Tim 1/9/13Glue Programme Discussion14
Responses Ingrid – I think that would be very interesting to join forces. I actually asked Luise to do a survey of what studies have been done earlier on when the current detectors were prepared. Especially to understand what kind of radiation hardness and aging tests were done. For me it was not really clear if it is reasonable to do standard aging tests were a humidity step is included (one cycle with higher humidity, percentage not really defined). Did you do something like this before ? Nigel – I am interested. For petals, we glue silicon to face sheet (or more correctly, silicon to parylene-C coating on face sheet). And we glue bus tape to face sheet. You could also add adhesives used inside the bus tape. – Our experience so far: Dow Corning 4445 for silicon to facesheet: Spec has break-down field on the edge for endcap use. We have bought some and will try it. It is very expensive in the Netherlands. Richard F – I spent from 1997 to 2003 studying polymer behavior including some glues so my expertise is more the glue structure and how to measure this such as contact angle measurements, surface measurements such as wetting effect etc etc. – Bottom line is this was something I could do, have the test kit in house to access and am interested in. However I'm not sure how much to bite off here but would like to help. Dave Lynn – BNL/Yale uses for (5) Hysol with 30% BN. There appears to work well although we have yet to do detailed testing. For (6) cooling tube to foam we seal the foam with hysol+30 % BN and then use CGL. We plan to move to an Hysol + 30% BN only solution as the UK has had good results with this (but I have not yet seen any thermal contraction measurements). But so far we seem to have good results with the CGL (based upon camera thermal imaging) but cannot yet quantify. For (6) foam to facing, we again use Hysol +30% BN with good results (again only based upon thermal imaging). 1/9/13Glue Programme Discussion15
… and more… Steve McM – During our discussion on Friday I said that I would pass on the headline topics that came up in the discussion with Martin on the Thursday Pixel in Manchester. Items to be added to the list of things to look at in the glue group. – In no particular order they were 1. Thermal conductivity 2. Loading the glues to improve TC (Boron nitride etc) 3. Glass transition temperature 4. Radiation tolerance 5. Pot-life 6. minimum order quantity 7. Time to delivery 8. Cost 9. Preparation details 10. Disposal at end of use 11. MSDS (datasheet safety) 12. Viscosity 13. Rework issues 14. Expansion or contraction of the glue on curing 15. Is the glue hygroscopic 16. Cure temperature 17. The temperature (value and stability) of the cure environment 18. Hardness 19……. 1/9/13Glue Programme Discussion16
… and the big one! Eric Anderssen – I've seen this request before, and wish the best, but what are you trying to get at? – An old trope in the adhesives business is that all adhesives are the same - it's all about how they get into the joint; their properties during application/assembly are as, or more, important than their cured properties, because they all perform similarly after cure. When I say 'old trope' I mean old--I heard this at industry conferences back in '89, and it was common knowledge before I started attending--admittedly from companies in the business of dispensing said adhesives, but they had (have) a point. – An adhesive joint isn't reliably predictable if you can't control how much adhesive gets there, how thick it is, or how well filleted the joints are. The properties of mixed 2-part systems or defrosted 1-part systems (pre-cure/dispensing) trump 'ideal' cured properties because control of application properties yields controlled cured properties. The 10-50% variation in 'measured cured properties' reported by various institutes are likely completely obscured by their assembly processes. That is a problem. If a syringe applied (low viscosity) thermal adhesive performs better than a screed-mask applied (thixotropic) adhesive, post-cure, is it because one adhesive is better than the other, or is it because the folk that report have better control over their process? – In my experience, assembly process dominates adhesive selection. Not to confuse the issue, but at LBNL we either select, or modify, adhesives specifically for their characteristics during assembly. For example, if we want or need to use 9396, but need it to be more viscous to stay where it belongs in a joint, either we wait during it's pot-life until it becomes more viscous (partially cured), but if the process needs more time, we add Cab-o-sil (microspheres and talc) to change it's viscosity in-pot-life allowing time to introduce the adhesive to the joint. In terms of performance, there is a lot of overlap. – How do you plan to normalize the data you request across the numerous and variable processes (and internal modification of adhesives)? 1/9/13Glue Programme Discussion17
…. And one from 2 day’s ago Tony Affolder – ASIC to Hybrid The Tra-duct 2902 (silver loaded epoxy) works well. It is thick enough to screen print and has a good working life. There are a few issues with it: It is fairly expensive and has a 3-4 month delivery time The cure time is too long. Ideally we would want something that hardens to the point that we can remove the vacuum in under 1/2 a working shift. So workable for 1/2-1 hour, hardened by 2-4 hours, cured in 12 hours – Hybrid to Silicon The Fuller Epolite 5313 is more problematic. It is too thin to screen on well. And pre-curing to thicken isn't great. So we need an electronics- grade epoxy with the following characteristics Thick enough the screen on workable for 1/2 hr-1 hr cures in 2-4 hour without shrinking or expanding which doesn't damage the silicon radiation hard 1/9/13Glue Programme Discussion18
Incomplete Matrix Gluing StepAdherends Required Properties Cure ConditionsCandidates ASIC to HybridSilicon to Au/Cu Thermal conductivity Electrical conductivity ? T ≈ 20°C P ≈ 0barg t ≤ 24h Tra-duct 2902 Hybrid to SiliconSilicon / Kapton Compatibility with silicon T ≈ 20°C P ≈ 0barg t ≤ 24 hr Fuller Epolite FH-5313 Silicon to Bus- tape Silicon / Kapton Thermal conductivity Compliance T ≈ 20°C P ≈ 0barg t ≤ 24 hr SE4445 Silicon to HV bias pad Silicon / Gold Electrical conductivity T ≈ 20°C P ≈ 0barg t ≤ 24 hr Tra-duct 2902 (?) CFRP face-sheet to core CFRP / (Nomex / CFRP) Mechanical integrity T ≥ 20°C P ≤ 0barg t ≤ 24 hr Hysol 9396 (+boron nitride) Hysol NA CFRP face-sheet to Th. Cond. foam CFRP / Carbon Thermal conductivity T ≥ 20 °C P ≤ 0barg t ≤ 24 hr Hysol boron nitride 1/9/13Glue Programme Discussion19
Status A low modulus adhesive is required to bond detectors An epoxy has been formulated at RAL – Low viscosity – Very low Tg (approx -40 ° C) – Very low stiffness at ambient temperature – Low tear strength above Tg (potential for reworking) Cure currently 50 ° C for 24 hrs, should be possible to reduce this temperature Detector epoxy - S Canfer, RAL 12/9/13
Compared to SE4445… This is an adhesive Lower viscosity so a thinner joint should be achievable so less material and potentially the lack of any filler is an advantage
Properties in compression by DMA STORAGE MODULUS MPa
Plans Irradiation at Birmingham syncrotron, hopefully October Make mockups of detector on CFRP and thermally cycle Try to reduce cure temperature (by catalyst choice and concentration) Further formulation development options: – Thixotropic modifiers – Silane bond promoter – Filler to increase K (but I prefer to aim for a thin joint)