Results 1 to 5 of 5

Thread: Thermal Interface Materials ... again ... sorry

  1. #1
    Anodized. Again. Konrad's Avatar
    Join Date
    Aug 2010
    Location
    Canada
    Posts
    1,060

    Default Thermal Interface Materials ... again ... sorry

    So, it's been about 5 years since I was last fascinated by TIMs (HOPG in particular), much has been learned and unlearned (and probably forgotten) since I started ye olde Gettin Greasy ... thread.

    As I understand it, the best-performing TIM available for processors is Indigo Xtreme. But:
    • Available from only one manufacturer, in limited (hyperspecific) applications and at a considerable cost premium
    • Application requires a thermal reflow procedure which places thermal stress on critical components, does not produce entirely consistent results, and is basically incompatible with large/heavy heatsinks - most PC users are actually confused, shocked, and intimidated at the procedure, which may be a contributing factor in randomly poor results
    • Removal can also be difficult, and requires unusually strong, nasty solvents

    The next best would be Coollaboratory Liquid Ultra/Pro products. Unlike Indigo, you can slather a bottle of this stuff on any part you like. But:
    • Results (as reported online) appear highly variable, suggesting multiple applications may be necessary to obtain best performance. It may or may not be compatible with all heatsink/block geometries.
    • Removal is difficult, typically requiring much chipping and carving and hours of mechanically subtractive frustrations.

    Then there's the "conventional" TIMs, the semi-metallic pastes and greases and oils. A little bottle can last you a long time and cover many assorted parts, and most of them aren't very costly. The best of these (judging by online reviews, testing, and benchmarks) appear to be Prolimatech PK-3, Gelid GC Extreme, Phobya HeGrease Extreme, and Tuniq TX-4. To be honest, many of the reviews and tests and benchmarks tend to be a bit scientifically amateur, with low reproducibility, poor controls, little or no thought to isolating (or empirically eliminating) possibly significant variables. And, of course, they are as often as not influenced by subjective bias, opinion, expectations, and perceived popularity of certains brands and products. To be fair, it's not really reasonable to expect people to comprehensively and methodically test multiple batches of TIMs on multiple platforms and multiple configurations, etc, not always entirely to my exacting engineer-like expectations, and especially not for free!

    Another new product, graphene-based Thermene, also has my attention with promising specification claims. But it is currently impossible to obtain any of the product nor even any believable comments from people who have used it.

    And to be honest, I realize that pretty much any premium-grade TIM will do the job well enough. "Mediocre" performers like venerable Arctic Silver are basically good enough for the job, I suppose. But my interest is on pushing the extreme threshold, even if it really only means a practical difference of perhaps 1-3 degrees in processor cooling. Well, perhaps I should only say "semi-extreme" since I'm talking about air, liquid, and basic phase-change cooling; I'm not talking about LN2 and other exotics. I like computers which can actually do things in the real world, they need to coexist in the same room as their operators and they need to be able to run tasks for more than the few minutes it takes to screenshot a lucky overclocking world record.

    One interesting consideration which I have not seen referenced on any of the online reviews/etc ... I want a no-maintenance no-hassle TIM which continues to perform for many months or even years. Not a TIM which performs miracles for a week or half a season and requires constant reapplication. I have some machines in which I constantly tinker and swap parts - but I also have machines which I expect to dust out no more than once or twice a year and otherwise leave alone.

    Some interesting scholarly article stuff: US Patent 8081468 B2 (Memory modules including compliant multilayered thermally-conductive interface assemblies) and Graphite sheets and graphite gap pads used as thermal interface materials. I am impressed by graphite, graphene sheets, and - as always - HOPG. Incidentally, I'm also currently interested in thermal pads (of whatever sort) to use in RAM heatspreaders, motherboard heatsinks, and similar stuff.
    My mind says Technic, but my body says Duplo.

  2. #2
    Undead Pirate d_stilgar's Avatar
    Join Date
    Mar 2005
    Location
    Philadelphia, PA
    Posts
    2,987

    Default Re: Thermal Interface Materials ... again ... sorry

    I've used IC-Diamond in the past with generally good results. Their big claim is that the life span of IC-Diamond is much higher than other high performance thermal interface materials, which have to be replaced every six months to two years.

    However, in my experience with it I found that it too would need to be re-applied every two years or so. Other factors may have contributed to system instability, but I think the drying, cracking, and breaking down of the TIM was the biggest factor in overheating and instability in my systems.

    Still, I would really like to have some long term control experiments on some of these materials since it obviously makes a huge difference. There's 8hr burn in, and then there's the 6 month torture test. I'd rather see the latter.

  3. #3
    Anodized. Again. Konrad's Avatar
    Join Date
    Aug 2010
    Location
    Canada
    Posts
    1,060

    Default Re: Thermal Interface Materials ... again ... sorry

    I've read good things about Diamond, but it seems slightly impossible to obtain in Canada so I sort of forgot about it, lol. Given a choice between two unobtainable glowing products, I think I'd rather gamble on the Thermene.

    Any thoughts about thermal pads (or whatever) for mounting heatsinks onto bare PCBs, DIMMs, and such? The Graphite sheets etc paper (second link in my first post) seems to demonstrate (in very few but very rigidly controlled tests) that two thin layers of typical thermal polymer/foam sandwiching a single very thin layer of graphite sheet can produce surprising gains in TIM efficiency.

    Agreed, I'm also very interested in TIM comparisons over periods of months or years. But I can't expect people to perform controlled experiments involving so many products and so many variables over such long timespans. Most of these TIM benchmarkers just clean off whatever still-perfectly-good premium TIM they reviewed last week to test out some new one this week.

    It would be interesting to see any actual analysis of the chemistry behind the products, across the board. And maybe some experiments about whether applying the TIM at certain ambient gas pressures or oxygen partial-pressures (or whatever) has an effect. I imagine many TIMs may oxidize and flux, not a lot but enough to degrade their performance a little, perhaps only exposing them to reducing atmospheres could prolong their efficiency. Perhaps their specific chemistry (along with thermal conductivity properties) changes at different temperatures? The list of obscure yet controllable variables goes on and on.

    Just kinda trolling for advice, recommendations, corrections, insights, etc. I don't mind spending an extra twenty bucks to get the "best" TIMs available for my next high-end builds.
    My mind says Technic, but my body says Duplo.

  4. #4
    Undead Pirate d_stilgar's Avatar
    Join Date
    Mar 2005
    Location
    Philadelphia, PA
    Posts
    2,987

    Default Re: Thermal Interface Materials ... again ... sorry

    I can send you some IC-Diamond if you can't get it in Canada.

    I'd like to do a controlled experiment with a layer of some kind of acrylic, glass, plexi, whatever (but transparent) so you could see the degradation over time of different TIMs. You could heat them all with a series of peltiers, which could control for temperature, etc. and would simulate a constant load over a long time. Energy performance is obviously one serious drawback since each TIM you test would take ~100W or so for the entirety of the testing procedure. Multiply that by 10 TIMs and you're drawing a kilowatt. Still . . . a kickstarter campaign could be used to fund the whole thing and it really wouldn't cost too much to do.

  5. #5
    Anodized. Again. Konrad's Avatar
    Join Date
    Aug 2010
    Location
    Canada
    Posts
    1,060

    Default Re: Thermal Interface Materials ... again ... sorry

    In a way, an internet populated by enthusiasts and overclockers already provides that sort of data. The hard part, as usual, is determining which of the data are most valid and which should be weighted lightly due to poor controls and experimenters' bias.

    I imagine that no matter how comprehensive your testing methodology may be, there will always be hordes of argumentive people out there with little better to do than naysay any conclusions you've drawn because your results differ from mainstream opinions. A bit of a negative outlook, I suppose, but the notion sort of takes the fun out of it. If, for example, Corsair marketed a new TIM, especially if they charged fifty bucks a bottle for it, then half the world would automatically assume it's just the very best of the best and happily jump on board. Quite sad, really - even moreso since that's exactly what seems to already carry a lot of the "premium" TIM brands on the market.

    The Prolimatech product is easy enough for me to obtain locally, or I could wait for Diamond/Thermene to become locally available, all of them seem to have "proven" themselves well enough for my needs. Again, a no-fuss minimum hassle TIM is one of my primary criteria for which product is "best".
    My mind says Technic, but my body says Duplo.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •