markn2wae said:
Do NOT use any SILVER based compound (Arctic Silver for instance) for audio use or any other "non computer" device cooling that has power applied to it's case or housing (computer chips have "grounded" surfaces with NO Voltage present on the case). The first application MAY show no leakage, but over time the boundary between grounded heat sink and "hot" device case began breaking down via small sparks jumping across the small gap, perhaps the "solvent" evaporates causing the silver particles to "get closer together" and become conductive?
Yeah, I've read that but nobody has any actual data, just repeating what they've read. Here's what I previously posted about that, edited together.
The issue about Arctic Silver possibly destroying transistors sadly is, as best I can ascertain,
an audio urban legend. The first claim about it blowing up transistors appears somewhere on AK. I haven't found any discussion elsewhere that is credible.
Some years ago I contacted Arctic about using its thermal compound for transistors. I spoke with one of the senior chemists at Arctic who told me that the company doesn't publish specifications for non-computer use for two simple reasons:
(1) Arctic it is not in the business of thermal compound for non-computer use (transistor, power MOSFET, etc.) business and
did not perform any testing. So, being agnostic as to performance, the company was unwilling to make claims about suitability for non-tested purposes. I am not surprised by the lack of interest in the non-computer business, as Arctic sells a premium product which is far more expensive than the mass-market thermal compounds used for transistor and MOSFET purposes. The company simply could not compete in the commodity market, and the boutique CPU market is far, far more profitable. I know I don't care how much a tube costs because I use so little of it.
As an aside, the rail voltage for transistors in transistor amplifiers is typically ≈ 30 to 40 VDC which is not much different from 5 VDC in terms of potential dielectric breakdown. The polymer coating on the silver particles should be more than resistant to this level. The breakdown voltage for polymer thermal compounds with metallic conductors typically is greater than 200 V / mm, often significantly greater. So breakdown is simply impossible at the sorts of voltages in an audio amplifier.
(2)
The particle size is formulated to mate from CPUs with relatively smooth ceramic carriers and covers to relatively smooth heatsinks or water-cooler blocks. The nature of the paste is such that the particle size must be carefully chosen such that a very thin layer is sufficient to mate the surfaces as the compound is not as thermally conductive as a metal-to-metal mating.
Older transistors and heatsinks are generally far rougher surfaces than modern ceramic CPUs, and the mating between the two will be poor because the particle size for Arctic Silver is too small, so the cooling may be inferior to other encapsulated metals specifically designed for transistors and heatsinks
.
Back in the day we used to lap heatsinks and transistors, and then heatsinks and CPUs. While lapping the bottom of a transistor to a heatsink is impossible, the transistor bottom may be polished to mirror bright (very smooth) using a flat piece of metal to ensure the fine emery paper is evenly applied. A similar process can turn the heatsink flat. The flatter and smoother the surfaces the better the mating and the less thermal compound will be required for a good meshing. (BTW: I suggest not asking Perkin Elmer to perform your polishing.)
Arctic Silver (the manufacturer, to differentiate from the product) specifically asserted that the Arctic Silver product is non-conductive and does not over-time become conductive. I have for years used it with high-performance CPUs which run very, very hot. I've been using that product for a decade and never had any problems.
Finally, a handwavy explanation of another possible effect, just in case someone wants to raise it. A possible issue for encapsulated metal products could arise out of capacitance. Whenever two conductors (silver particles) are separated by an insulator (polymer) a capacitor is formed. A thick layer of encapsulated-metal thermal compound might add enough capacitance that, via the Miller Effect in a high-gain circuit, rolloff might occur and frequency response could suffer. Remember, the CPU is not amplifying, only switching, so the Miller Effect does not apply. BUT the effect of a few pF would require high gain to be significant, and even then would not especially matter for audio frequencies. I note that zinc thermal compound would have similar products. The question becomes a difference between silver and (non-conductive, but still a metal) zinc-oxide in terms of conductivity. I doubt it matters.
For further reading: www.arcticsilver.com/arctic_silver_thermal_adhesive.htm
No credible source has ever supplied data showing it is conductive. I have cleaned up CPUs with dollops by over-application (not my mess) and couldn't remove all of the glop. CPU perfectly worked both with globs of it all over the place and after most of it removed by me with a solvent. The product simply isn't conductive. So no way to arc over and destroy a transistor.
