Mineral Oil as Thermal Grease

I came across a recent measurement that shows the cpu temperature and motherboard temperature using mineral oil. Room temperature was around 25C.

This is on Ubuntu 10.04. The motherboard was one of the ones that died recently. For some reason, my current motherboard won't make the measurement. I have another due next week that should.

Nov 24, 2020 command | tee output.txt

The standard output stream will be copied to the file, it will still be visible in the terminal. If the file already exists, it gets overwritten.

cpu temperature: watch -n 2 sensors | tee mytemp.txt

cat mytemp.txt

atk0110-acpi-0 Adapter: ACPI interface Vcore Voltage: +1.10 V (min = +1.45 V, max = +1.75 V) +3.3 Voltage: +3.31 V (min = +3.00 V, max = +3.60 V) +5.0 Voltage: +4.87 V (min = +4.50 V, max = +5.50 V) +12.0 Voltage: +12.16 V (min = +11.20 V, max = +13.20 V) CPU FAN Speed: 3110 RPM (min = 0 RPM) CHASSIS FAN Speed: 0 RPM (min = 0 RPM) POWER FAN Speed: 0 RPM (min = 0 RPM) CPU Temperature: +25.0C (high = +90.0C, crit = +125.0C) MB Temperature: +34.0C (high = +30.0C, crit = +90.0C)

Mineral oil works. It stays put and doesn't dry out.

No, oil won't dry out. But it may leak out leaving less oil and more air than you expect. What is your CPU orientation? How long have you worked with it so far?

What is the base of the thermal compounds that dried out for you?

Years ago I was into building my own PCs. Some tried the lapping thing, but it never worked as well as any grease. You just can't get the grit fine enough.

They make some heat sinks by adding fins to a base using high pressure to make the metal flow, even with different metals such as copper fins with an aluminum base. That gets good contact. I don't know of other ways to get good thermal contact between separate metal surfaces without grease.

Don't stop turning! Someone tested the dry stuff you removed, and it passed (but aged poorly). The fact that 'mineral oil' (that's a large category of materials) seems to work, is not a complete study, yet.

Thermal compounds include pastes, wax-like sheets, rubber-like sheets, and some exotics (fluorinert, silicone oil baths, transformer oils). There are no reasons to expect that 'dry' stuff you removed is similar to any particular compound in use today; it's not clear what you'd want to avoid.

USP (food-grade) mineral oil should be inert, and better than air at conduction, but is NOT a safe solution for long-term heatsinking. It wets surfaces, and slowly spreads out, so keeping it in place means... making a wick-like structure in the region of interest. For instance, a bunch of powder (large surface area) stirred in. That's why the old solution worked for years; possibly was still working, but while cake-dry wouldn't reflow to make a new thermal connection reliably.

Use a good-quality commercial thermal paste, or sheet, after removing all traces of the old stuff.

I don't think you understand. There is no problem keeping the mineral oil in place. I examined the surfaces carefully when I separated the heat sink from the cpu, and it was obvious they were wet with mineral oil. There was no evidence it had spread out past the mating surfaces and wet the surrounding area. I checked very carefully for this.

It is also obvious from measuring the cpu temperature that the mineral oil was working. Even at idle, the cpu temperature was about the same as room temperature, where the motherboard was considerably warmer.

When the new motherboard and cpu arrive, I will do some full-power measurements and post the results.

After these examples, I fully expect the mineral oil to provide the same performance over a long period of time.

Based on zero long term experience. How long had the previous heat sink grease been in place before it failed? Why do you think a paste would dry up and your liquid won't run out?

If mineral oil is so great, why would it not be discussed in the many, many computer freak venues? Actually, I expect it has been discussed and found to be lacking.

The classic Dow 340 filled silicone doesn't seem to dry out.

Mineral oil is fine if the sufaces are pretty flat.

Oh, it's not immediate leakage, but slow (the oil wet spot gets bigger) that is the problem. Years passed and the original compound seemed dry (certainly didn't reflow well enough to make a new thermal bond), but that doesn't mean it was thermally defective, just that it was non-fluid.

For how many years?

I finally figured out why the mineral oil won't spread. Capillary action keeps it stuck between the surfaces.

Anyway, JL just blessed it, so it's fine.

At least several. As I mentioned in a related post, capillary action keeps it in place.

Capillary action where you want it isn't the whole story. Entropy says it doesn't have to stay stuck, but can wick through microcracks at a very slow pace... or diffuse into some materials, or evaporate, or even get nibbled by microbes.

If the oil every leaves, it won't be pulled back by capillary action.

Not too many cracks in the cpu or heatsink.

Not too worried about it ever leaving. Capillary action keeps it stuck in place.

No worries about evaporation. The spacing is very thin so very little is exposed. Anyway, mineral oil doesn't evaporate.

Quote:

Mineral oil itself doesn't evaporate. ... This means there is negligible or no evaporation of the material.

You keep talking about how wonderful this is, but you have not tested it to see how long it lasts and you have not even given thought to why others aren't using it.

I seem to recall someone testing butter and found it to be a good material for this, but it only lasted a few weeks.

We'll see how it goes. If the oil dissipates and the CPU overheats, you will let us know, right?

Hmmm... I did a quick search and this topic was covered here just last year. Mr Allison reported an amp using TO3 devices greased by error with bearing grease failed after a year when the grease oozed out of the joint. So I'm not expecting oil to last any longer.

Any dust that settles near it provides an alternate capillary path, which grows more rapidly as it enlarges. It may be that your test cases were in a very clean environment, but that won't be the case for everyone.

In mechanical clocks, the axis bushings get oiled, but gear teeth never. Dust collects quickly and jams the mechanism.

CH

Every oil lamp indicates a problem with that theory, as does every bearing that ever ran dry. Zinc is disallowed in high vacuum because it evaporates too fast. All the problems that go with loss of material, are EXACERBATED when it's a small 'very thin' space being filled.

Thermal compounds have granular fillers, to make lots of micro-spaces from the mini-space being filled, and retain the oil better. You reported that after some years, the oil was gone from your example anyhow. Where'd it go?

What do you mean "applying a uniform coat"? There doesn't need to be a uniform coat on the whole CPU enclosure - you don't spread the thermal grease on it like cake batter. Put a pea sized drop in the center and put the sink down on it and let it spread out on its own.

Cake-battering it will leave gaps and give poor thermal transfer and the grease will dry out, yeah.

The heat produced within a CPU enclosure isn't uniform anyway so why there needs to be a uniform layer of thermal interface material across the whole surface of the enclosure IDK. the path from the center of the die to the case, out to the edge of the case, through the interface to the sink is a relatively high-resistance path it contributes little.

A smaller amount of grease in the center will transfer heat to the sink nearly as effectively and with less interface area there's less opportunity for any air gaps to form in the first place.

Less...is more...

Don't need to apply a "uniform coating" anyway that's not how you apply thermal grease to a CPU. If you slather it on like icing a cake you get gaps and the s*it dries out, yeah.

It's not like icing a cake or greasing a baking pan! Was OP a f****ng chef?

Oil lamps burn kerosine, which does evaporate. Anyone who tried to use WD-

Mineral oil does not burn, and cannot be used in oil lamps. The vapor pressure is <0.5 mmHg, which is fairly low:

Another issue is the amout of mineral oil being exposed. The separation between the heat sink and the cpu is very small. Only the edges are exposed, which is the only place evaporation can take place.

From my examination, the rate of evaporation of the mineral oil is negligible over a period of years.

Bearings do not evaporate oil. The oil is forced out of the bearing, which will run dry if the oil is not replaced. This is called lubrication, andis very important in engines. We are not talking about engines.

We are not talking about zinc, and we are not talking about a vacuum.

I do not see any of the problems you speak of.

Conventional thermal compound dries out and must be replaced periodically.

We do not know what kind of liquid they use. Also, the filler means the mating surfaces are held further apart, which reduces the thermal transfer.

This problem does not occur with mineral oil since there is no filler.

I see no evidence of loss of mineral oil from the heat sink over seveal years of use.