Home » Questions » Computers [ Ask a new question ]

Is it OK to use toothpaste instead of thermal paste when fitting a CPU?

Is it OK to use toothpaste instead of thermal paste when fitting a CPU?

I was told many years ago to do this by someone who at the time knew more than I did. The CPU was a celeron in the Pentium 2 era. It ran cooler with the toothpaste between the chip and the heatsink than what it did with nothing between.

Asked by: Guest | Views: 231
Total answers/comments: 5
Guest [Entry]

"This is the standard ""saran-wrap-in-place-of-condom"" question. While some toothpastes may provide the correct type of thermal conductivity, ""toothpaste"" is too big a category to answer the question accurately. Worse, unless you want to make it a fun science project, nobody is going to be testing different types of toothpastes for thermal conductivity.

That said, the answer is probably ""yes."" Toothpaste is definitely better than nothing, because air (i.e., nothing) is a terrible heat conductor. That said, there are other properties to consider. From Wikipedia

How well it fills the gaps and conforms to the component's uneven surfaces and the heat sink
How well it adheres to those surfaces
How well it maintains its consistency over the required temperature range
How well it resists drying out or flaking over time
How well it insulates electrically
Whether it degrades with oxidation or breaks down over time

I think that toothpaste might work for the thermal part, but you might have other problems in the short- or long-run.

On A Different Note: That said, if you need to stick a note to the wall, you can definitely use chewing gum (after chewing a bit). Sometimes it's hard to get off when your lease is over, though :)"
Guest [Entry]

"It might be better than nothing, but unless toothpaste has unsuspected thermal conductivity, I'd say it's a bad idea.

You also have to consider what the ingredients of the toothpaste might do in contact with your CPU. I suspect it would be pretty conductive in an electrical sense. You don't want electrical conductivity.

Better get some real thermal paste. It's more expensive than toothpaste, but worth it."
Guest [Entry]

"I've been using a rebuild with toothpaste on my Thinkpad T42 for about 3 years, with no problems. These laptops, however, are known for running cooler than most thinkpads, to begin with, and we don't play very many 3-D games on those ATI 9600's, so the load on the toothpaste is not very high. If you are worried about the paste drying out, just use A LOT. The only part that will dry out is the stuff on the edges of the CPU, and if you are generous with the paste, there should continue to be a large amount in the middle of the CPU that is thoroughly wet ... or use 25% vaseline and 75% paste... or 25% grease and 75% paste. Grease is basically oil & soap, to keep the oil from drying out.

Remember, toothpaste drops the temperature by 20 degrees, even after 12-hours of burn-in, and arctic silver 5 drops it by 29 degrees at most. So toothpaste gets you 69% to cooling nirvana, even after the dry-out period."
Guest [Entry]

"You acutally would be surprised.

In a pinch, yes, it is OK but a pinch is basically 12 hours.

If you want some other common goods, try this:

http://forums.overclockersclub.com/topic/164465-you-got-your-peanutbutter-on-my-cpu-you-got-your-cpu-in-my-peanut/

http://forums.overclockersclub.com/topic/164465-you-got-your-peanutbutter-on-my-cpu-you-got-your-cpu-in-my-peanut/?p=1680970

All of these ideas will help you in a pinch but ALWAYS get proper thermal paste."
Guest [Entry]

"You (or anybody else in kind of emergency) can just use regular (comestible) oil, or mineral oil which will be better if avail. for several weeks will work perfectly. Howerver, it will be mandatory to replace the oil with thermal paste after a while, because oil will polymerize and dry up in time.

Long time ago (in 198x) i did this for heatsinking high power audio transistors in a 2x300W amplifier, because thermal paste was not easily available at those times in Romania. The oil runs well for years, but the surface was quite large and thermal requirements were not so high as in modern processors (some 3 cmp per power transistor case versus less than 1 cmp for processor)."