
We have received several hundred emails from readers hailing from 10 countries in response to Strip Out The Fans, Add 8 Gallons of Cooking Oil - and the enthusiasm shows no sign of abating anytime soon. Thanks to the huge amounts of reader feedback, we have been able to collect a great deal of invaluable information.

Upgrading the oil-PC requires completely draining the coolant, which proved very complicated in our case.
We'd like to take this opportunity to present some of the most intriguing reader responses and add our own comments as well.
g'day Frank,
WARNING
Vegetable oils slowly polymerize (turn into a solid, rubbery mass) upon exposure to oxygen. This process is accelerated by higher temperatures.
Using vegetable oils to cool hotspots like CPUs and GPUs may lead to a slow buildup of a solid polymer layer on the electronic component. This will impede heat dissipation, and lead to overheating. The most likely outcome is component failure rather than a catastrophic failure, such as fire.
I suggest that you append a warning to your highly entertaining article, and advise people to use good quality motor oil instead -which is designed for sustained high-temperature operation.
cheers, Mat Ballard
And Another Thing About Vegetable Oil
Sorry to spoil your meal, but cooking oil will go rancid, i.e. grow BACTERIA by the gazillions and get someone sick. Your trick is sort of a laugh for someone who is bored, but you've just ruined your PC permanently, and even if you try to get the cooking oil out of your printed circuits, (they are porous weaves of fiberglass and epoxy) they too will rot. You'll see soon enough - you'll have to throw out your experiment just because of the stink, never mind your health in the longer term. Not to mention that silicon won't seal oil for long, it will seep out the "seals". The permanent way to do this is to use a higher temperature inorganic oil - like a silicon-based lubricant, but it still will be hell on any elastomer seals, but at least won't rot...At Cray they used to use fluorinert liquids, but these are not permitted anymore due to halocarbon destruction of the atmospheric OZONE layer from the vapors ( plus you need a sealed system ) BTW silicon based oils are Hell to remove from anything they touch - you thought the cooking oils was messy, silicon oils are almost non-removable.
BTW You must have been VERY bored.
In your article concerning the vegetable oils you wrote:
"Our initial attempts had shown that the poured oil sporadically led to crashes. The cause was also quickly found: The processor base together with the CPU and the heat sink had to be made impermeable to the liquid. Quite a bit of labor and time are necessary for this, since just like for the case, we first had to use special glue and then silicon. After successful sealing, the system works without a hitch."
We have the following explanation for this phenomenon: On the motherboard in the area of the CPU base, the oil is responsible for increasing the capacitive resistance between the individual wiring. In short, the oil acts as a dielectric material. Since very high frequencies occur on the motherboard, the capacitive resistance goes down. Accordingly, this then influences (or tampers with) the digital signals, particularly in the area of the CPU base. After all, 939 pins are located there in a very tight space. I believe you're on the right track, as proven by your experiments.
I'd like to offer this additional information: Vegetable oil is primarily a triglyceride. What this means is it's primarily a big molecule that consists of a glycerine [compound] holding onto three long-chain ester groups. This makes for a nice non-conductive medium that's fairly stable and readily available. However, at high temperatures this triglyceride will lose one, or more of its ester groups, which are then referred to as free fatty acids. These free fatty acids have vastly different dielectric properties, and (I'm assuming) a higher conductivity. I postulate that the higher temperature immediately surrounding the processor, and thus the pins, is increasing the concentration of free fatty acids, until a suitable conductivity exists to afford an errant discharge between pins. While the immediate fix is to seal off the processor/pin sockets/pins from the oil, I can't help but think that over time, the free fatty acid concentration may increase throughout the system, thus lowering pH, and increasing conductivity. This may give rise to the necessity for an oil change, or buffering of the acids with a material that would not increase the dielectric properties of the medium. I don't know how much time this lowering of the pH by vice of the free fatty acids (FFA) would take, but a calculation of the watt-density of the hottest parts acting on the volume of oil would be able to estimate the FFA concentration increase, and thus the pH decrease. Then again, this increase may take a year - or it may take a millennium - so it might be a moot point for the system as a whole.
Ed's Experiences With Liquid-filled Electronics
Nice work Frank.
Just a few comments from someone that has worked on a lot of liquid filled electronics. Water would work, but:
1. As you learned, you must protect all high frequency areas
2. You need continuous demineralization treatment. You need to keep the resistivity of the water high.
Better than vegetable oil would be white mineral oil (baby oil without the smell). Vegetable oil and auto oil will both attack many plastics and rubbers over time. There are some exotic fluids, but mineral oil or a synthetic with no lubricant additives is the best. And they are very clear.
Good Luck, Ed Blessman
Sir:
I read your article about the oil-filled case, and I found it interesting. I was thinking of doing the same thing, but instead of vegetable oil, using either Exxon's Coolanol or Polyalpha Olefin, which are actually oils designed to have electronics, even high frequency electronics, submerged in them. The most common usage for these oils is high-powered military attack radars, when lots of cooling is needed, and protection from the oil is needed as well. I am commenting on this, because you suggested motor oil, and am offering these oils as something else for you to use as a suggestion.
Jonathan Jones
Joel's Waterless Water
Thought you might be interested to know. Concerning your article "Strip Out The Fans, Add 8 Gallons of Cooking Oil," you guys tried oil and distilled water. What would be really cool is if you used what's called "waterless water". It's a fire suppression agent that looks like water only it doesn't conduct electricity. There have been a few impressive displays of its nonconductive properties whereby they immerse a monitor and laptop in a bath of this stuff without any problems.
Joel
Chris's Suggestions For Holding It All Together
Hi Frank,
You use special glue in your oil PC article; I would suggest a couple of different ones. I have used Loctite UV curing anaerobic glue in glass work; this glue is expensive ($40 Australian a bottle) but is used sparingly (use less than superglue even!). The glue is cured by exposure to UV light in the absence of air. This means only the contact surface is glued and excess can be removed with solvent. The curing is almost instantaneous (two-three seconds exposure) and is extremely strong. The glue is perfectly clear-another reason I used it in glass work.
Another option would be fusion bonding with Weldon #3 (methylene chloride, trichloroethylene and a methyl methacrylate monomer mixture- nasty!). This is a very toxic solvent and it evaporates at a fantastic rate (its caused ice frost when I've used it before). The draw back is the whitening it can cause; all parts have to be kept in an air-tight box in a warm place with a dehydrator pack to suck all the moisture out of the plastic. This then stops the white frosting that is common with super gluing and solvent bonding. I find the solvent good for repairing cracks as it wicks into the tightest places.
Hope this comes in handy sometime!
Chris Knowling
Hi,
I just read your article on the silent oil-filled PC (coolest idea ever, BTW!) and was just wondering if you ever planned on testing out the overclocking capabilities of such a cooling solution? I'd be interested in seeing how much harder you could push a CPU/GPU over an air-cooled or other cooling solutions? Maybe some kind of overclocking showdown with various coolers to see what performs best, or which is the best bang-for-the-buck?
I was also wondering why you recommend motor oil? And why didn't you use it in your test? Given the price of the electronics involved, motor oil wouldn't be a huge cost increase...
Thanks for a superbly interesting article! :)
Dan
Reader Response From Mr. Rau, Germany: Using De-ionized Water As A Coolant
Dear THG-Team!
I just finished reading your article on cooling a PC by submerging the components in oil and would like to comment. You wrote that you used water in your original trials. Did you use distilled water or de-ionized water? I ask because there is a huge difference between the two. Compared to de-ionized water, distilled water looks like slurry.
In a very complex process, the conductance of de-ionized water is reduced to a bare minimum (µS = micro-Siemens) using deionization filters. (The filter material looks a lot like those little hydroponic fertilizer pellets.) It is so sensitive that it has to be transported and stored in containers that can effectively protect it from any light.
This type of water is used to cool krypton lamps in lasers, since it is of the utmost importance that none of the substances found in the coolant accumulate on the lamps (4kW per lamp, up to four lamps per laser). These high-powered lamps require an initial ignition voltage of up to 30,000 Volts and then continue to "burn" at 200 Volts. Since the electrodes of the krypton lamps are submerged in water during this process, the only choice here is the aforementioned de-ionized variety. However, this type of water is highly aggressive and requires constant monitoring and filtering to clear out even the most minuscule particles (down to 2µm) and thereby ensure the required purity. Also, the aggressive nature of de-ionized water puts certain strict limitations on the materials that can be used in such a system. For example, any steel components such as screws, connections or casings need to be made of stainless steel. Glass and almost any type of plastic can be used as well. All other metallic materials (steel, copper, brass, aluminum, nickel, etc.) will suffer extreme corrosion almost immediately as the water mercilessly tries to replenish its missing substances (i.e. ions) from them, if present. That would mean that de-ionized water would make short work of any circuit paths, soldering terminals and even the pins on the processor or its socket as well as any conductive materials if used as a coolant in a PC. Therefore, the entire cooling circuit would need to be completely free of such materials.
I'm still trying to figure out what could have caused the short-circuit you described. Even if the conductance of the water had not been low enough or the aggressive water had drawn ions from the PCB's exposed (brass-covered copper) areas very quickly, the voltages used in a normal PC are not sufficient to cause a short-circuit, in my opinion.
Finally, long-term usage would also require the cooling system to be made impervious to light to prevent algae formation. Since the coolant in such a system would require a very high degree of de-ionization to ensure low conductance, adding anti-algal agents would not be possible.
Incidentally, water is a much better coolant than any type of oil, since it can dissipate more heat per unit of volume than oil. On the other hand, oil-cooled components will never rust, although the thermal load they can bear is also lower if the radiating surface is the same. In either case, a good circulation of the coolant should be ensured. I hope that I have been able to make at least a small contribution.
Eberhard Rau, Erdmannshausen, Germany

Can't submerge these in the oil bath: All hard disks are sealed in, and equipped with air vents to permit access to the external atmosphere
Andreas Mösli: Why Not Put Disk Drives And PSU Inside The Oil Bath, Too?
Dear Mr. Völkel,
My congratulations on your article on the ultimate oil-cooled PC. I couldn't help but notice that not all components were submerged in oil in your build. Did you make any attempts to completely submerge the hard disk and power supply in oil as well? My idea was to build an all-aluminum enclosure with cooling fins sealed from within, and further sealed on the outside with elastic caulk, so that I could submerge all system components in a special type of mineral oil called silicon oil. This way, you can finally build an absolutely silent PC, which nevertheless delivers top-notch performance! By comparison with conventional oil, silicon oil is more expensive, but also confers absolute certainly that no current flow can occur. This type of oil has long been used as a coolant in large-scale computers, if not in such a rudimentary fashion.
In general, I'd also like to give THG a big thank you for all the various topics and articles you bring to my monitor every day. I find myself dropping in several times a day, just to see what's up!
With best wishes,
Andreas Mösli (Richterswil, Switzerland)
Most honored Mr. Völkel:
Your attempt to cool a PC with oil was exemplary. But whenever you try water - even distilled, de-ionized water - your attempts are bound to fail. As Newton put it, "for every reaction there is an equal and opposite reaction:" You can put oxygen and hydrogen together, whether in an uncontrolled reaction or in an action controlled through a fuel cell. In both cases, water will be produced. The compatible opposite reaction is the electrolysis of water. Put an anode and a cathode in water, apply voltage, and off it goes. In any case, water is an ampholyte, which means it possesses the characteristics of an acid and a base, and is capable of reacting chemically as either an acid or a base. In small, both thoroughly measurable regions it reacts to itself in the following way:
H20 (water) + H20 (water) < - > H30 (trihydrogen monoxide) + OH- (a hydroxyl ion). This reaction cannot be stopped because both sides of the reaction share the same energy state. In addition, the relationship between trihydrogen monoxide and hydroxyl ions gives water its ph-value. If that relationship is 1:1, the fluid is neutral in pH, but if there is more of one than the other, the liquid will either be an acid (more H30) or a base (more OH-). In both cases, that fluid is conductive.
Sincerely,
Stephan Wilden

Lots of detail work is required before starting a project like this one. Just the selection of individual components takes time to decide, even when everything is already at hand.
My dear Mr. Völkel,
I read your article about the oil-cooled PC with great interest, but I have an improvement to suggest as regards the coolant you chose. You stated that water's clarity and transparency helped it make a better visual impression inside the case. My hair stood on end, however, when I saw you using deep-frying oil and then you also recommended motor oil.
There are two alternatives you might want to consider:
1. Neutral oil is a colorless oil that's half-synthetic in manufacture and that has a lower fraction of free fatty acids in composition. It's colorless and somewhat thinner in viscosity than cooking oil. You can purchase this type of oil in small quantities at any good drug store. Naturally, it's not as cheap as cooking oil - in part because it has valuable pharmaceutical characteristics - but I don't have exact prices at my fingertips.
2. Paraffin comes in various forms and is likewise colorless. It's a mix of hydrocarbons and petroleum distillates and contains no free fatty acids, which could cause corrosion - it's used to create a protective film on stainless steel in the pharmaceutical industry. Paraffin also comes in different viscosities: the thinnest varieties differ only slightly from water, while others are solid at room temperature.
These items are probably more costly than cooking oil, but they're almost certainly less expensive than most types of motor oil.
With friendly greetings,
Kai Hückstädt
Sven Reischauer: Cover The Board With Clear Coat
Hello THG-Team,
Great article on the oil-cooled PC. Because using cooking oil as a coolant also makes rancidity inevitable, I couldn't help but ask myself if using some kind of thorough clear coating on the circuit boards wouldn't help avert current flow in distilled water. This is best accomplished in a clear lacquer bath that's stimulated with current at 50 Hz, so as to cause the lacquer to penetrate every pore on the surfaces it coats. Remark: similar technology is used when pouring cement in the building trades. It has to be this way to counter the potential dangers of algae bloom that can otherwise occur. If I had time to try this out, I'd do it myself with an old circuit board. Do you have any interest in providing your readers with a sequel to your "PC water bath" efforts?
Sven Reischauer, Tübingen (Baden-Württemberg, Germany)
Hello!
I followed your oil-cooled PC article with tremendous interest and suspense. I've also been playing around with the idea of conducting this experiment myself. But vegetable oil really doesn't have an appealing color. That's what led me to think of silicon oil, which is crystal clear and is also available in many different viscosities and for many temperature ranges. Next, I dug into some of the related forums on this subject online. The results might best be summed up as "Silicon oil is too aggressive, will corrode silicon and acrylic components, and is also too expensive." I wasn't quite ready to give up on this option, and went ahead and researched the cost of silicon oil. In my opinion, the cost is still in the realm of what's tolerable. What do you think of this idea?
Friendly greetings from:
Jörn Achtelik
Martin Göbel Writes: Plexiglas Is Unsuitable
Dear Mr. Völkel,
I read your article with great interest and attention. As someone who works with plastics by trade, I wanted to provide you with two important bits of advice about what I read therein:
1. Plexiglas is very sensitive to contact with chemicals. Also, the brand of silicon sealant you chose isn't compatible with Plexiglas. In the short or long run, your case could fall prey to stress fractures. Under some circumstances this could permit the fluid in the case to escape, and might even negatively impact the stability of the case itself. I'd recommend that you substitute a pH-neutral, acid-free silicon sealant instead.
2. In one of the photos in your story, the image shows you working with a material named "Hobbyglas". This material is not Plexiglas; rather, it's a type of polystyrol (PS). This material is even less chemically resistant than Plexiglas, and should therefore be kept away from any contact with oil, silicon or glues of any kind.
Because your article is so likely to provide impetus for others to follow in your footsteps, I urge you to consider this advice very carefully. Should you have any questions or concerns about what I've said, I'll be happy to respond at length.
Best wishes from
Martin Göbel, Bochum (Nordrhein-Westfalen, Germany)
Good morning:
I find your oil-cooled PC project fascinating, but I have a couple of ideas and questions I'd like to share on this subject. A nice gimmick would be some kind of long-chain fatty acid, because it would be solid at room temperature, making transport of the PC much simpler and easier.
I would recommend stearin (C16) or palmitic acid (C18). These are saturated fats that are very stable, and that give off no (or only a very few) H+ ions. If these substances melt too slowly, however, it might make sense to use a C14 or C12 fatty acid instead. To protect the Plexiglas, you could also coat it with an emulsifier (by painting it or using a film of some kind) to repel the fatty acid itself. With a chemist at your disposal, you should find it no trouble to brew a fatty acid that would be liquid at 84° F (30° C).
At the end of your article you recommend motor oil. Should you use completely- or partly-synthetic oil? I would be sure to check to see if I can also build something like this for myself. It would certainly be the prime attention-grabber at any LAN party, if you were to show up with a PC deep-fryer. If I come up with any new ideas, I'll be sure to get back in touch!
Best wishes,
Urs Rudolph
E-mail From Mr. Wimmer: A Special Coolant
Hello Mr. Völkel,
I found your article about the oil-cooled PC very interesting. I had already thought of immersing a complete PC system within a fluid of some kind myself, primarily as a way to stymie noise output. That was a thought that had long since vanished from my musings, however.
I have a couple of thoughts as they relate to your test. Sadly, the choice of oil as a coolant is not the best option. I have worked for a while now on the development of temperature measurement systems for semi-conductor production. This permits the wafers to be cooled to temperatures as low as -76° F (-60° C). For many tests, this makes absolute electrical isolation of the carrier substrate from the local environment possible.
For our coolant, we use a special fluid named "Galden ZT 150". This fluid is colorless and looks better than vegetable oil in every situation. It's also completely neutral chemically. In addition this fluid possesses phenomenal volume resistivity of 2.0 * 1010 Ohms/cm. This makes sealing up the CPU socket completely unnecessary. The hard drive can also be immersed. This would also dampen drive noise, as well as provide additional cooling.
The only downside is the price: 11 lb (5 kg) of this fluid costs about $70 (I've e-mailed the company rep for North America for this material and asked for pricing information and where one can buy this fantastic sounding stuff - maybe we can drop this information into the story later when it's made available to me?). That's a lot more than you'd pay for cooking oil! But if you were to make the enclosure smaller, as the improved heat transfer characteristics of this fluid readily permit, you wouldn't necessarily need eight gallons, either.
With best wishes,
Ferdinand Wimmer
Use Silicon Oil
Too bad that you didn't make your first try with silicon oil or something similar. Then you could have let the system run at a heavy load for at least a couple of weeks. I hope you're inclined to try something along those lines in the future.
It's clear that extended use of salad oil would quickly lead to gummed-up components and a stinky, rancid mess reminiscent of rotting fish. I also believe that in any case use of silicon oil would lead to a longer system lifetime.
I'm also curious about what would happen if the oil should penetrate between the individual silicon and metal layers on the circuit boards, insofar as that is possible.
Your friend,
Josef Fischer
P.S.: You have my special respect for your attempts with distilled water!

The Oil PC under construction: We filled the case for our test machine with eight gallons of vegetable oil while a benchmark was running to finish up our silent Athlon FX-55 system! It also includes a GeForce 6800 Ultra as well as other components. Video 18 presents a completely new chapter in the annals of PC cooling for high-end systems: Here, you'll see a custom-built case with working components filled with eight gallons of oil. All this, while our benchmark keeps on running! These videos from the Munich THG labs have been viewed by a huge audience since their introduction in 2001.
As in earlier videos, we offer Video 18 in three different formats for our readers. We've also taken all possible steps to deliver optimal image quality while using a relatively modest data rate for our video stream.
To grant the wishes of various readers who didn't want to have to install another video codec, we offer this new video in WMV-9 Format for the MS Media Player as well as in Apple QuickTime (H.264) format as well.

Apple's QuickTime makes up for earlier versions and is now easier to serve up than competitive offerings. The video is encoded using the H.264 standard.
In addition, the file is compressed using a ZIP utility, so that users must download a file that's about 20 MB in size. With a playing time of 3 minutes and 45 seconds including stereo sound, this means an average data rate of 5.7 MB/min or 93 kB/s.
Our preference for the DivX Codex has also changed, because the new version 6 does a poorer job of encoding as compared to the preceding 5.2.1 version. Thus, we decided to go with the "old" version for this video. But this whole market has been transformed by new compression methods, in that the Microsoft WMV 9 codec shows equally good characteristics when compared to the DivX 5.1.2 version - and it is also much easier to obtain. Apple's QuickTime Pro 7.0.3 has also made strong efforts to catch up, and now works with a modern H.264 Codec. Handling is better and easier than for DivX or WMV 9. Anybody who downloads the Apple version of the video can quickly confirm this for themselves.
You can download the new video as follows:
Download video 18 in the DivX format
Download video 18 in the WMV9 format
Download video 18 in the Quicktime format
If the video won't play back on your system, please download and install the compatible player: Divx Player, Microsoft Media Player Apple Quicktime Player
| Technical Data: THG Oil PC Video | |
|---|---|
| Video resolution | 540 x 432 @ 25 fps (5/6 NTSC) |
| Aspect ratio | 4:3 |
| Color Depth | 16 Bit |
| Audio signal | Stereo, 16 Bit, 48 KHz |
| Audio data rate | 96 kBit/s (12 kByte/s) |
| Video data rate | 650 kBit/s (81 kByte/s) |
| Total data rate | 746 kBit/s (93 kByte/s) |
| Video compression | MPEG-4 DivX, 5.2.1 Pro Codec, 2 Pass, Bidirectional Encoding |
| Compression Audio | MPEG-1 Layer 3 (MP3), Fraunhofer |
| Color space | YUV |
| Run time | 3:45 (m:ss) |
| File size | 20 MB |