Silicon fluid immersion?

mwin

2[H]4U
Joined
Jun 24, 2004
Messages
3,355
I've been brainstorming on building a peltier/H2O system that I won't have to worry about condensation with (at all). Dielectric grease and neoprene just doesn't seem fool-proof enough. So I started searching for acceptable fluids to use for immersion. Here's what I'm thinking for requirements:

1)Has to be affordable. I don't mind spending money, but I don't want to spend hundreds of dollars a gallon (I'll need around five).
2)Has to be plastic friendly. Can't have it pulling the plastimers out of my cables and components.
3)Has to be relatively safe. No risk of explosion, terrible health side affects, things like that.
4)Has to have low vapor pressures. It'll be in a sealed system, but I'd really like to not have to worry about fumes or the stuff evaporating away through the acrylic enclosure.
5)Has to be able to cool anything on the motherboard/grapics card that needs cooling (except CPU, chipset, GPU)

I found this fluid silicon that looks like it might work. The dielectric properties look like they'll be okay, and I think it'll meet all my requirements except perhaps #5. Anybody have any insight into using this stuff? Other fluid recommendations? Links to/pictures of other immersion projects?
 
Hey...I assume you've seen my Thinktank mod , I thought I'd chime in , as I've done a LOT of research on this. ;)


Quote : " 1)Has to be affordable. I don't mind spending money, but I don't want to spend hundreds of dollars a gallon (I'll need around five) "

Ok....This eliminates any of the flurochemical dielectric fluids such as 3m Fluorinert or Galden from solvay solexis.
All of these fluids are really expensive.

The fire/explosion thing eliminates alchohols,ethers etc......

Basically , this leaves commercially used dielectric oils : mineral oils/silicon oils/synthetics........This is also where 'plastic friendly' becomes an issue.
A complete motherboard assembly has a lot of different plastics :rubber/pvc/nylon/acrylic...
Finding an oil based dielectric that won't affect any of them , is'nt going to happen.
The biggest issue is the rubber.Every electrolytic capacitor on a motherboard is sealed with a little rubber endcap.Almost any oil will dissolve rubber.These must be protected (i sealed mine with acrylic).

I ended up choosing a fluid called Midel 7131 based on 3 things :

1) The only thing the midel should affect is the pvc.....it'll leach the plasticizer out....but so what ?....the pvc bits are wires and connectors.They're never gonna move anyways so it's not an issue.My machine was submerged for 2 months....I havnt seen any degradation.

2)Midel is clear.There are other similar oils , but they have color.....

3)Midel is Biodegradeable....most silicon oils are'nt.

I hope i've given ya some food for thought....good luck with the project.

:D

ps.....after closer inspection of your link : it lists the flash point of this liquid at 39 degrees C (!).....I'd cross this one off the list.
 
BrainEater said:
Hey...I assume you've seen my Thinktank mod , I thought I'd chime in , as I've done a LOT of research on this. ;)

Actually, I'm working with a new project, and I'm going to say that ThinkTank was definately my inspiration. I can't give everything away yet, but I did find a fluid that meets everything in the above list and then some. It's a hydrocracked, hydrorefined mineral oil called Luminol TR-I. The "I" in the TR-I stands for inhibited. It's not supposed to attack any plastics or rubbers, but just to be safe, I have a system that's been "curing" (for lack of a better term) in it for about three weeks now. So far, no discoloration, no contamination, no eating of plastics (even the PVC's), and it seems to be leaving the rubber end on my capacitors alone. The stuff cost me about $3.00 a gallon, and it's been performing incredbly. I've been testing it on a cheap motherboard just to be sure.

I'm an engineer for an electric utility, and started picking the brains of our transformer cooling guys a while back. They were the ones who suggested this stuff. It's fairly new, and they've been very happy with it. Good enough to cool a 768KV transformer, good enough for 300 Watts of heat dissipation.

So far, the results have been pretty good. The only issue that I have with it is that heat does not want to travel very far within the fluid. Water likes to keep heat homogenious, but the thermal transfer rate of Luminol just does not match it. Luckily, the refinining process keeps the fluid very thin, even at lower temperatures. As such, I should be able to get away with using a submerged pump to move the fluid around. I've noticed in ThinkTank that you have three circulation pipes. What pump ar you using to move the Midel around?

I tried messing around with Silicone based fluids for a while, and kept running into two problems. First, they are rather expensive, with the cheapest I could find being about $7.00 per gallon. Second, as you mentioned before, Silicon is NOT bio-degradable, and I don't like the idea of pouring it down the sink.

I was going to keep this one under my head until all my testing was done, but for the original poster's benefit, you should look this one up. It might be the fluid you're looking for. Midel 7131 is great, don't get me wrong, but I can't justify spending the amount of money they wanted for it, especially shipping.

Oh, and Luminol is clear, as clear as water, and it reacts incredibly to black light. It puts off an amazing purple glow. So far, the ONLY downside I can see to this stuff is the smell. It most definately has a petrolium scent to it. However, my systems are running sealed, so as long as you don't run it open, you won't smell it. According to the MSDS sheets, Luminol is almost 100% non-toxic, but I still am not going to chance breathing it in. Working with Methyl was bad enough.

Matt.
 
BrainEater said:
ps.....after closer inspection of your link : it lists the flash point of this liquid at 39 degrees C (!).....I'd cross this one off the list.
Yeah, that probably is a little too low. They've got another one with a viscosity of 3 cSt that has a flash point of 100 C (212 F). I didn't know if that would be too viscus or not. There's another with a viscosity of 1.5 cSt and a flash point of 63 C (145.4 F). If the temperature ever gets that high in my room, the flash point of my computer is probably the least of my worries. Do you know how viscus that Midel is?

BrainEater said:
this leaves commercially used dielectric oils : mineral oils/silicon oils/synthetics........This is also where 'plastic friendly' becomes an issue.
I saw this on their list of properties: Chemically inert to most materials including rubbers, plastics and metals. You think this includes the plastics in the motherboard components? I don't really know anything about silicon oils, besides what little knowledge I've gathered from their site.

My plan is to build myself as small of an acrylic enclosure as I can. Cut holes in the acrylic to pass the cables through, and then use epoxy to re-seal the enclosure. I'll run distilled water in .5" ID tubing through the acrylic to waterblocks/peltiers on the CPU, GPU, and chipset. This way, I won't have to worry about the bath having to absorb and dissipate the majority of the heat.
 
enraged78 said:
Actually, I'm working with a new project, and I'm going to say that ThinkTank was definately my inspiration. I can't give everything away yet, but I did find a fluid that meets everything in the above list and then some. It's a hydrocracked, hydrorefined mineral oil called Luminol TR-I. The "I" in the TR-I stands for inhibited.
I'm having trouble finding anything on that one. You have a link?
 
heh....
I knew someone would find a comparable fluid for waaaay cheaper. :p
Luminol tr-i , manufactured here in Canada by Petro-Canada , Is very much like Midel7131.

Got any links to materials compatability charts for luminol tr-i ? I'd be VERY surprized if it doesnt attack rubber.
quote :"and it reacts incredibly to black light."
Id be very careful with this too : 'biodegradeable' often means "UV sensitive"...I know Midel is.

The pump in the thinktank is a little giant submersable....it will not be in ver 2.0.

I can't wait to see these projects finished ! good luck guys !
 
BrainEater said:
heh....
I knew someone would find a comparable fluid for waaaay cheaper. :p
Luminol tr-i , manufactured here in Canada by Petro-Canada , Is very much like Midel7131.

Got any links to materials compatability charts for luminol tr-i ? I'd be VERY surprized if it doesnt attack rubber.
quote :"and it reacts incredibly to black light."
Id be very careful with this too : 'biodegradeable' often means "UV sensitive"...I know Midel is.

The pump in the thinktank is a little giant submersable....it will not be in ver 2.0.

I can't wait to see these projects finished ! good luck guys !

Thanks, I appreciate the comments. Finding Luminol was the result of MONTHS of experimentation, hard work, and research.

There unfortunately isn't too much available on Luminol's chemical compatibility, but here's what I had to work with.

Safety Information (Luminol IS biodegradable):
http://www.petro-canada.ca/eng/prodserv/specialtyfluids/6940.htm

MSDS Sheets (Luminol TR-I):
http://www.online.petro-canada.ca/datasheets/en_US/lumtri.pdf

Luminol TR (Haven't tried this out yet, and don't plan to.)
http://www.online.petro-canada.ca/datasheets/en_US/lumtr.pdf

It doesn't seem to react to anything but Oxidizing agents, and considering that it's used in transformers, it SHOULD not eat rubber. I stress should because I have data that suggests it will not, but I don't trust it, hence all the testing. Also, no one has ever tried using this product in the manner that I am, so I'm taking as many precautions as possible.

BrainEater, what are the drawbacks from using a biodegradable fluid if you take measures to ensure the system is sealed from air and you have very little in the way of contaminants in the system? Are there any? Also, thanks for the name on the pump. If my Eheim can't cut the mustard, I'll have to try a Little Giant.

Any more comments, please post.

Matt.
 
I got in touch with Clearco and they recommended the 50 cSt fluid. The low-viscosity stuff was alot more expensive (like nearly Flourinert prices) and volatile. From what I can find, 50 cSt should still be pretty runny, but I'm don't know anything about the measure of viscosity. It should make my acrylic enclosure easier to seal, anyway.
 
The viscosity of water is 1 centistoke. Many light oils (organic) are around 50 cst. Not too viscous, but not watery either.
 
enraged78 said:
BrainEater, what are the drawbacks from using a biodegradable fluid if you take measures to ensure the system is sealed from air and you have very little in the way of contaminants in the system? Are there any? Also, thanks for the name on the pump. If my Eheim can't cut the mustard, I'll have to try a Little Giant.

hmmm.....
Well , these biodegradeable oils break down via oxidization , so oxygen and UV are generally bad.Sealing the tank is a good step , but be aware of pressure changes due to thermal expansion/contraction...In my case the tank is sealed except for a breather valve.

I can tell you : The Midel7131 I've been using has started to oxidize despite my efforts to prevent it.....(It has yellowed a little)....not a big issue really as my lighting is green. ;)

The last thing to be aware of is water....often these oils are highly hydroscopic -I.E. will absorb water from the air....now this isnt an issue electrically , but it can lead to oxidization.

As far as the Little giant pump goes : Great pump for water , but it won't pump cold midel worth a $h^t........I'm trying for below 0 C in ver 2.0 and midel is thick like paint at that temp.....Therefore I'll be using a 'paint mixer' for circulation.....I'd use a peristaltic pump , but they are $$....


Mark305TBI said:
It should make my acrylic enclosure easier to seal, anyway.

I'm curious what you plan to seal it with....remember : silicon oil will dissolve silicon rubber.(usually). I'd suggest contacting the company you buy your liquid from and asking them what the best sealant would be.

---------------------------------------

On a side note , I have the silver /copper /aluminum I need to start building the 500W water blocks for the pelts.....I should have some more pics soon...
 
BrainEater said:
hmmm.....
Well , these biodegradeable oils break down via oxidization , so oxygen and UV are generally bad.Sealing the tank is a good step , but be aware of pressure changes due to thermal expansion/contraction...In my case the tank is sealed except for a breather valve.

I can tell you : The Midel7131 I've been using has started to oxidize despite my efforts to prevent it.....(It has yellowed a little)....not a big issue really as my lighting is green. ;)

The last thing to be aware of is water....often these oils are highly hydroscopic -I.E. will absorb water from the air....now this isnt an issue electrically , but it can lead to oxidization.

As far as the Little giant pump goes : Great pump for water , but it won't pump cold midel worth a $h^t........I'm trying for below 0 C in ver 2.0 and midel is thick like paint at that temp.....Therefore I'll be using a 'paint mixer' for circulation.....I'd use a peristaltic pump , but they are $$....

On a side note , I have the silver /copper /aluminum I need to start building the 500W water blocks for the pelts.....I should have some more pics soon...

Awesome. Thank you very much for the reply. I spoke to some other transformer engineers about what can make this stuff oxidize this morning, and they basically mirrored what you just told me. They did not know about the UV setting off a reaction, though. I'm going to have to think about a non-UV replacment for the blacklight in the enclosure that I'm using. One thing that they did suggest was a little out there as far as I'm concerned. The transformers we use here also use Luminol TR-i in some of the newer equipment, and the older stuff that has been retro-fitted. The transformers are sealed, but they also have to worry about thermal expansion as you just mentioned. The way they get around this in a transformer is to rely on something called a "Nitrogen Blanket". What they will do is fill the transformer up to its fill level, seal the transformer, use a vacuum to suck the remaining air out, and fill the vacuum with nitrogen. I'm not sure if that kind of deal will be vaible for me. As of right now, the Luminol has not oxidized despite being in direct contact with air for a month. So far, so good. I believe that the oxidation level is rather low for this chemical, so I might be able to fend it off long enough to only change the fluid once a year or so. Water is most DEFINATELY a problem for Luminol. As you suggested, it will use the humidity in the air to complete the oxidation process necessary for it to degrade. This may become a problem in the future, but I will do my best to prevent it. I think a breather cap with some sort of filter should help a little bit. I also pinged the engineers on the possible issue of using Luminol with rubber. According to my coworkers, older transformers used rubber insulation on some electrical lines, and Luminol has been engineered to not interfere with it, so it should work without having to seal the caps with acrylic or conformial coating. I will probably do it anyway just to be safe.

One other thing, I've noticed according to the sheets at Midel, that 7131 becomes extremely viscous at 0 degrees. The Luminol does in fact become more viscous, but certainly not to the degree that Midel does. I'll have to put my Eheim to the test via a trial by fire. We'll see how it goes, I'll keep you all posted. Once my project is done, I'll submit it here and to Kyle.

P.S. - Sounds like ThinkTank 2.0 might have a lot in common with what I'm doing. This should be very interesting.

Matt.
 
Being as I live in Calgary , (petroCanada's head office is here) , Im going to look into more information on luminol.....

I'll let ya know what I find out.

:D
 
BrainEater said:
Being as I live in Calgary , (petroCanada's head office is here) , Im going to look into more information on luminol.....

I'll let ya know what I find out.

:D

Sweet. Thanks. We appreciate it.

Matt.
 
BrainEater said:
I'm curious what you plan to seal it with....remember : silicon oil will dissolve silicon rubber.(usually). I'd suggest contacting the company you buy your liquid from and asking them what the best sealant would be.
I was planning on gluing the acrylic sheets together with acetone for 5 of the sides. For the removeable side, I was planning on a gasket...with automotive silicone gasket sealer. Might have to rethink that plan.
 
I've been in conact with PetroCanada , but I'm still waiting for some compatibility data.
Just for kicks , I put together a comparison chart between the various liquids.
Mineral oil data is for common mineral oil.
Silicone oil data matches 50 cSt polydimethylsiloxane ( Clearco )

Fluid Comparison chart

:D
 
DaRkF0g said:
Does any of the above liquids/oils put off any odor?

enraged78 said:
So far, the ONLY downside I can see to this stuff is the smell. It most definately has a petrolium scent to it. However, my systems are running sealed, so as long as you don't run it open, you won't smell it. According to the MSDS sheets, Luminol is almost 100% non-toxic, but I still am not going to chance breathing it in. Working with Methyl was bad enough.

Matt.

Up a few posts
 
Ok....

I've gotten a response from PetroCanada.Here's the compatibilty data for Luminol Tr-i :



Not Recommended:
• Natural Rubber (Natural Isoprene, Latex)
• Styrene Butadiene Rubber (SBR, Buna S)
• Butyl Rubber
• Ethylene Propylene Diene Monomer (EPDM)

Recommended:
• Nitrile (Buna N)
• Epichlorohydrin (Hydrin)
• Poly Urethane (Adiprene)
• Fluorosilicine (Sylon)
• Fluoroelastomer (Viton, Fluorel)
• Polyacrylate (Hycar)
• Phosphonitrilic
• Fluoroelastomer
• TFE/Propylene (Aflas)


Good:
• Silicone (Silastic)
• Chlorosulfonate Polyethylene (Hypalon)
• Polysulfide (Thiokol)
• Ethylene Acrylate
• Vamac
• Polychloroprene (Neoprene)

Paints
Recommended:
• Epoxy
• Polyurethane based Paints
• Oil Resistant Alkyd
• Acrylic Enamel

Not Recommended:
• Latex
• Water Solvent

Plastics
Recommended:
• Fluorocarbon
• Nylon
• Delrin
• Celcon
• Polyethylene
• Polycarbonate
• Polystyrene
• PVC
• ABS



:D
 
Awesome. This is the data that I was looking for. Great work. Any idea on what kind of rubber capacitor caps are made out of? I was under the impression that they were synthetic rubber. Any truth to this? So far, after five weeks of immersion, no discoloration, cloudiness, or degradation of components on an untreaded motherboard.

Matt.
 
did you guys try out rubbing alcohol? it doesnt conduct elecrtricity, its really fluid and i doubt it attacks motherboards parts.
 
nope.
I'm not gonna either.The biggest issue is the flammability of alcohols....although the vapor pressure would be an issue as well.

I do use isopropanol for cleaning mobo's etc tho......take's a sticker right off. :cool:
 
There's another fluid candidate that I read about on a Madshrimps article in the [H]ard News. It's called OptiCool. Don't know anything about it, just adding it to the discussion. The MadShrimps article can be found here.
 
i was thinking about the problem with pressure due to thermal expansion, and i think there is a better way than a breather tube with a filter. if you want to make it a completely closed system, so no air can get in/out, therefore less chance of oxidization, then this might work better.

create the "breather tube" at the top of the case...or where ever...it doesn't matter, but instead of putting a filter on it so that air can get in or out, use something air-tight but flexible...like a balloon (don't use a balloon though, they aren't really air-tight)...but some sort of stretchy material...that way, when the inside of the case heats up and expands, the air (or nitrogen) inside is pushed up into this "balloon"

this way you wouldn't have to worry about oxidization, because there wouldn't be any new air introduced into the system. i don't know much about the fluids or cooling or any of that stuff...but after reading the thread this seems like a viable solution to that particular part of the problem...
 
Sheazle said:
i was thinking about the problem with pressure due to thermal expansion, and i think there is a better way than a breather tube with a filter. if you want to make it a completely closed system, so no air can get in/out, therefore less chance of oxidization, then this might work better.

create the "breather tube" at the top of the case...or where ever...it doesn't matter, but instead of putting a filter on it so that air can get in or out, use something air-tight but flexible...like a balloon (don't use a balloon though, they aren't really air-tight)...but some sort of stretchy material...that way, when the inside of the case heats up and expands, the air (or nitrogen) inside is pushed up into this "balloon"

this way you wouldn't have to worry about oxidization, because there wouldn't be any new air introduced into the system. i don't know much about the fluids or cooling or any of that stuff...but after reading the thread this seems like a viable solution to that particular part of the problem...

Not a bad idea at all. Mind if I use it?

Matt.
 
enraged78 said:
Not a bad idea at all. Mind if I use it?

Matt.
not at all, i'd like to see how well it works...actually, it might work better if you cut a bigger hole, instead of a tube...like a hole for an 80mm fan on top, and cover that with rubber, that way there is more surface area for the pressure to exert on, and less pressure required to make it expand, therefore less worry about breaking a seal somewhere else...

the trick will be finding a rubber that is stretchy enough and safe for use with your liquid...i know a lot of the dielectric fluids don't get along very well with most plastics/rubbers...
 
If you do go with the filter route, perhaps a moisture remover made for air compressors would work. I have one inline on my framing compressor to remove moisture. Plus only costs around $10 bucks.
 
darw_n said:
quick question, what are the wieghts of the fluids you are using?
Here's a few for what I was considering (this is the weight for 5 gallons):


HFE-7100 = 28,769.13g = 63.43 lbs (Novec Engineered Fluids)
HFE-7500 = 30,472.56g = 67.18 lbs
FC-84 = 32,743.81g = 72.19 lbs (Flourinert)
FC-77 = 33,690.17g = 74.27 lbs
FC-43 = 35,582.87g = 78.45 lbs
FC-40 = 35,390.60g = 78.02 lbs
50 cSt = 18,143.69g = 40.00 lbs (Liquid Silicone)
 
Just wondering if any of you all considered working with R/C boat parts (i.e. propeller or impeller) to circulate the fluid. They use nylon propellers so you don't have to worry about incompatability of the fluid with the prop. You can mate any kind of motor to the propeller as well, if you don't want the super loud noise that r/c boat motors make. Of course, it would require more work on your part (motor housing, sealing the location where the actual propeller enters the case, etc, etc). It'll also look pretty dumb, though.

Just a thought...

-m
 
mkim797 said:
Just wondering if any of you all considered working with R/C boat parts (i.e. propeller or impeller) to circulate the fluid. They use nylon propellers so you don't have to worry about incompatability of the fluid with the prop. You can mate any kind of motor to the propeller as well, if you don't want the super loud noise that r/c boat motors make. Of course, it would require more work on your part (motor housing, sealing the location where the actual propeller enters the case, etc, etc). It'll also look pretty dumb, though.

Just a thought...

-m
I don't think RC parts could push 24/7
 
Some really interesting ideas here. I am interested in using submersion cooling for a MythTV box that I am building in order to keep it cool and avoid noise. How has research and testing on this stuff been going? Have any problems occurred with the Luminol TR-I or the Opticool (dissolved rubber, breakdown, etc)? Also how much more economically feasible are these than Midel? Thanks for any info, I'd really like to give this stuff a shot.

-JTC
 
What about sillicone transformer oil?

Ive seen this stuff for around $60.00 a gallon, and its fairly runny stuff.

The thinktank mod looks like it would use a lot of power pumping around all of that jello-ey goop, hmm.

I have an idea...

Why not fill the whole submersible setup about 1/5th of the way, maybe 6" deep. Have the pump still suck up the fluid, and have it spray it all over the motherboard/cpu waterblock.

Wouldnt that be a much lighter and easier setup? It should work similar to watercooling, without the cost of the fluid, the weight too...

You know, suck up the fluid, pump it through a radiator somewhere first, out the radiator, spray it on the mobo/ram, through the waterblocks, then let it dribble back down to the bottom where the pump sucks it back up again and does it all over. It should work just like watercooling, but the entire motherboard would be cooled by the oil dribbling down, and the case would be the resiviour.
 
Midel and Luminol-tr are actually both replacements for silicone transformer oil.Silicone oils are toxic , and non-biodegradeable.

Midel , like any transformer fluids , is only "jello-ey goop" when It's cold (>10 C).Midel's close to water at the ~40 C I was running at.

There are other reasons midel/luminol-tr are better : The fire characteristics are much better , and the cold temp properties are superior.

The idea you are describing bob, is called 'spray cooling' .Certain Cray supercomputers use a variation of it.Here's the catch tho : In order to use spray cooling effectively , the liquid needs to be cold (> 0 C) , but in order to be pumped effectively , it needs to be thin (not viscous).

Basically , that means all transformer liquids are out.The only liquids that will work are the super expensive (400$/gal) hydrofluorocarbon liquids ( fluorinert , novec , galdol ,etc).

:D
 
Interesting thread.
I am wondering whether a giant paddle-wheel would be better to move the liquid within the case. I realize that there would be quite some resistance to the paddles, but it could perhaps be powered by a geared high torgue motor. Seems like paddles would creat a better stir than just "jet" pumps.
 
I've been researching this topic myself somewhat, too. I've looked up some possible solutions, and I came up with teflon. A teflon spray applied to any particular area of concern should prevent any degradation of the plastic/rubber, according to my research. The teflon and most of the cheaper immersion cooling solutions should not affect each other. Teflon will not interact with almost all oils, plastics, rubbers, etc.. making it a safe solution.

In regards to transformer oils being used, Shell also has a fluid that would be great for cooling. It's called Diala HFX. The bonus is that it's completely biodegradeable.
Shell Diala HFX info link (includes MSDS, Technical data sheet, and general info.):Shell USA
the downside is that they ship only from the US (Shell Canada doesn't carry it) and it only comes in 55-gallon drums.

If you guys are interested, I could buy a drum or two and divvy it up into smaller containers and ship it to ya for a small fee.

~Darren Crook
 
A good pump for viscous liquids would be any kind of sewage pump. After all, they're made to pump solid waste, so viscous liquids should be easy.

~Darren Crook
 
Back
Top