11. Cooling Loop
The liquid cooling system for the dual laser MK580 Nd:YAG + ruby system is water/glycol based. Ruby benefits from cold and is the first in the system, followed by the Nd:YAG.
The cooling system consists of the following components:
EGW (Ethylene, Glycol, Water) + inhibitor
silver coil to remove bacteria (UV lamp would destroy plastic)
magnetic float level cut off switch in reservoir
filler route to reservoir
drain route from reservoir
Hydor 1200 lph pump inside an enclosed acrylic reservoir
1/5hp Aquavia chiller modified for 0°C operation
optional acrylic reservoir with vapour phase head down to -50°C
flow rate detector cut off switch in loop
ruby laser
MK580 laser
4 fan assisted radiator
return to inlet reservoir
drain taps on both laser ends to facilitate removal
Rationale for choice of cooling liquid
The higher power lasers need liquid cooling. In the dim and distant past when a 25% over-clock of a 1GHz PC was a considerable feat, I bought a number of water cooling parts to over-clock a PC. The parts I bought included an imported modified American 1/5hp CC50 AquaVia chiller and 1200 lph (litres per hour) Hydor water pumps to match the chiller instructions 'Water flow should be a min. of 100 US gph (gallons per hour) but less than 400 gph'. 1 lph = 0.264 US gph; 1200 lph = 316.8 US gph. I also bought some anti-freeze and corrosion inhibitor, and silver strips to combat bacteria build-up.
Later I bought a CPU vapour phase cooler which I intended to use whilst immersing the entire PC in
2 litres (5.8 US gallons, 4.8 UK gallons) of Armari Inertx PFX-90, their cheapest non-conductive synthetic coolant (and to give the system a head start) cooled to 0°C by the chiller. I negotiated a bulk price discount and bought it but just as I was about to put it all together, the Intel Core 2 Duo processor appeared and I found I could over-clock it by 50% using cooling fans alone. This and the PC's subsequent tendency to boot, crash and reboot regularly afterwards (despite exercising great caution in the over-clock), deterred me from pursuing the extreme over-clock route further.
I could now use these parts to liquid cool my lasers, but I wanted to refresh my mind of the options.
The following informative site provides a wealth of useful information, much of it reproduced below:
[G22] The Importance of Using Good-Quality Water in Heat Transfer Fluid Solutions,
The Dow Chemical Company, Form No.180-01396-1099QRP.
Water, deionized water, glycol/water solutions, and dielectric fluids such as PAO and fluorocarbons are the heat transfer fluids most commonly used in high performance liquid cooling applications.
WATER / GLYCOL
[T15] Minimum recommended mineral limit for good quality water:
Calcium < 50ppm
Magnesium < 50ppm
Chloride < 25ppm
Sulfate < 25ppm
Total Hardness < 100ppm (5 grains)
PC cooling typically uses this method. 'Water’s ability to corrode metal can vary considerably depending on its chemical composition. Chloride, for example, is commonly found in tap water and can be corrosive. Facility or tap water should not be used in liquid cooling loops if it contains more than 25 PPM of chloride. The levels of calcium and magnesium in the water also need to be considered, since calcium and magnesium can form scale on metal surfaces and reduce the thermal performance of the components....Water must be unpolluted and needs some form of anti-freeze. Automotive glycol should not be used in a cooling system or heat exchanger because it contains silicate-based rust inhibitors. These inhibitors can gel and foul, coating heat exchanger surfaces and reducing their efficiency. Silicates have also been shown to significantly reduce the lifespan of pump seals. While the wrong inhibitors can cause significant problems, the right inhibitors can prevent corrosion and significantly prolong the life of a liquid cooling loop. Inhibited glycols can be purchased from companies such as Dynalene, Houghton Chemical, or the Dow Chemical Company and are highly recommended over non-inhibited glycols.
The two types of glycol most commonly used for liquid cooling applications are ethylene glycol and water (EGW), and propylene glycol and water (PGW) solutions. Ethylene glycol has desirable thermal properties, including a high boiling point, low freezing point, stability over a wide range of temperatures, and high specific heat and thermal conductivity. It also has a low viscosity and, therefore, reduced pumping requirements. EGW has more desirable physical properties than PGW, which is used in applications where toxicity might be a concern.
As the concentration of glycol in the solution increases, the thermal performance of the heat transfer fluid decreases. Therefore, it is best to use the lowest possible concentration of inhibited glycol necessary to meet your corrosion and freeze protection needs. Dow Chemical recommends a minimum concentration of 25-30% EGW. At this minimum concentration, the ethylene glycol also serves as a bactericide and fungicide.
With recirculating chillers, a solution of 30% ethylene glycol will result in only about a 3% drop in thermal performance over using water alone but will provide corrosion protection as well as freeze protection down to -15°C (5°F). The quality of the water used in the glycol solution is also important. The water should meet or exceed the limits specified above, even if you’re using an inhibited glycol. Ions in the water can cause the inhibitor to fall out of solution, resulting in fouling and corrosion.'
It is well known that glycol anti-freeze and corrosion inhibitors are hazardous to heath.
If water is to be used, the best choice is bottled spring water as it omits the chlorine found in domestic tap water.
DE-IONISED WATER
https://www.uswatersystems.com/deionized-water-vs-distilled-water
Deionized water is ultra-purified water containing no ions other than H+ and OH-. Its resistivity is close to 18MΩcm. After distillation, water is passed through a reverse osmosis membrane and a deionizing membrane which allows only H+ and OH- ions to pass through, to finally obtain deionized water. Most industrial liquid cooled lasers use de-ionised water, but this comes with its own set of problems, most notable the need to keep it non-conductive as it leaches metal ions which in turn makes it very corrosive to metals. Cheap de-ionised battery top-up fluid is a potential source but may contain contamination in the form of bits of ion exchange resin. Multiple filters in series are needed to catch these and micro amounts of metallic impurities, but conventional filters restrict fluid flow rates, necessitating alternative and expensive high flow reverse osmosis membrane types that need regular replacement. Bacteria can be inhibited with DUV lamps instead of silver strips. Carbon pellets should remove metal ions but need replacing once a month. Magnets can be used to attract metals out of the solution provided the method does not expose the fluid to conductive ions. Stainless Steel piping is required throughout; other metals will slowly leach into the solution, making it conductive. Ideally the level of impurities should be constantly measured using a TDS (Total Dissolved Solids) meter and the water and filter replaced before impurities reach 100ppm. Cheap eBay China TDS meters may fail. Given what is being protected, a quality TDS is required, together with an industrial quality calibration solution. A suitable TDS meter may be the Extech EC400. Although not recommended, it is possible to drink de-ionised water without ill-effect.
DISTILLED WATER
Distilled water is highly purified water that does not contain any salts. It is produced by boiling water into steam and then condensing the steam. As a result, organic substances having boiling point less than that of water and oils present in water can be present in distilled water. Like de-ionised water it has a high resistance to electricity, and over time ions from metal will leach into it and make it conductive. However unlike de-ionised water, it is cheaply available, although sometimes it too is produced by the same membrane process, leaving unwanted solids present that need to be filtered out.
For the purposes of this assessment, I am only considering water distilled by boiling.
https://www.quora.com/How-do-metals-react-with-distilled-water
J.Peters, Materials Scientist and Engineer, University of California, C Berkeley:
'The metal corrodes until an equilibrium concentration of metal ions in the water is reached. The corrosion process puts metal ions into solution and creates hydrogen gas:
2H_2O + M -> H_2 + M^2^+ + 2OH^-
At equilibrium, metal atoms are corroding off the metal at the same rate as ions being deposited onto the metal. Once this point is reached, a human observer would see the metal neither shrinking nor growing. For some metals (like platinum) this requires a very, very small concentration of ions in solution, so the metal will not measurably corrode. For some metals (like zinc) the equilibrium concentration is impossibly high (aka the solution becomes more metal than water), so it will keep corroding forever. For alkali metals, this process is hot enough for the hydrogen gas to ignite when it comes into contact with air. These explosions can be seen with a quick Youtube search of sodium or caesium in water. If the metal is constantly flushed with new distilled water, an equilibrium concentration will never be reached, and the metal will corrode completely. This process can occur across a massive spectrum of rates depending on the conditions.
There is oxygen in the water. The metal will corrode faster and to a higher equilibrium concentration than without oxygen, but otherwise this is exactly the same as above.
O_2 + 2H_20 + 2M -> 2M^2^+ + 4OH^-O_2 + 2H_20 + 2M -> 2M^2^+ + 4OH^-
An oxide forms (e.g. M_2O_3 or MO_2^2^-M_2O_3 or MO_2^2^-). Sometimes the oxide is an ion. These ions also have an equilibrium concentration, so a similar process to the previous cases occurs. Other times the oxide is solid (think rust). Solids don’t really have a concentration, so the metal will "rust" itself into oblivion.
A passive layer forms (e.g. stainless steel). These special solid oxides form in thin layers coating the metal and prevent any further corrosion. The passive layer is only a few atoms thick, so it is usually invisible and really hard to measure. Passive layers are very useful because they can protect metals with high equilibrium ion concentrations from corroding. However, they only form in a very limited set of circumstances and may not be practical.
Distilled water is different tap water or natural bodies of water because it lacks chlorine ions. Tap water has only very small amounts of chlorine, but it is enough to have a measurable impact on metal corrosion. Chlorine ions catalyse all the above reactions. This means they make everything happen much, much faster, but have no effect on the end result. Chlorine can also destroy the protection given by passive films.'
HYDROCARBON-DERIVED
1 SYNTHETIC OIL
'Poly-alpha-olefin (or poly-a-olefin, abbreviated as PAO) is a polymer made by polymerizing an alpha-olefina synthetic hydrocarbon. It is also used as a base stock in the production of some synthetic lubricants. PAO is used frequently in military and aerospace applications for its dielectric properties and wide operating temperatures. PAO compatible recirculating chillers are also available. PAO has a thermal conductivity of 0.14 W/m°C (0.081 BTU/hr ft °F). Although dielectric fluids provide low risk liquid cooling for electronics, they generally have a much lower thermal conductivity than water and most water-based solutions', but PAO is notably higher.
The statement 'PAO compatible recirculating chillers are also available.' implies special materials are required, and my chiller and radiator may be incompatible. Advanced PC over-clocking projects have used oil, some cooling it using LN2 (Liquid Nitrogen). However I am allergic to oil and I discarded it at the outset for this reason, as well as the difficulty of cleaning it off. The following rationale only affirms my decision:
http://beowulf.beowulf.narkive.com/0Mnedziq/oil-immersion-cooled-blades
'So here's my experience using oil as an insulator/coolant:
1) oil wicks up insulated wire, particularly stranded. Put the mobo in oil and the power supply
outside, and pretty soon your power supply will be full of oil.
2) oil leaks. There is *nothing* that is oil insulated that doesn't have a fine film of oil on its
surface eventually, unless it is in a hermetically sealed can with welded/crimped seals.
3) oil is a mess when you need to fix something.'
Material Safety Data Sheet (MSDS) for ATI PAO-4 [D10], health issues:
Inhalation: May cause respiratory irritation.
Ingestion: If ingested, aspiration may result in chemical pneumonia.
Skin Contact: Repeated or prolonged skin contact may cause irritation or dermatitis.
Eye Contact: May cause eye irritation.
Immediate Effects: May cause respiratory system irritation.
Delayed Effects: Repeated exposure to synthetic oil mists via inhalation may cause lipoid pneumonia,
inflammation with or without pnuemonia, fibrosis, and paraffinomas.
2 ELECTRONIC FLUID 1
HFE-7100:
According to 3M's datasheet: 'Novec™ Engineered Fluid HFE-7100, methoxy-nonafluorobutane (C4 F9 OCH3 ), is a clear, colorless and low-odor fluid intended to replace ozone-depleting materials in many applications.'
http://forums.procooling.com/vbb/showthread.php?t=10144
'HFE is the replacement for traditional applications of Fluorinert (FC-series fluids).
HFE is a selective solvent (it will leech similar materials out). I use HFE-7100 on a daily basis for work and things I've noticed with that fluid are:
a) if there's the smallest leak, even a pinhole, it'll find it (and the smaller it is, the harder it will be to find the leak since it is odourless and evaporates rapidly),
b) depending on the material, it will react to your tubing/o-rings/etc and will become statically charged due to it leeching material. When we had originally switched from FC-77 to HFE-7100, we found that out when various individuals were getting shocked by handling instruments in the machines.
Lo and behold, the fluid in the system was generating a charge differential of excess of 2kV.
So for now, we're using a special additive to the fluid to reduce the charge build-up in addition to material changes in the coolant loop to minimize that build-up.'
I mention HFE-7100 first, because I discarded it the moment I learned it was a static hazard.
2 ELECTRONIC FLUID 2
FLUORINERT FC-77 (now superseded by FC-770):
The final cooling medium I checked was fluorocarbon, since I already had some. Due to the high voltages involved and its advertised neutrality, it struck me as the route to go. However I had not completed my PC project, so I explored it in finer detail:
The Armari PFX-90 coolant I bought is reclaimed
3M Fluorinert FC-74 from a Cray-2 supercomputer.
Armari also used it in their own brand PCs.
In 2015 I could find no spec for FC-74 on the
internet, but 3M confirmed FC-74 was FC-77
re-badged for Cray. They also confirmed it does
not have a shelf life or degrade with age if it
is kept in its original container and not exposed
to extreme temperatures or radiation.
One of many Cray history sites:
http://ed-thelen.org/comp-hist/vs-cray-res.html
Freely downloadable brochures for Cray supercomputers can be found here:
http://www.computerhistory.org/brochures/companies.php?alpha=a-c&company=com-42b9d5d68b216
5 Woodshots Meadow, Croxley Green Business Park, Watford, Herts, WD18 8YS, UK.
support@armari.co.uk. Sales: 01923 225 550, Technical: 01923 221 102. Contact: Dan Goldsmith.
Dan said all they had to do was remove general fluff, and it still has the same thermal spec as FC-74.
However once it had been cleaned, 3M no longer allowed them to call it FC-74 so they called it PFX-90. Apparently 3M Minnesotta was the only 3M division that dealt with it, specifically for the Cray supercomputer. 3M confirmed the detailed spec of FC-77 has changed over the years, but the basic properties remain more-or-less constant.
I uncovered a glowing review from the era when I bought it:
http://hexus.net/tech/reviews/cooling/823-armaris-inertx-pf5080
'InertX is Armari's brand name for Fluorinert perfluorocarbon liquids, in the past they have used 3M as a supplier, but recently [2004] they managed to source higher performing Perfluorocarbonates in order to offer it as a mainstream solution. The three that Armari have brought to market all share the same basic properties. They're all similar in viscosity to water, meaning the vast majority of pump systems used in water-cooling systems will have no trouble pumping it round a closed loop or system. They all have a specific heat capacity that's suitable for moving large amounts of heat from block to heat exchanger. They're all completely non-conductive. You could submerge your mainboard in InertX and it would run happily. They're all clear, colourless and don't have any smell. They don't mix with water and they evaporate into the air safely. Finally, they don't transfer ions between differing metals in your cooling system (think copper block and aluminium heat exchanger) and they're all bad places for algae to live, so you don't have to suffer that either. Typical Fluorinert thermal conductivity is 0.06 vs Water 0.6 and copper 300. Cray used Fluorinert as a substitute for air. It has 2-3x the conductivity of air, so it cools better than air when pumped over a circuit board. Armari's InertX coolants freeze between -108°C and -90°C depending on the variant, and boil between +80°C and +95°C.
At the time of writing [2004], InertX pricing is as follows.
InertX PFX-90 1kG (560ml) Pro liquid coolant - £29.95
InertX PF5070 1kG (570ml) Pro liquid coolant - £39.95
InertX PF5080 1kG (560ml) Pro liquid coolant - £49.95'
http://forums.bit-tech.net/showthread.php?t=55455
'We sell PF-5080 as reclaimed Perfluorooctane. The difference between Perfluoroctane (CF-5080) and 3M Fluorinert a.k.a. Perfluorcarbon (PF-74) is: Chemically, CF-5080 is exactly the same C8 F18, chemical name Perfluorooctane. The FC fluids by 3M have a tighter spec than the PF fluids, this is why we sell them under the chemical name.' Robert Ansell, Eastern Sales Manager, TMC Industries, Inc.,
1423 Mill Lane, Waconia, Mn. 55387, Phone: 800-255-5789, Fax: 952-442-1160, rob@tmcindustries.com,
www.tmcindustries.com
I compared the specs of coolants offered by Armari with water and other Armari / 3M PFCs:
Type Pour point Boiling Point Thermal Conductivity Comments
Water 0°C +100°C 0.580 W/m/K
FC-40 - 57°C +165°C 0.065 W/m/K
FC-43 - 50°C +174°C 0.065 W/m/K
FC-74 -110°C + 97°C 0.063 W/m/K PFX-90 = reclaimed FC-74 = FC-77
FC-77 -110°C + 97°C 0.063 W/m/K
FC-770 -127°C + 95°C 0.063 W/m/K Modern replacement for FC-77
PF-5070 - 95°C + 80°C 0.060 W/m/K
PF-5080 - 30°C +101°C 0.064 W/m/K
For general scientific use, thermal conductance is the quantity of heat that passes in unit time through a plate of particular area and thickness when its opposite faces differ in temperature by one kelvin. (for reference, 0.063 W/m/°C = 0.036BTU/hr ft°F):
http://www.saylor.org/site/wp-content/uploads/2011/04/Thermal_conductivity.pdf
For a plate of thermal conductivity k, area A and thickness L this is kA/L, measured in W/K:
http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html
Convert W/cm/°C & W/m k to other units:
https://asm.matweb.com/search/GetUnits.asp
https://www.theunitconverter.com/watt-centimeter-to-watt-meter-k-conversion/
http://www.translatorscafe.com/unit-converter/en/thermal-conductivity/1-2
3M ID number for FC-74 = 98-0211-4191-0:
http://www.computerhistory.org/collections/catalog/102653085
CAS Number for FC-77 is 52623-00-4, formula = (C8F18)n.(C8F16O)m:
http://www.chemblink.com/products/52623-00-4.htm
PFC resistance to UV radiation:
A paper I read said Freon E-3 had been used for laser cooling but was slowly degraded by UV radiation. Given DUV would be produced by my quartz flashlamps, I asked 3M if FC-77 had the same issues. They said tests run on FC-104 indicated high resistance to photolysis (the decomposition or separation of molecules by the action of light), and FC-77 would not be substantially different, attaching corroborative documents, reproduced below:
Here are more links I found on the subject of PFC cooling:
http://forums.bit-tech.net/showthread.php?t=55455
'...since fluorinert is considerably heavier than the same volume of water, the same flow rate will result in a greater mass of fluorinert passing through your blocks. Therefore the fact that its spec heat capacity is less than that of water is misleading.'
http://arstechnica.com/civis/viewtopic.php?f=7&t=566337
'...When a pump is used, the flow should be turbulent. This allows PFC to carry away almost as much heat as water would (about 75%), on a watt-seconds/gallon basis.'
The following Russian site discusses FC-74 in the Cray (photos) and discusses the chemical composition of other FCs, implying the Russians tried to copy FC-74 but failed (use Google translate):
http://engineering-ru.dreamwidth.org/169960.html
PFC viscosity vs water
I found a comment on a cooling forum: 'FC stuff is generally a more viscous fluid than water making it a pain to pump with a normal water pump.', but this doesn't seem to apply to PFX-90 according to Armari's original sales literature, and the curves below:
Curves: FC-77 vs water, viscosity in Centistokes (cSt) over temperature:
(different scales)
The above right viscosity chart for fluorocarbon fluids was taken from the following 3M document:
MATERIALS COMPATIBILITY
Next I scanned the internet for any compatibility issues associated with FC-74, and discovered something worrying: PFC degrades most flexible plastics:
http://forum.corsair.com/v3/showthread.php?t=25976
'It hurts because it will leak or permeate in systems that will hold air and or water. This hurts because of its price. You will have to use metal pipes because of this. It will permeate through PVC, Tygon, and silicon tubing. You will also want your pump inside a metal reservoir. I might recommend using copper tubing because it is fairly cheap and relatively easy to work with. It also evaporates faster than anything I've ever seen. I saw someone accidentally spill a quart into the open air 30 feet off the ground on a 100°F day. It never hit the ground. This stuff evaporates at several times the rate of alcohol at room temperature. It will also evaporate on a very cold day. So, this added to its leakage rate means you could easily be topping off your system once per month or so and that gets expensive. Not only is the stuff expensive, the permeation levels require costly sealing solutions.
3M Novec and Fluorinert are extremely difficult to contain in a basic sealed system for a home-brewed liquid-cooling PC. A person would need to know industrial refrigeration prior to using it. As for having it work in the [Bosch] Hydrocool pump, the answer is nope. The fluorinated fluids are very difficult to seal, and highly corrosive to materials containing **plasticisers. A specialised pump would be required for such a complex cooling system. The Bosch pump materials would in no way hold up.'
**A plasticiser is a substance which when added to a material, usually a plastic, makes it flexible, resilient and easier to handle. The most common plasticisers used for PVC are phthalates.
According to the 3M Thermal Management Fluids and Services brochure above, FC is generally ok with most hard plastics .
The 2006 Sandia Labs report (below) measured FC vapour and liquid absorption for a number of plastics:
ABS (Acrylonitrile Butadiene Styrene)
Acrylic
Nylon
Phenolic
Polycarbonate
PEEK (Poly Ether Ether Ketone)
Polyethylene
Polypropylene
Polysulfone
PVC (Poly Vinly Chloride)
PTFE (Poly Tetra Fluoro Ethylene) 'Teflon'
Thermoplastics
The report found up Most of the materials tested absorbed a little FC. but Teflon (PTFE) in particular absorbed up to 8%. RTV silicone absorption was less but still significant. The report suggested the materials should still function as intended however it should be noted their tests only spanned 4 weeks. IMO if 8% absorption occurred in just one month, I hate to imagine what happens after a year and I am surprised the report conclusion did not consider recommending follow-up long term (e.g. 1 year) evaluation, since any industrial equipment exposure to FC would be far longer than 1 month.
I asked 3M for clarification of the above and this was their reply:
'Hard plastics like Delrin (and Nylon, PEEK etc) are considered highly compatible.
All the PFCs - FC-77, 770, 74 included - will swell or deform fluoriniated polymers: PTFE, FEP, etc. (I’ve even seen thin films dissolve…) - these are generally to be avoided - they can also act as a ‘wick’ rather than a seal.
Heavily plasticised/filled polymers can be embrittled as the plasticisers leach out - PVC cable sheathing most notably; "pure" polymers are normally OK.
More "basic" elastomers such as butyl or Nitrile rubbers are normally considered compatible and would be worth looking at for flexible tubing and for seals.
For metallic tubing - copper is not so good! The ability of Fluorinert to dissolve Oxygen gives rise to discolouration (Verdigris) in the tubes. However if it is a closed system (i.e. no atmospheric contact between the fluorinert and air) and the liquid is de-gassed before filling the system (by heating to BP) then there will be no oxygen to affect the copper.
De-gassing a sealed system before filling is recommended, otherwise it will de-gas within the system when it gets hot. If it’s an open "atmospheric" system - no point in de-gassing, air is readily dissolved back into the fluid on contact.
All other common metals/alloys are considered compatible.'
I had bought the Inertx PFX-90 because it was supposed to be inert! I had discarded de-ionised water because it forced me to use stainless steel fittings! Only now did I realise PFX-90 is incompatible with most flexible plastics and for it to work with my chiller and radiator which have some copper piping, it has to be a closed system. Boiling it and filling the system without exposure to the air is not something I think I can do, which leaves me with an open system that cannot use copper.
HEALTH ISSUES
FC is not harmful to the ozone layer, but is a known source affecting global warming. If the cooling system is totally enclosed, that risk is minimised. Facilities exist to process it for disposal (fire is the usual method) after use.
However it would be a good idea to avoid exposure to FC liquid, particularly to eyes, as the evaporation process would instantly remove all fluid from the eye. Gloves and safety glasses should be worn when handling.
Boston, MA., 2012: 'A new study finds that perfluorinated compounds (PFCs), widely used in manufactured products such as non-stick cookware, waterproof clothing, and fast-food packaging, were associated with lowered immune response to vaccinations in children. It is the first study to document how PFCs, which can be transferred to children prenatally (via the mother) and postnatally from exposure in the environment, can adversely affect vaccine response. The study appears in the January 25, 2012 issue of the Journal of the American Medical Association (JAMA).'
http://www.hsph.harvard.edu/news/press-releases/pfcs-childhood-vaccinations
If FC solids have this issue, it seems likely liquid PFCs have it too: exposure is to be avoided.
Then I found this really informative post discussing cooling the Cray-2 with FC-74:
http://ed-thelen.org/comp-hist/vs-cray-res.html
'I heard the temps: the Fluorinert came in at 80°F (27°C) at the bottom near power supplies and exited the top of the 2 around 90°F (32°C). One critical problem was the formation of bubbles. Too big a bubble could create heating problems, so bubbles could not get too big. Cray 2 was cooled by FC-74, product now obsoleted. Called 800-833-5045, asked for "Thermal Management". Lou Tousignant 3M 1-651-736-5242 handled Cray cooling said cooling was definitely convective, not boiling, and FC-74 boils at +97°C. He said most of the bubbles you see are actually air that dissolved when the Fluorinert was cool and exposed to air. When the Fluorinert warms, it can dissolve less air (like water) and the air appears as bubbles. Any steady bubbling was definitely a trouble. There are occasionally pockets of air at the top Cray-2 but no problem as no heat released there. Terry Greyzck: (still works for Cray) The fluid ran directly over the circuit boards, referred to as "direct-immersion cooling", at about a rate of 1 inch per second. The Cray-3 ran at about 10 inches per second, fast enough that they had to worry about erosion. The Cray T90 also uses direct immersion cooling. Fluorinert can also be used as artificial plasma, but it's way too expensive for that purpose. If you heat it up enough (around 500 degrees) one of the decomposition products is **phosgene - which is why a big vent hood can be found over Cray-2 installations in case of fire, to vent the phosgene outside. Aside from that, it's about the most harmless stuff around.'
**The MSDS reveals it's not phosgene, it's Perfluoroisobutylene (PFIB), and it only has to be heated to 392°F (= 200°C), easily achieved on the wall of a flash lamp in an air bubble or void.
After this, the more I dug the worse it got.
http://en.wikipedia.org/wiki/Perfluoroisobutene
'PFIB is about 10 times as toxic as phosgene. Its inhalation can lead to pulmonary edema, which may be fatal. Onset of symptoms can take 1-4 hours after inhalation. Treatment is based on management of the pulmonary edema (usually with high-dose corticoids and other medication/measures) and associated disorders (e.g. heart failure, hypocalcemia etc.). Many cases resolve within 72 hours without major long-term effects. In contact with water PFIB undergoes rapid hydrolysis, producing various reactive compounds and fluorophosgene. PFIB is a product of pyrolysis of polytetrafluoroethylene (PTFE), one of the substances causing polymer fume fever.' (so common PTFE is equally dangerous!)
My original Inertx PFX-90 MSDS dated 30/09/2003 [D12] identified these chemicals, but omitted the warnings present in 3M's FC-77 MSDS. When I bought my PFX-90 in 2004, everything suggested to someone without a chemistry background, that all was well and it was inert as advertised. The worst it said was 'May cause long-term adverse effects in the aquatic environment.' It was only when I enquired in 2015 that I learned it was FC-77, and read its full MSDS, dated 03/09/2002
[D13].
I checked the Material Data Safety Sheet (MSDS) for HFE-7100 [D11] and found above 300ºC it also produces Hydrogen fluoride and Perfluoroisobutylene, so it's just as bad. I have deliberately chosen older MSDS versions because I found newer ones such as the 2016 MSDS for HFE-7100 omit the decomposition temperature, but if it still has the old part number, it must be the same stuff:
3M's MSDS for FC-77 [D13]:
'SECTION 10: STABILITY AND REACTIVITY:
Stability: Stable
Materials and Conditions to Avoid: Finely divided active metals; Alkali and alkaline earth metals
Heat greater than 200ºC: Hazardous Decomposition or By-Products
Hydrogen Fluoride At Elevated Temperatures - greater than 200ºC
Perfluoroisobutylene (PFIB) At Elevated Temperatures - greater than 200ºC
Hazardous Decomposition: If the product is exposed to extreme condition of heat from misuse or equipment failure, toxic decomposition products that include hydrogen fluoride and perfluoroisobutylene can occur.
Hydrogen fluoride (CAS No. 7664-39-3) has an ACGIH Threshold Limit Value - Ceiling of 3 ppm (as fluoride), an OSHA Permissible Exposure Limit - Time Weighted Average of 3ppm (as fluoride) and a revoked OSHA Permissible Exposure Limit - Short Term Exposure Limit (which is enforced by some State Right-To-Know programs) of 6ppm (as fluoride).
Hydrogen fluoride may cause respiratory tract irritation, dental or skeletal fluorosis and irritation or burns to the eyes or skin, particularly when dissolved in water (hydrofluoric acid). The odor threshold for HF is 0.04ppm, providing good warning properties for exposure.
Perfluoroisobutylene(CAS No. 382-21-8) has an ACGIH Threshold Limit Value - Ceiling of 0.01ppm. Perfluoroisobutylene may cause respiratory tract irritation, pulmonary edema, cyanosis, and effect on the hematopoietic system.'
Additionally,
http://koregronews.blogspot.co.uk/2012/12/frying-potato-chip-in-fluorinert-fc-40.html
'Hydrogen Fluoride is an incredibly dangerous colourless, odourless gas which upon contact with bodily tissue, forms hydrofluoric acid. It's very good at penetrating tissue and is known for almost instantly destroying lung tissue and corneas.'
I found the above comment attached to a duplicate of the original video produced by Ben Krasnow, below, followed by its own original comments:
http://www.youtube.com/watch?v=a4gYv2BK-HQ
David Shepherd 2 years ago (2015)
'At Cray Research we used Flourinert extensively in the Tritons, and if we ever had a 'burn', where a circuit failed and heated it to boiling, we had to evacuate the building immediately, as the gas is the most powerful carcinogen known... 1 part in 5 billion if memory serves... at any rate, in theory anyway, we avoided it more than the plague.'
The last comment really cans it for a DIY PFC cooled laser. One of the issues with liquid cooled lasers is the possibility of hot spots occurring in the fluid where eddies form due to voids in the cooling block cavity around the flash lamp and rod. With my DIY ruby block, this is exactly the thing I fear could occur and the last thing I want is a pocket of Perfluoroisobutylene. It's simply too dangerous.
Even without this danger, I cannot envisage an easy way to boil the stuff to remove oxygen in order to run a closed system cooling loop. There is bound to be a leak somewhere, and again, physical exposure.
I could perhaps seal the PFC with a lighter layer of water, but I would then face all the issues associated with it so I might as well use water to begin with.
Lastly, PFC severely limits the materials I can use: I would have to replace the silicone O-rings I already bought for the flash lamps and ruby rod, and find an alternative for the neoprene sheet I plan to seal the ruby chamber. In fact, every material in contact with PFC (reservoirs, hoses, pumps, sensors) would need to be run by 3M for suitability; impractical for experimentation.
SUMMARY
Water/glycol
Conducts electricity - potential danger of electrocution
Readily available, cheap
Minimal filtering required
Avoid tap water contaminated by chlorine
Compatible with most plastics and metals
Glycol/inhibitor hazardous to health
Distilled water
High resistance to electricity
Compatible with most plastics and stainless steel
Highly corrosive to many metals
The corrosion process puts metal ions into solution and creates hydrogen gas
Cheaper than de-ionised water for a reason - it has more impurities in it
Needs expensive filtering equipment with continual replacement
De-ionised water
High resistance to electricity
Compatible with most plastics and stainless steel
Highly corrosive to many metals
The corrosion process puts metal ions into solution and creates hydrogen gas
Needs expensive filtering equipment with continual replacement
Oil
High resistance to electricity
Contaminates everything
Difficult to remove
May need specialist seals
Potential source of allergy
Hazardous to health
HFE-7100 & PFC
High resistance to electricity
Incompatible with most flexible plastics
Incompatible with copper unless pre-boiled in a closed system
Difficult to seal within a system
Highly volatile and expensive to replace
Hazardous to health
Lethal above 200ºC
CONCLUSION
When I started out I had every expectation I would be using the PFC left over from my old PC cooling project, but my new research drove me back to square one. Despite the potential for electric shock, I decided the best choice for the LHC cooling system is bottled spring water with added glycol. The bottled water should have a mineral analysis label; only water meeting the mineral quantities permitted in [T14] above will be used. This looks feasible.
An alternative could be to use cheap distilled water and add the minimum amount of minerals necessary to stop its corrosive properties, but that would require active regulation and be just as much hassle as filtering de-ionised water.
A water based system allows the greatest flexibility in component materials and setup, particularly important in an experimental system, and there are minimal issues with an open system. If a metal optics breadboard is employed, it will be suitably earthed, with system power derived via a Residual Current protection Device (RCD).
The very low pulse rate and thus 'on time' of a LIBS system helps reduce the risk of electric shock.
However the flash lamps can no longer be triggered by external electrodes as these would be in the water, which will certainly lethally conduct at such high voltages. The flash lamps will therefore require series triggering, the simplest solution being, inducing a high voltage pulse on one electrode.
IMPLEMENTATION
The 1/5hp Aquavia chiller has been modified to go down to 0°C. However there is potential to increase ruby laser power by lowering its cooling water temperature even further using a vapour phase cooling system I have left over from the PC cooling days, immersing its copper head into a reservoir before the ruby laser. A safety interlock will be needed to ensure it cannot drop below the anti-freeze limit.
Text