About Me

My Photo

I previously worked on Virtual Reality and other hardware at Valve.  I currently work at Google[x].

Prior to starting at Valve, I built computer peripherals such as keyboards, mice, and joysticks that were designed to be used inside MRI machines.  My company, Mag Design and Engineering, sold these devices directly to researchers at academic institutions who used them to publish scientific papers in peer-reviewed journals.

After work, I spend time on many different types of projects that usually involve circuit design, machining, material selection, and general fabrication/hacking.  My favorite place to be is my home workshop.

ben dot krasnow at gmail




Saturday, September 3, 2011

A close look at supercritical carbon dioxide CO2

I built a pressure vessel from aluminum and acrylic and filled it by placing pieces of dry ice inside. The dry ice melts under high pressure, and forms a liquid and gas phase. When the vessel is heated, the CO2 becomes supercritical -- meaning the liquid and gas phases merge together into a new phase that has properties of a gas, but the density of a liquid.

Supercritical CO2 is a good solvent, and is used for decaffeinating coffee, dry cleaning clothes, and other situations where avoiding a hydrocarbon solvent is desirable for environmental or health reasons.

If you have a suggestion for what I should do with the supercritical CO2, please leave a comment.

Here are a few engineering calculations that I used to determine the pressure capacity of the chamber:

1. Hoop stress in the aluminum ring:
The aluminum alloy and heat treatment is unknown unfortunately, which makes a huge difference in its material properties. Since it is a structural tube, I will assume 6061-T4, which has a yield strength of about 40 ksi.

ID = 1.1", OD= 1.5" (to the inner edge of the bolt circle)
Chamber pressure = 3000 psi
Hoop stress at inner edge = 10ksi

So, there is a safety factor of 4, but the additional material outside the bolt circle will actually add to this factor. In theory, the aluminum will yield at 12000 psi chamber pressure.

2. Bending force on the acrylic windows:
Acrylic ultimate strength: 10 ksi. It doesn't yield. It is elastic, then breaks. Modulus: 400 ksi
The plate is not a thin plate, but the results show only a 0.004" deflection at the center under a chamber pressure of 3000 psi.


This shows a stress of about 4.3 ksi for a 1.25" thick acrylic plate with 1.35" radius. The pressure-bearing radius is larger than the inner radius of the aluminum ring. This has a safety factor of 10/4.3 = 2.3. In theory the acrylic will break apart when the chamber reaches 7000 psi.

3. Stress on the bolts:

Total window area is about (pi)(1.35)^2 = 5.7", so total force when chamber pressure is 3000 psi is (5.7)(3000) = 17,200 pounds! I will use six bolts, so each bolt must hold 17,200/6 = 2860 pounds.

1/4-20 bolts are NOT strong enough -- even at grade 8!

5/16 bolts would be OK in grade 8, but I wanted a higher safety margin, and I don't like 5/16 bolts.

I chose 3/8" grade 8 bolts, which have a working load of almost 7000 pounds. I wanted to be sure bolt failure could not possibly be the failure mode that breaks the whole system. I also used grade 8 nuts, which should ensure the failure happens within the fastener, not by shearing the threads out of the nut or bolt. I am not positive about this, though.

4. Pipe threads:

I wasn't sure what 1/8" pipe threads are capable of holding, but McMaster sells such fittings that are rated for 5000 psi (like the gauge that I used), so I assume a brass part can hold such a load. I cut threads into the aluminum so it's possible that the pipe thread in aluminum could fail (ie the gauge or valve could be pushed out, shearing the threads right out of the aluminum ring). It might be possible to add up all of the area of the pipe thread cross-sectional area, but it seems silly and unlikely to be at all accurate.

5. Temperature concerns:

The acrylic has a glass transition temperature of at least 180*F, but it should not be heated anywhere near this temperature or else its ultimate strength rating may not be valid. I would say 130*F is the upper safe limit.

6. Effect of supercritical CO2 on the acrylic and O-ring:

I used buna-n O-rings, which may affected by exposure to SC CO2. They are very unlikely to fail in the short term, and I can change the O-rings for every experiment if I want.

The acrylic showed signs of crazing after just one supercritical CO2 cycle. I think the crazing is unlikely to affect the acrylic's ability to hold pressure, but there is a slight concern.

The most likely failure mode would occur when the acrylic reaches its ultimate strength, and suddenly breaks. Unlike pressure vessels made from ductile materials, which can be designed to yield and leak before breaking, the acrylic will suddenly blast apart without leaking first. If the equations and material specs are correct, 3000 psi should be OK, but I would not want to go much higher.


  1. Hey Ben, Thanks for this interesting video! How do you end up safely venting the chamber, since it has no valve?

  2. Joey, it was indeed a problem: http://www.youtube.com/watch?v=M4CTkicgKtE

    I added a valve, and it works well now.

  3. Very cool Ben... Are the o-rings made of any special type of rubber, or just standard o-rings?

  4. Hash, I used plain buna-n O-rings. Someone mentioned the supercritical CO2 might penetrate the buna-n, and cause a failure upon decompression. I am not sure if this happened during my O-ring mishap. After adding the valve, I don't anticipate a similar problem, but it's true the O-rings may not last long in this environment.

  5. Hi Ben,

    A CF flange view port design with copper gasket seal should solve the O-ring problem.


  6. Have you thought of using a bar clamp to temporarily hold pressure while you are tightening the bolts? There are some quick-clamp units with rubber facing that might work.

  7. This is a really cool project, Ben. I love being able to see the stuff that's usually hidden inside metal pressure tanks.

    Supercritical ethanol can be used to "see" high energy particles such as cosmic rays as with cloud chambers, I wonder if you could use co2 in the same way. You'd have to get just the right temperature and pressure where the co2 sits just above the condensation point - at least that's how the cloud chambers work.

  8. Very cool setup. Makes me want to go build one. :)

    One thing to watch out for, though; you mentioned trying to fill the chamber more, so you need to be very careful not to overfill it. If the liquid phase doesn't have the room to expand into the gas/supercritical stage, pressures can skyrocket rapidly.

    You might consider machining the aluminum ring for a burst disc, such as from a paintball tank, for safety.


  9. I think I might cool the chamber in something like a dry ice/isopropanol bath. Then you could close it up before the pressure builds. Also, you can weigh the CO2... 'just toss some in' is a common approach in chemistry, but ill-advised.

    If you have any worries about the bolt holes letting loose unexpectedly, why not do experiments with the chamber in a pressure cooker or other metal containment vessel? Your posts are very cool. Having you taken out by this will not do.

    You can do lipid extractions from plants or from food. Extract the oil from chips or cake. See how much of the food is fat. Be horrified. Remember how good they are, and ignore your results.

  10. I second the thoughts about burst disks, too.

  11. Hi there,

    Beautiful demo, and a nice bit of machining too!

    What are the engineering formulas you used to calculate the thickness of the acrylic? also, will you share an engineering drawing or dimensions of the vessel? I'd like to have a go at making one of these, except perhaps with a valve and safety burst disc thrown into the mix.

    Recently graduated mechanical engineer here, so feel free to go technical on me...



  12. Seb, I added some engineering info to the blog post. I don't have any drawings, and the dimensions were chosen solely to suit the material that I had on hand.

    I am going to try caffeine extraction, but I will be using an all-aluminum chamber with a safety disc. I will be heating it to temperatures much higher than the acrylic could handle. Stay tuned.

    Let me know if you have any input regarding my engineering analysis.

  13. Keep up the good work, Ben!

    Have you thought about seals using Indium metal? Could support long-term display of liquid CO2!

    How about liquid Oxygen? It would be supercritial at about 750psi, I think.

    Liquid Xenon would also be doable at just below room temperature (or a fridge) at less pressure than your CO2.

  14. Same Anon as before, I had a recommendation- Aerogels! It would be wonderful to make these with a window, most ESPECIALLY lanthanide oxide aerogels!


  15. Look up Dr. Goodling from the Mechanical Engineering department (now retired) in Auburn University. Back in 1984 he had this set up to show the triple point exactly. He could have a storm inside the vessel. Maybe he can give you some pointers and a proper clear wall material.
    Miguel Reznicek

  16. I think you should stop messing with your device and CO2.
    Acrylic has a high affinity for sc CO2. The crazing you saw is the effect of CO2 dissolving through the acrylic window surface, and followed by the creation of bubbles when you depressurized the vessel.Under long exposure (several hours), acrylic will soften into a rubber, and the whole thing will blow up. Believe me, you don't want to be around when this happens.

  17. Anonymous, you should see my video regarding the acrylic chamber after a one-week exposure to liquid CO2, and my video that shows the aftermath of an O-ring blowout. I am building some new acrylic windows and will continue playing with things that interest me. Thanks for your concern.

  18. It's great! Thanks for share.
    By the way, could you introduce some other transparent materials which can sustain with pressure from 3000 to 5000psi?
    Thank you so much!

  19. Your calculations are helpful in guiding design of the vessel...except for the bit where your O-ring is probably the weakest point - even if it wasn't already extruding into the space around it.

    +1 for a bursting disc too.

  20. Fantastic job!!
    Off course the acrylic isn't the best(and safe) material for sc-co2 using, but it's cheap!!! and it is really cool to see someone building a high pressure view cell with these material!!!
    I'm work with SC-CO2 phase equilibria and extraction of biocompounds!

    Cheers man!!!

    André Zibetti

  21. Have you thought about using thin glass panes reinforced by acrylic? Glass should be chemically inert and you won't need it for structural purposes.

  22. hey im at a point where o-ring failure cant happen. my vessel rating is 6,000 psi and wondered your experience w seals and what works

  23. Very cool to be able to see the phase change!

    We work with s.c. CO2 for subsurface sequestration of (gaseous) CO2 as a climate change mitigation strategy. When CO2 is injected into the subsurface storage reservoir, the CO2 needs to be s.c. due to P,T from overburden.


  24. Hello Ben;
          I have been looking everywhere for answers to these I guess unusual questions. Please help if you can or send me to the person who can, if you know of somebody.
    1. Assuming you have a container that can withstand the pressures, can you fill a container completely full of liquid co2?
    2. If you can, what pressure would it exert at say 70 degrees F.
        Note: I be leave you probably can and the pressure would be whatever the vapor pressure 
                  Would be at that temperature, but not sure.
    3. If you can fill it completely OR even if you can only fill it to say 90% full, can you now change the entire liquid content to a gas by only raising/lowering Temperature or pressure in the closed container?
          Note: I have seen on you tube a video demonstration on supercritical co2 and it showed the 
                    Container about 60-70% full and by changing temp and pressure it all changed to 
                    Vapor in the cylinder before it went to scco2 ( supercritical co2). I am curious if it can
                    Be done with container being full or if not how full would it work before the saturation
                    Level would become so high it would force it to liquid form.
    4. Approximately liquid co2 has a density/weight of 5.5 lbs per gallon, gas co2 is 3.5 lbs per gal,
        And scco2 at about 4 lbs per gallon. My 4th question is does this density have anything to
        Do with gaseous concentration? Meaning for example if you have to identical closed container
        Of co2, 1 with only 1 inch of liquid co2 and the other just 1 inch short of full with liquid co2, then 
        You change both to gaseous co2, would both containers still have the same density in the gas
        Of the reported 3.5 lbs per gallon, or would the gas in the container that was almost full of 
        Liquid  co2 have a higher density due to its much higher concentration?

    I know these are unusual questions and I have been trying to find the answers to them for 2 weeks now looking online. You seemed very  knowledgable about co2 in your video are excel ant. I am hoping you will know the answers, if not hopefully you can steer me the right direction.
    Thank you for your help.  

  25. Timothy,

    1. Sure.
    2. Yes, just the vapor pressure -- about 850psi at room temperature.
    3. If you raise the temperature of a container mostly full of liquid CO2, the entire contents will eventually become a supercritical fluid. The density of the fluid will be the average density of the container contents (liquid and gas) before heat was applied.
    4. A gas has no intrinsic density. You must specify the temperature and pressure in order to calculate the density. When comparing two sealed containers with different levels of liquid CO2 at room temperature, the properties of the gas phase and liquid phase should be the same in each container because each system is at equilibrium, and positioned at the boundary of the liquid/gas transition.

  26. Thank you Ben if I could trouble you one more time please
    Here is my idea. Please tell me what you think.

    Rough Example:
    I would take a 24" inside diameter pipe 40' long and mounted vertical? I would fill  as full as possible with liquid co2. Then take a floating piston of about 1000 lbs constructing it so it displaces enough liquid co2 to float but still be much heavier then vapor co2. I would connect this floating piston to a 1 inch diamiter 40 foot long straight bar, connected to the center of the top of the piston where it would then go straight up through an appropriate sleeved/seal (that could stand the pressure) through the top lid of the pipe then go straight up into the air. Around this bar (shaft) On the lid of the pipe Icould put two adjasent rollers that would be pressed inward against the rod thereby  transferring  the up/down travel of the rod to a spinning motion then connect that to a gear system that would cause the spinning rotation to be in the same direction if the bar was going up or down. I would then connect that to a very heavy flywheel system which would help with the intermitancy of the system as the rod slows and changes direction. Then I could connect that to a generator or pump etc.

    Operation example:
    The Basic cycle  would be to maintain the temperature to (well under the 88 degrees F required for the co2 to go critical) say about room temperature 72 degrees F. Then I would adjust the pressure by  lowering the pressure of the container by venting off vapor  bringing the pressure down below the liquid/vapor phase line and the liquid changing to gas allowing the 1000 lbs piston to fall per gravity, then once the piston is on the bottom I would inject liquid co2 (from another tank at higher pressure) into the pipe system raising the pressure thereby changing the co2 back to a liquid, which in in turn would raise the piston back to the top using buoyancy. The second tank would be constantly maintained at the higher pressure by a pump circulating co2 that was bleed of the system in the downstroke. And depending on how fast the co2 changes phase would determine the speed of the system.
    The amount of energy it would take to keep the ( higher) pressure tank pressurized I believe would be much less then the energy created by the system.
    I am pretty sure I am missing something that would prevent this to work as there are none in operation that I know of and there are many smarter then me but in my mind it seems to work. Looking at the phase change diagram and just using a straight up line at say 70 degrees F and only changing the pressure, also a YouTube video I saw seems to suggest it would work. The sight is 
    Thank you for reading all this, please let me know what you think.
    Timothy Dennis

  27. Timothy, you've described a variation of a basic heat engine. It should work, but you may not have found one yet because there are more efficient ways to do it. In thermodynamic engines, heat flows from a source to a sink, and performs mechanical work along the way. The temperature differential between the source and sink place an upper limit on how efficient the system can be. In practice, there are lots of other factors that create inefficiencies. In your system, you are using the buoyant force created by changing the density of the working fluid via pumping and/or heating. Unless you are burning a fuel, using electricity, or inputting energy in some way, the system will not produce any mechanical work for you. If you have a heat source, your best bet is to boil water into steam and spin a turbine. Water is cheap, non-toxic, well-understood, and performs well in turbines. Using a different working fluid will not help unless your system has a particular set of temperatures which must be maintained.

  28. Hi, Ben!

    You had a wonderful experiment.
    I am from Russia. I have an idea for you.
    Carbon dioxide can be evaporated and condensed otherwise.


    You can take the case as an engineer.

    Regards, Emil Kutin, 21/11/2012 12:57

  29. which pump is best suited for pumping liq CO2.

  30. hey ben, thanks for sharing all your hard work much appreciated...Out of simplicity Ive noticed on your aerogel video you use plumbers pipe for the sc vessel, can one just get an oversized Tee fitting, ptfe tape and plug up the openings and put a needle valve on it? would this hold enough pressure to turn super critical?

  31. You should make sure that all of the components are rated to hold the pressure. You'd probably want to add a gauge as well so that you can detect an overpressure situation.

  32. Hello, Ben!
    You mentioned dimensions of aluminum ring as ID = 1.1", OD= 1.5", but according to video I can estimate that they are larger. Did you mean radii rather than diameters? Asking, because want to assemble same chamber.

    Best wishes,

  33. I have been building a working super-critical CO2 extraction machine, I am wondering if you know of any alternatives to using Teflon for thread sealer, as Teflon is soluble in SCCO2.

  34. mattkc, it might be possible to use indium metal. It's very soft and is used for crush seals in high vacuum systems.

  35. Hi Ben,
    To arrive in the critical point (31 °C - 77 bar), the CO2 in the vessel, has to follow the vapour-pressure line in the (p,T)-graph. For that reason it is important to use an exact mass of CO2, depending on the volume of the container.
    When you take not enough dry ice, all CO2-liquid will be evapored before reaching the critical point and the system will be in gas-phase.
    Because the density of liquid and vapour are nearly equal close to the critical point, it may seem as if the meniscus disappears and as if the system reaches the critical point, which is not ...
    If you take too much dry ice, the liquid expands, absorbs the vapour and the system will be in liquid phase.
    Again, it may seem as if the meniscus disappears and as if the system reaches the critical point, which is not ...
    The only way to examin it: heat very slowly, so that temperature measured is really temperature of the CO2-system.If you heat to fast, there is a retardation!
    At 31 °C you will see disappear the liquid surface and you have to measure 77 bar pressure.
    I can send you the graph by email as attachment
    Please let me know if you are interested

  36. sorry, critical pressure is 74 bar, not 77 bar!

  37. Hi Ben,

    Thank you for sharing your works. I have 2 short questions about making supercritical co2.
    1. In the video, dry ice was used to make a liquid co2. Was heating also used?
    2. I wonder if the final state of co2 has a pressure more than 1070 psi which is near critical point of co2.

    Thank you for your time.

  38. I really wouldn't use acryllic, it is far more brittle than polycarbonate, which is why acryllic is banned from use as safety guards in factories.

    Also, the crazing may be from thermal shock, who knows what that has done to the strength of your acryllic.

    Finally +1 for a burst disc or pressure relief valve - best you calculate the stored energy before you next pressurise this puppy, then think about how fast a 20 gram fragment might travel with say 30% of that stored energy converted to kinetic energy.

  39. Hello Ben. I am a student at Colorado Mesa University and i am building a super critical co2 extractor system. I just finished my first test of getting dry ice to turn to liquid and hold at 800 psi with 65 degrees F. I depressurized my chamber with a needle valve and took it apart to find my inner o-ring grew an extra 2 inches. Do you have any ideas on what could have caused this. I am using the same o-rings you used.

  40. Run some pipe tobacco through it and see if you can make a tobacco concentrate that could be used for ejuice and not cause gunking