Your Science Zine

Build your own Liquid Nitrogen Generator

liquid nitrogen generator

I have always been intrigued by science demonstrations using liquid nitrogen, and often made trips to a local welding supply store with my stainless steel vacuum flask to purchase liquid nitrogen and satisfy my cryogenic craving at home. After a few fill-ups, I wondered about the possibility of making liquid nitrogen on demand. Some companies have already produced self-contained liquid nitrogen generators that are designed for small laboratories. The Elan2 would be ideal for home experimenters, but the cost is over $10,000, so I decided to build a similar device with less total output, lower purity, and at much lower cost. The device that I built cost less than $500 and produces l liter of liquid nitrogen per day.

Nearly all large-scale liquid nitrogen is made by compressing, cooling, and expanding air. This process removes heat from the air and can be repeated until the air liquefies. The condensing gasses are then separated using fractional distillation. This process cannot be easily scaled down because it relies on maintaining a complex, large distillation column to separate nitrogen from the other gasses in air. To avoid using a distillation column, one could use a nitrogen separation device to strip out the nitrogen from air at room temperature. Then, the room temperature nitrogen can be liquefied via the standard compression and expansion method. This is likely the process used in the Elan2 generator. However, it still requires the use of a very high pressure compressor and heat exchanger, extensive insulation and many other custom parts.

Nitrogen Storage Tank

I purchased the Superfilter on eBay and extracted the cryocooler. In order to test the device, I attached a small heatsink to the cooler’s cold tip, placed the tip into a household vacuum flask, and powered up the unit. After 30 minutes, I took the cryocooler out of the flask, and noticed a small amount of liquid air had collected at the bottom. Inspired by this success, I continued construction of a more complete liquid nitrogen generator. I already owned a 30-liter dewar (large vacuum flask) and fabricated an acrylic plate that would seal the top of the dewar while the cryocooler was also mounted to the plate with its heatsink hanging down into the neck of the flask. I also removed the cryocooler’s finned heatsink on its heat rejection area and replaced it with a liquid- cooling manifold. Liquid cooling lowered the heat rejection area temperature more effectively than forced air cooling, and this ultimately lead to higher system efficiency.

The Stirling-cycle cryocooler with custom heatsink and acrylic dewar-mounting plate. The mounting plate has a rubber gasket that forms a seal with the neck of the dewar.

The liquid nitrogen generator has two basic sections, the dewar with cryocooler, and the air processing equipment that creates dry nitrogen from atmospheric air. The dry nitrogen is fed into the dewar at just above atmospheric pressure where the cryocooler chills the nitrogen until it liquefies and drips off the heatsink. Surprisingly, most of project’s time budget was spent designing and building the equipment to produce dry nitrogen from air. There are some companies who make dry nitrogen supply devices, but even small units are meant for much higher throughput than what is needed by this liquid nitrogen generator. Each liter of liquid nitrogen requires about 700 liters of room temperature nitrogen gas. 700 liters per day is only 0.5 l/min, a very modest flow rate. One popular, but unnecessary use for relatively low-purity nitrogen is filling car tires. I tried to purchase such a machine, but the cost and flow rate were much higher than anticipated. Instead, I found a very small nitrogen separation membrane on eBay. It’s original use was unknown. The separation membrane is the actual component inside commercial nitrogen generators that perform gas separation. The membrane is formed into a large bundle of hundreds of 2mm dia tubes. Air is fed under high pressure into one end of the bundle. The tube walls are semi-permeable and allow oxygen, water vapor, carbon dioxide and other “fast” gasses to permeate relatively quickly. Nitrogen and heavier gasses do not permeate as quickly, so the concentration of nitrogen is much higher at the exit end of the tubes than it is at the input end. Higher purities of nitrogen can be achieved by restricting the flow rate through the tubes, thus allowing plenty of time for the unwanted gasses to permeate the tube walls and leave the system. The resulting nitrogen will contain trace amounts of argon and even smaller amounts of other noble gasses.

Air Processing Unit

Air processing unit with two aluminum desiccator chambers, two small black coalescing filters, two beige carbon filters, nitrogen separation membrane at lower right, and a flow meter with needle valve.

I also built a dessicator from aluminum cylinders filled with silica gel and plumbed this into to the system before the air reaches the separation membrane. These units are available commercially, and the one that I built is not particularly specialized. Separation membranes also exist for removing water, and this would be an improvement over silica gel dessicators, which require the gel to be dehydrated in an oven after it becomes saturated with water.

The liquid nitrogen generator has proved to be a reliable, but fairly slow method to produce small quantities of liquid nitrogen at home. The initial cool-down of the dewar takes about 12-18 hours, after which liquid nitrogen is produced at a net rate of 1 liter per day. The generator uses about 300 to 400 watts of electricity (includes the water chiller, which cycles on and off), so the energy cost for producing one liter of liquid nitrogen is about 8.5 KWh, or $1.10. This is substantially less expensive than having a thermos filled at a local welding supply store.

LN2 Generator overview

System overview showing a window air conditioner converted into a liquid cooler with liquid cooling lines running to the cryo- cooler which is mounted atop a Linde dewar. The additional refrigeration compressor mounted on top of the air conditioner unit compresses atmospheric air for liquid nitrogen production.

  • André Esteves

    The link to the book or documentation of superfilter is incorrect or badly malformed. correction, pls?

  • Ben Krasnow

    Andre, here is the correct link:

    It seems that one of the characters in the link has broken the HTML in this page. I’ll ask the editor about this. -Ben

    • Jacob Shiach

      Link updated. And page fixed.

  • Brett Kuehner

    Excellent article, thanks.

  • Jared Arand

    It seems like you have a lot of knowledge about liquid nitrogen and this type of engineering so let me see if you might have some ideas or info that could help me.

    They commonly use liquid nitrogen for overclocking in computers, however the way that they use it, is to place a pot with the base on the cpu, then pour in liquid nitrogen, and the liquid nitrogen expands into a gas and carries the heat away with it. This requires using a lot of liquid nitrogen for rather short periods of time and is not practical at all for any sort of continued use, they basically just use it for setting records. However, would it be possible to create a closed loop system that runs liquid nitrogen though metal lines to the cpu, and then back to some sort of compressor and heat exchanger?

    • Ben Krasnow

      Jared, it would certainly be possible to liquefy nitrogen at the same rate as it is boiled off an overclocked CPU, but the energy required to do this would be quite large. My LN2 generator uses about 400 watts to produce 1 liter of liquid nitrogen per day. If the CPU boils off 1 liter per 10 minutes (just a guess), my LN2 generator design, scaled up to task, would consume (6*24*400) = 57KW. Obviously larger LN2 generators would be quite a bit more efficient than mine, but the point remains that creating LN2 at such a high rate takes a lot of energy. A better design approach would be to couple a large cryocooler or just plain HCFC refrigerant-based cooler to the CPU. The two criteria are 1. How cold does the CPU need to be and 2. How much heat (in watts) need to be pumped. The warmer the CPU is allowed to be, the easier it is to pump large amounts of heat away from the CPU. If the CPU must be 77K (LN2 temp), pumping heat away from it takes a lot of energy.

      • Douglas Minton

        I am trying to find a way to keep LNG in it’s liquid state in a storage tank. I would be curious if your liquid nitrogen process could be adapted to this use, and what it may involve. Thanks for your time.

    • Dave

      Ok here’s an analogy..

      You’re asking someone, who’s teaching us how to raise a dinosaur, if this method is good enough to make fuel.

  • Jared Arand

    Well I was already planning to employ a phase change unit that utilizes standard refrigerants to give a -50c idle temperature and around 44-35c under full load.

    I want to cool the cpu and gpu, both overclocked. At factory clock rates i believe their combined tdp is around 275-300watts. I dunno if this translates directly into watts of heat energy but of course increasing their clock rates and raising voltages would increase the watts needing to be dissipated pretty rapidly. A factory build phase change designed to handle that kind of load is approximately 1600 dollars.

  • Ben Krasnow

    Building a liquid nitrogen closed-cycle refrigerator would probably cost much, much more than $1600. 300W is not much of a load for a small air conditioning unit. If you are trying to get this done on the cheap, I would suggest getting an old window air conditioner, and building a custom evaporator that sits on the CPU and expands the refrigerant over a very small area. You’ll need a vacuum pump to empty the lines, then you can refill it with r134 from an auto parts store if that is was the air conditioner used originally.

  • Kevin

    Do you have any recommendations on what supply stores to get liquid nitrogen?

    • Ben Krasnow

      Kevin, try welding supply stores. Most of them will fill a stainless steel vacuum flask for $20 to $30. Many will not fill a glass flask. Good luck!

  • Conner

    Finally something to do with that old air conditioner sitting in our basement. I’ve been thinking about doing something like this for my High School. Very nice article.


  • Dr Sheldon Cooper

    Can I use this to make ice cream (seriously).

    Dr Cooper

    • Ben Krasnow

      Yes. Search for liquid nitrogen ice cream. You’ll likely need to make the LN2 first, then use it to make the ice cream. The boiling nitrogen is what whips up the cream as it is freezing to create a pleasing texture.

  • sayan

    Hi Ben,
    i’m interesting your machine to product Ln2 that will support my home experiment to make a liquid bio gas, so if i need Liquid n2 40 lite per day what is the approximatley cost? Can you built for me?

  • Mike

    Coca Cola using compressed co2 Cardon Dioxide in all 2014 coolers . So you can have a cold drink. Wouldn’t want to be around the cooler when the pipe breaks at 5000 PSI .

  • michael

    Hey Ben;
    Quick inquiry…
    Where did you get your nitrogen separation membrane?
    Did you buy it, or build it?
    If you purchased it, where from?

    • Ben Krasnow

      Michael, I bought the membrane on eBay. It’s easier to find very large N2 separation membrane units, but I happened to find a very small one that was pulled from some unknown lab equipment. In fact, it was too small, and broken, but I managed to patch up a couple leaks and get it working well enough to produce some LN2. Search for N2 tire inflation stations. These all contain large separation membranes, and are probably the easiest way to find them.

      • michael

        Thank you for the helpful info,
        much appreciated.

  • Dylan

    is it noisy?

    • Jax

      It uses a compressor…

      Of course it is noisy….

  • Marco

    Hello Ben,
    congratulations for your work! I am trying to produce some “nitrogen enriched air”, so just the first block of your device. I am not able to find nitrogen separation membranes, can you help on this?
    thanks in advance


  • Anders

    Very interesting post, and I’d really like to do the same thing.
    I find it most unlikely to be able to copy this setup for anything remotely close to $500.

    Not to be pesimistic, but I have been looking for stirlign cryocooling units on eBay for a many years (on and off), and still havent found them. Not for the price mentioned here anyway. Can you please be more specific on how you found yours, and lead me to where I can find similar ones? I have found some on twinbirds website (only for sale in japan), but they don’w go nearly as far down in temperature as is needed here. And I expect these to be considerably more expensive as well.

    The stirling cooler unit is the key element of the whole thing here, and I would really like to know more on how to get one myself. All I could find in the post ablut it was “I purchased the Superfilter on eBay and extracted the cryocooler.”
    I’m sorry but that doesn’t tell me anything…. So I’d really appreciated if you could tell me more details here so I could be able to replicate this experiment.
    The rest of the article is wery well written with good enough details to duplicate what is done. But unfortunately not the part on how to get a stirling cryocooler.

  • sfsdf

    Maybe adding some insulation could increase the nitrogen production?