"Cooking" oxygen is possible!

This post is to update the process done so far.

At a time when inhaling with a high proportion of oxygen is becoming increasingly relevant for treating patients with covid, we have started exploring options that can be decentralized and enhance community capacity and know-how.

There are several DIY approaches, some of them listed in a previous thread here:


Chemical method

We are following a chemical process, while other teams are working on oxygen concentrators.


We managed to successfully generate oxygen as this video shows below.


We used a 20litre mineral water bottle, filled it with tap water, and displaced water for the released gas, which is definitely oxygen (with some expected gases and humidity). The following reaction is used;


which is Sodium Percarbonate releasing hydrogen peroxide, and


hydrogen peroxide in turn releasing oxygen. We roughly calculated that more than 20 liters of oxygen can be generated from 200gms of the reagent. We used KMnO_4 to increase the rate of reaction.

Next Challenge

The released oxygen to be compressed in either oxygen cylinders when available, or a large size pressure cookers (we still need to design how to do this).

This post is also to keep tinkering on this part of the puzzle, viz., how to compress and deliver oxygen to the patients.

Parallel Challenges

DIY methods of making nasal canula. For example, see the following post:

If any of you are interested in supporting this, please reply below. We need people to document, write safety procedures to be followed, alert us about possible precautions to be taken, etc.

Testing the quality of oxygen generated: we need help in this. Go and look out for sensors, beg, borrow or bring as soon as possible.

At this moment we are focussing on compressing the gas into cylinders or pressure cookers, based on availability.

We will post updates as soon as possible.


If a sick person is able to inhale as much as 5lpm, an empirical figure based on widespread reports from homes with oxygen concentrators in use in the emergency, while waiting for transfer to a full fledged hospital, the actual O² needed, at concentrations of 95% and above, in humidified condition, will be 10lpm. For safety, the device that delivers this needs to have a capability to supply about 15lpm, which is why the machines under development (eg the Houston Oxi-kit being completed by the various teams pulled together by the Maker’s Asylum initiative) are working out quite expensive (final figures awaited, but seems like tens of thousands of rupees). Reliable oil free pumps/compressors of that capacity are not cheap.

However, any system that can fill readymade pressure vessels, such as modified pressure cookers for storage, means that the production system and delivery system can be divorced, and operate at different scales. This could be an effective way of lowering the capital cost.

For the delivery system, if a valve can be devised that senses when the user is breathing out, and stops the oxygen flow for that period of time, the oxygen requirement will be roughly halved. This can also be a major area of cost efficiency.


120 ATM agriculture spray pumps are available online at Rs.900/- . These are diaphragm pumps and use two 12 v 4 A motors. I have been using one for a month without problems. A standard car battery works well as a supply source. If the grid is available a regular pc smps will provide power. O2 can be pumped into tyres, which can hold substantial volumes and at pressure. But will require pressure regulators before being released to a patient. Before use tyre can be cleaned by pumping soapy water in and out.

Two ball cocks in series can act as inlet for chemical charge. Open the top valve add chemical and close. Open the bottom to let it flow into chamber, then close. Ball valves are commonly used in agriculture. They are rated to work upto 80^oC.

One can automate the valve with a 10 rpm geared motor.

Rural areas have a fantastic variety of pipes, tubes, valves and accessories to hook them up in any imaginable configuration. Cities have a very limited choice in such items.


Googling long forgotten high school physics tells us that 20 Li of gas at STP (standard temperature and pressure ) compressed to 120 atm will occupy .16667 Li. Therefore a 20 Li can capable of withstanding 120atm will hold 2200 li when expanded to STP.

1 Like

which is your diaphragm pump please? For solar water pumps, my analysis indicated that a 100 watt 3 litre/minute can fill a 500 litre overhead water tank in 3 hours. 12 v 100 watt solar panel available for Rs 3500. I did find a suitable Shakti diaphragm pump but could not get it from Kochi or Coimbatore.

1 Like

How much pressure can a standard mineral water bottle withstand? how do we test?

Can we use pumps made for liquid for gases? If the pump is made liquid leak proof, can the same pump be gas leak proof as well?

Is this the kind of pump you have in mind?

1 Like

Dental compressors I was told are better. for example this one:

This is the same as the one I am using.

It is rated at 10bar.
A thumb test shows it to be gas leak proof too. Liquid pumps like this one dont use lubricant oil as the liquid acts as lubricant. With gas the diaphragm is guaranteed to wear out without lubrication. Spares are easily available in rural areas.

1 Like

A scuba tank of about that size costs $500, and is normally charged at 140 bar.

I don’t think that this scale of compression may be needed. Perhaps we should work backwards, say, to half an hour per tank. At the worst case 15 lpm, based on empirical reports from patients and caregivers, this is 450L, which in a 15L bottle means 1:30 compression. I should look up a table, but it is probably not very different from 30 bar. This study describes the strength of the smaller PET bottle, which can handle up to 9 bar. Given the higher mass of 15L of water, the container is probably much tougher, but this needs to be evaluated.

In any case, in our chat yesterday, we are ruling out the worst cases, given that such people may need much higher standards of care than only oxygen feed. There will still be many larger numbers of people who might avoid getting so ill if they get sufficient high concentration oxygen earlier, to compensate for the reduced conversion function caused by the virus. If we assume that the 15L bottle is no stronger than the smaller PET bottle, it will still hold 130 to 150L of oxygen concentrate, sufficient for about 15 minutes at the ‘heavy end’ of the spectrum, a flow rate of 10lpm.


Sealing the mouth would be difficult with only the cap. A soft silicon rubber bunge inside the neck should work.


Nowadays there are glass bottles with a neat bent wire lock that presses the bung into place, using the flange on the bottle as a latch.


Glass and compressed gas might be a dangerous combination.

1 Like

Some more updates:

  1. The gas that we are obtaining from this process is collected in balloons and we got the percent of oxygen in the gas, and we now know that is 91%. If we can eliminate the humidity from the gas, we may be able to reach much higher percentage. This number is obtained from both gas chromatography method (help from TIFR labs) as well as oxygen sensor (we used Envyteck by Honeywell OOM201), with help from Gorbochev Anthony.
  2. Next step we will get the gas checked for any poisonous gases like CO (carbon monoxide).
  3. Meanwhile, we are looking for designs for dispensing the chemical x amount for each t time. Please suggest methods.
  4. Seeking suggestions for humidity reduction methods
  5. Controlling the compressor speed to match with the production speed. Regulation of compressor speed, we are told would also reduce the power of pumping. So, we need a design that can decrease the speed without the reduction of the HP.
  6. The precipitate is known to have Sodium bicarbonate and Manganese oxide. Any ideas of how this can be dried and recycled. We don’t want this process to generate chemical waste and spoil our environment.

Any other issues if any of you think about, please reply and contribute to the thread.

1 Like

I think the idea is to try and use readymade 15L plastic bottles.

They need to be pressure tested, of course, but an attempt to locate readymade data on standards yielded only some information about 1.5L PET bottles. Still, they may well be able to withstand about 10 bar, allowing enough compressed gas to last about half an hour, for somebody who is not very seriously ill, and can get better with medication and enough oxygen to help keep blood saturation levels from falling dangerously low.

I think it makes sense to mention here that this process, unlike the zeolite concentrators, could lead to a grassroots solution for many oxygen related medical conditions, whose value remains after the pandemic, especially if hand or foot operated pumps can manage 10 bar, freeing rural medical specialists from dependence on external electricity. The details of such conditions are listed in the WHO guidelines on oxygen as a basic medical necessity.

1 Like

One very simple, readily available, metered dispenser is an injection syringe. The ones that are about the size of a finger handle 5ml of liquid, and if used without a syringe can deliver the equivalent amount of powder very easily. The density of the two powders needs to be measured regularly, as the actual density will probably vary considerably from the standard as the monsoons continue to approach.

The working end of the dispenser is a nicely soft tipped hollow plastic a couple of mm in diameter. It should be very simple to rig a self sealing valve/orifice on the top of the pressure cooker, in place of the safety valve, in order to inject the ingredients.


The usual method is to use a vibrating incline cone. The incline is kept to a minimum . The cone is vibrated with a motor having an eccentric fly wheel. We could easily use the vibrating disc motor to move the powder or granule. Vary the speed and you can control the flow rate. Additionally we have many patterns with @Ashish_Pardeshi controller.