An alternative approach to CO2?

[Yugang]

CO2 in the planted tank is challenging for many hobbyists, and it would be nice to have more innovations to manage CO2. How about a technology that gives us just a remote control to set CO2, then forget about it? How about some serious CO2 savings, and unprecedented stability?



Two years ago I researched new MEMS technology, sensors for CO2 detection through Photoacoustic spectroscopy. Essentially a full physics experiment integrated on a chip for very affordable costs. We could apply Henry's law, and calculate that we need less than 1% CO2 concentration in the air above the tank, have that controlled by our sensor, and set the CO2 in the tank water at typical values used in our planted tank hobby.

Unfortunately the idea has not been picked up by a pioneer hobbyist who could build and test it. I believe we could create a breakthrough for the hobby, but lack the electronics skills and the time to take it further. I am posting here again, and hope someone will understand the concept and see an opportunity to create progress.

I hope the mods can accept the following link, as I currently lack the time to do a full write up here.

An alternative approach to CO2?

Inspired by the thread on CO2 controllers, and @Art experiment with modified pH probe / drop checker, I came up with what is to the best of my knowledge a new approach for our hobby. I am not strong on the chemistry of CO2 in water, so it would be great if others chime in and comment if this...

scapecrunch.com

 

[Marcel G]

Hi Yugang,

It's great that you're so enthusiastic about further innovations! I personally appreciate it very much. I wish there were more aquarists like you. Unfortunately, I'm not the person who can help you with this, but I'll keep my fingers crossed that you find someone who can and that you manage to test it successfully in practice. I just wanted to ask what you see as the advantage of this solution over your previous "CO2 spray bar" project (https://scapecrunch.com/threads/co2-spray-bar-a-summary.1009/)? I use a slightly modified version (as a "bell-type CO2 diffuser") of your previous project in my experimental aquariums and I am extremely satisfied with it. I also find this method very simple, reliable, stable, safe, cheap, and non-overdosable.

Marcel

 

[Yugang]

I just wanted to ask what you see as the advantage of this solution over your previous "CO2 spray bar" project

The short answer is that I believe we should explore as many directions as possible, build experience over time and let the best solutions win. I believe especially for large tanks, advanced applications, this solution could be the winner.

For this particular solution I would see a couple of benefits:

• Controlled by electronics, CO2 can easily be set and adjusted by the user. Manufacturers could offer it as a user friendly integrated solution with the tank.
• Very stable, as the sensor accurately sets the CO2 concentration in the air (less than 1%), and concentration in water follows from Henry's law.
• As we measure CO2 in the air, it is independent of water chemistry. This is the benefit/functionality of professional CO2 meters that currently cost several thousands USD.
• Potentially big savings of CO2 consumption as we work with a semi-closed system. I imagine probably at least 90% savings.
• We can save on the CO2 regulator, just need the electronics controller to release puffs of CO2 via the solenoid. The CO2 sensor is cheap. System cost can be lower than good CO2 setup today.

I do believe some development work would be needed for the final solution

• We don't want water vapour condensation on the sensor or electronics. Probably need to operate them at slightly elevated temperatures.
• We want the system to be as much closed as possible to save on CO2, but still some decent gas exchange for the plants and livestock. We need to find the optimum between competing requirements. I have operated nearly perfectly closed aquarium for months, don't expect problems with this but it is a factor that needs to be considered.

In summary, we have an idea based on simple physics, and affordable technology to experiment with it. Time will tell if it will be a winner.

 

[Andy Pierce]

I do like the idea. Key is having a tank with a lid so the whole water surface becomes the 'spray bar'. This is how the bell-type CO2 diffuser works... essentially as an external tank with a lid.

 

[Yugang]

I do like the idea. Key is having a tank with a lid so the whole water surface becomes the 'spray bar'. This is how the bell-type CO2 diffuser works... essentially as an external tank with a lid.

If I may use another analogy (I hope it helps) - One could compare the conventional method of injecting CO2 while relying on surface agitation for the stabilisation with heating a home in the winter time, with the windows wide open.

Wasting 95-100% of CO2 to stabilise the tank is not very efficient, nor is it easy to have CO2 homogeneous everywhere in the tank.

Let's "close the windows", have a decent "thermostat" and have the owner set a comfortable temperature without unnecessary waste or instability. That is the essence of this system that I see as compared to our current approach of stabilising CO2 in the tank.

 

[jaisol]

Looks like one of the many "sensors" you can get from china, a glance at ELECFREAKS Octopus CO2 Gas Sensor (MG811) gave me the impression it's analogue. An ESP32 would probably be the cheapest easiest way to go unless you have ardino, pie, or some other micro controller around. Without the sensor, I'd say you could do the electronics for under £20, or nothing if you have a 5v psu and a micro controller laying around, then ask chat gpt for code, confirm chatgpt with co pilot, assuming you have compiled programs for your micro controller in the past, then 1/2 a days work, after check it with a variable voltage before you buy the co2 sensor. That will give you a proof of concept, it will be the testing, reliability, air sampling, and making a semi commercial prototype that will consume money and time. If no-one is interested I'll make you a prototype but it will be mid Feb before I have time to do it. mmm 1. monitor co2 level. 2. turn co2 on /off, 3. open windows. 4. ? monitor temp? ... all easily do-able but Chinese themo-couples tend to be temperamental .. might be Boyles law I'm thinking of..

I only glanced at the data sheet, but one thing that did make me think that this may not be ideal was that the sensor runs at an elevated temp, and if it is not housed properly when not in use may take 48 hours to become stable. those to me rings alarm bells.

 

[Yugang]

@jaisol , later in the thread I posted the Infineon sensor, does that look good to you?

 

[jaisol]

It's late & been watching films, the actual sensors look very similar from the photos, one is analogue the other is I2C bus, the second I2C one sounds like it may be "mapped" - the sensor is tested and the results are then used to calibrate readings, the basic way the sensor runs maybe the same and that may be where your problems start... one would have to look at the cost. The analogue one is around £25, the other, I2C is easier to work with and mapped what is the cost and stability? The sensor sounds a lot like the sensor that is used in ph probes, in that it needs looking after, checking and calibrating, when I say sensor I mean the part that does the detecting not any electronics bits after it. I2C is very easy to interface to a micro controller. I will be busy until end of jan, then can give it a week or two, what you need to know is the probe stability, I mean the actual bit that measures it, not the bits in between, the actual probe - silver / copper even exotic wires, £25 probe says iron to me ... just a guess. Ohms law. If no one want to bite I'll give you a working proof mid feb
Hope everyone has a great new year

 

[Yugang]

Thank you so much for bringing in your expertise @jaisol, much appreciated.

In 2025/2026 I am very busy, hardly have any time for the hobby. I am so grateful for several other hobbyists supporting the Yugang reactor, because I cannot. For this idea, I can help a bit as I've done more experiments with semi-closed systems, and have some other ideas that I have not posted yet. But I cannot build anything, or do testing in my tank.

I am posting this thread, as I feel it may be a missed opportunity for the hobby when nobody has a try with it. If you, perhaps with some others, are willing to set up some collaborative project to try it out, I can bring in some physics expertise, but not much more I am afraid.

It is quite unfortunate that the manufacturers seem so passive, and it entirely depends on hobbyists to push new ideas forward.

 

[Aquahorti]

Just to save you some time I will give you a link to an article about a sensor of this type, where they describe what they did and some of the things you have to adjust for. Granted I do not know if the sensor you are looking at will have the same problems with the concentration of water in the gas samples as the setup they have used: (Redirecting), but it might.

If your sensor have the same problems as the setup described in the article I would suggest a much simpler setup (although I am of the school where you always look at each thing you add to your setup and ask yourself if you are adding this to try and compensate for a problem rather than eliminating the problem). My suggestion would be using 3 modified drop checkers with different KH solutions in them and have 3 light sources and 3 photodetectors control the flow of CO2. By tweaking the KH solution in the drop checkers you should in theory be able to get the CO2 to be in the range you want it too. Granted you will not get a RT readout of the CO2 in PPM with what I suggest but I do not see what good that will be to anyone outside research applications.
Both setups are depending on the transfer rate of the CO2 from the water to an atmosphere, so in theory you will end up with the same response times for both setups (you will need the drop checkers to have the gas go through the solutions or in some other way increase the reaction speed with the KH solution). In the end it will be a cost and size issue.
One last suggestion, If you make this try and make it an inline sensor and not something that needs to go on top of the aquarium.

 

[hypnogogia]

sets the CO2 concentration in the air (less than 1%), and concentration in water follows from Henry's law.

Would this limit the dimensions of the tank in so far as that the surface area would need to be a certain ratio to the volume of water in the tank? Would it work equally effectively with a deep tank as with a shallow tank?

 

[Yugang]

Granted I do not know if the sensor you are looking at will have the same problems with the concentration of water in the gas samples as the setup they have used

The Infineon sensor has a compensation mechanism for humidity, and is specified until 95% RH. This is the graph from the application note:


Indeed, it would be good to test how the sensor performs above 95% RH, non condensing. It for sure would be good to operate the sensor at slightly elevated temperature (maximum 50 degrees in spec), to avoid any condensation in an aquarium environment.

Both setups are depending on the transfer rate of the CO2 from the water to an atmosphere

This is a misunderstanding. In my concept we measure and control the air above the tank water, so that it has a stable (<1%) concentration of CO2 in the air. As we know the relationship between partial pressure in the gas above the tank and the water in the tank (Henry's law), we know then what the CO2 ppm in the tank water will be.

Would this limit the dimensions of the tank in so far as that the surface area would need to be a certain ratio to the volume of water in the tank? Would it work equally effectively with a deep tank as with a shallow tank?

No, the tank dept is not relevant here. Following Henry's law, the partial pressure of CO2 in the water will be in equilibrium with the partial CO2 pressure in the air. As there is no way for CO2 to "escape" from the water, other than plant consumption, there will be very little nett gas exchange at the water/air interface and furthermore we expect extremely good homogeneous distribution of CO2 through the tank water (as there are virtually no CO2 nett transports involved). We create a much more stable situation than with traditional open-top CO2 injected tanks.

 

[Yugang]

May I ask some help from the various experts in water chemistry on this forum? I am really happy that both on UKAPS and Scapecrunch there are volunteers to try this out, and perhaps the best would be if we could all collaborate as a team.

I would appreciate if someone can double check my calculations that a range of 0.5-1% CO2 in atmosphere would yield via Henry’s law a good testing range CO2 ppm in the planted tank. It is good to have some peer review, as I would hate that we build the hardware and then discover a calulation mistake in the CO2 targets.

 

[PeerUnk]

Hi all,

please note using sensors in air to measure CO2 and oxygen with an esp32 is already done by @hax47 and copied by me. We used it to measure a gas pocket, like a digital drop checker. The thread is on this platform: gas exchange experiments
Hax47 disclosed his software on github including a guide to build the entire hardware device: github link hax47

@Yugang came up with this interesting idea, so I’m willing to change my device to monitor a airtight lid. I will reply later on with some progress I made.

Cheers!

Somewhat offtopic, but I extended the ESP32 software of Hax47, by adding a LOT of other configurable features, like
a webserver, display, calculation of running averages, using even more accurate physics calculations, able to trigger a relay, supporting sensors like light sensitivity, pH, conductivity, relative humidity, air pressure, etc.
I did not share it as open source yet, still a private project.

 

[PeerUnk]

I would appreciate if someone can double check my calculations that a range of 0.5-1% CO2 in atmosphere would yield via Henry’s law a good testing range CO2 ppm in the planted tank. It is good to have some peer review, as I would hate that we build the hardware and then discover a calulation mistake in the CO2 targets.

I can check your target, but have to go now. Please read gas exchange experiments, the answer is in there, I'm sure.

 

[PeerUnk]

I would appreciate if someone can double check my calculations that a range of 0.5-1% CO2 in atmosphere would yield via Henry’s law a good testing range CO2 ppm in the planted tank. It is good to have some peer review, as I would hate that we build the hardware and then discover a calulation mistake in the CO2 targets.

I've entered your 0.5% as CO2 in air, which will diffuse in water to 7.5 ppm or mg/L @ 25 degrees Celsius. Seems like a reasonable target.

For others to check:

 

[PeerUnk]

Hi all,

When creating a air tight setup, the most important factor for all (aquatic) life is oxygen! This might make or break the approach to be tested... Think about what happens when oxygen consumption would be very high for any reason when lights are off or nutrients are depleted so oxygen production in plants comes to a halt for days!

Therefore I will include a oxygen sensor as well.
The target I set would be 20.95% (like we daily breath in) which will equilibrate into 25*C water to 8.71 mg/L. ChatGPT comes up with ≥ 6 mg/L to support nearly all fish, invertebrates, and plants and widely used as a safe management target (so atmospheric lowest threshold would be 15%)

A CO2 sensor module like the SCD41 from aliexpress is 25 euros. But the oxygen sensor module (SEN322) I'm using comes at 50 - 90 euros.

Cheers!

 

[hax47]

Hi!
Interesting idea, but let me throw in my two cents.

I think that sealing an aquarium truly airtight is very challenging. I have been trying to make an airtight container to take a water sample, equilibrate it with air, and then measure the CO2 concentration. I have yet to find a genuinely airtight container; they all leak slowly. In the case of an aquarium, you also have inlets and outlets, and it is very hard to seal all of them perfectly.

CO2 diffusion is much faster in air than in water (on the order of 10 000×), so leaking into the air through imperfect seals might actually be faster than dissolving into the water through the water surface. Just a guess, but it might be that CO2 loss through a typical aquarium lid from the headspace is faster than through the water surface in an open aquarium.

No, the tank dept is not relevant here. Following Henry's law, the partial pressure of CO2 in the water will be in equilibrium with the partial CO2 pressure in the air. As there is no way for CO2 to "escape" from the water, other than plant consumption, there will be very little nett gas exchange at the water/air interface and furthermore we expect extremely good homogeneous distribution of CO2 through the tank water (as there are virtually no CO2 nett transports involved). We create a much more stable situation than with traditional open-top CO2 injected tanks.

If you look at this plot from here, the CO2 drop during the lighting period (indicated by yellow background stripes) is quite significant compared to the dark period, which means that the diffusion rate will matter unless you are OK with dropping CO2 levels during the light period. I am OK with that, I have always had this drop in my CO2-injected tanks, but I also start from higher levels (30–40 ppm).
The rate at which CO2 can enter or leave the water depends on the gas–water surface area. So while Henry’s law tells you the equilibrium, the tank geometry (surface area/volume) influences how fast you approach that equilibrium, or how fast you can replenish the used-up CO2.

The diffusion rate is also affected by the partial pressures of the CO2, so compared to 100% CO2 (1 atm partial pressure) in a bell diffuser or CO2 bubbles from a regular diffuser, in the case of the 1% CO2 (0.01 atm partial pressure, resulting in about 15 mg/l CO2 in water in equilibrium), the diffusion rate will be much slower. Lets say, we have a partial pressure of 0.005 atm in the water (gives about 7.5 mg/l CO2 concentration), we have 1-0.005=0.995 atm partial pressure difference (driving force for diffusion according to Fick's law) in the case of the 100% CO2, but only 0.01 - 0.005 = 0.005 atm in the case of 1% CO2 in the headspace. That is almost 200x difference. If the water CO2 level is 11.25 mg/l, the driving force will be already 400x smaller in the case of 1% CO2. You can compensate with a larger surface area (another parameter in Fick's law) and stronger surface agitation to some point, but I think that this low diffusion drive could be a serious limitation.
Also you might want to keep higher CO2 levels in the headspace to compensate for the slower diffusion rate (and variable plant consumption rate), maybe higher during the day, and lower during the night, but probably you would need to calibrate the right levels for each individual aquarium (each with unique shape, surface agitation and plant density). You will most likely never reach a true equilibrium, and calculating with the dynamic aspects might be challenging.

When creating a air tight setup, the most important factor for all (aquatic) life is oxygen! This might make or break the approach to be tested... Think about what happens when oxygen consumption would be very high for any reason when lights are off or nutrients are depleted so oxygen production in plants comes to a halt for days!

In light of the above, I do not really see oxygen as the main limitation here, as long as the system is not perfectly airtight (which is likely). The O2 partial pressure in ambient air is about 0.21 atm. Even if it drops to 0.14 atm in the headspace and water (which still gives around 6 mg/L O2 in equilibrium), the driving force for diffusion from the room air is 0.07 atm. That is more than 10–20× higher than the CO2 partial-pressure differences I used in the examples above. So oxygen will leak back in much more readily than CO2 is retained.

 

[Yugang]

These are all very relevant remarks @hax47, thank you. I can't say that I disagree to any of your observations, qualitatively but the challenge is how we quantify these. At least for me, gas transport processes are very hard to estimate, or model, and I believe the best thing is to build and test in the real world rather than trying to calculate.

A couple of remarks, on your post:

I think that sealing an aquarium truly airtight is very challenging.

This is also my main concern. In the worst case, as you also say, the leaking may actually give a higher CO2 consumption than injecting in the water and an open tank top. I believe very hard to quantify without testing it. I do have some hope, or expectation that it is actually working out:

• I have experimented with nearly perfectly closed lids for several months, and saw no negative effect on the tank whatsoever. I could verify that the tank was nearly gas-tight, as evaporation was reduced by more than 95%, which may be a proxy of gas transport (water vapour) and proxy for CO2 transport.
• The CO2 concentration in the "air" above the tank, in the semi-closed system, is less than 1%. So there may be some gas transport, but only <1% of that is CO2.

The rate at which CO2 can enter or leave the water depends on the gas–water surface area.

This is true, but when testing prototypes of horizontal Yugang reactor I have always been surprised how fast CO2 (pure in that case) gets absorbed into water. Also here, I believe it is hard to quantify (at least for me), and probably the best would be to just test it out.

Also you might want to keep higher CO2 levels in the headspace to compensate for the slower diffusion rate (and variable plant consumption rate)

Agree, qualitatively at least. I have done numerical modelling in the past, using the dip in pH as a starting point and calculating the plant CO2 uptake from that. At least from my estimations at the time, I believe that plant consumption is a very small part of the total CO2 injection. Also here, I find it hard to quantify whether the plant uptake will be significant as compared to CO2 transport from the gas pocket into the water. Again, I believe we only know when we test for it.

In summary @hax47 I am in agreement with all your observations, but would you concur that only a test would point out if in practice these are significant impacts, or would even lead us to the conclusion that the concept won't work very well in practice? Overall it looks like the extent that we can perfectly seal the tank is going to make or break the concept. Perfect sealing, it will almost certainly work. Imperfect sealing may lead to trouble and too much CO2 consumption, but the key question seems to be how much perfectly the sealing needs to be.

 

[hax47]

In summary @hax47 I am in agreement with all your observations, but would you concur that only a test would point out if in practice these are significant impacts, or would even lead us to the conclusion that the concept won't work very well in practice? Overall it looks like the extent that we can perfectly seal the tank is going to make or break the concept. Perfect sealing, it will almost certainly work. Imperfect sealing may lead to trouble and too much CO2 consumption, but the key question seems to be how much perfectly the sealing needs to be.

I agree that the best approach is to test things directly — that is exactly why I started running gas-exchange experiments myself, to better understand the different processes involved. But you will need a strategy to handle the challenges this approach reveals. The main one is the leakage, I agree, but I also think that plant consumption is significant (you can also check this plot), and you will need to address the orders of magnitude difference in the diffusion rate compared to the horizontal reactor, for example. We do not need experiments to identify what controls the diffusion rate here; Fick’s law already tells us the relevant parameters.

One option I have been thinking about before, although for the reverse diffusion — although originally for the opposite direction (letting CO₂ from the water equilibrate into the headspace for measurement)— is to use an air pump and diffuser that takes air from the headspace and injects it into the water, therefore increasing the surface for gas exchange. My issue was that I could not find an air pump that performs reliably in a high-humidity space under a tank lid.

Maybe you could also inject 100% CO2 into the water instead of the headspace to have more efficient diffusion into the water (exploiting a higher pressure difference between the bubbles and water, and a larger surface for efficient delivery of CO2), and still measure the CO2 in the headspace. Combined with the air circulation with an airpump between headspace and water, maybe you get the efficiency you need. But maybe you will be underestimating the CO2 levels in the water this way if you measure it in the headspace.
What I also know from my experiments, that a closed lid (sealed with duct tape) reduced the dissipation rate of CO2 many times. So yes, the sealed lid itself would decrease the CO2 dissipation by many folds, but you still need a good delivery of CO2 into the water.