-
You are viewing the forum as a Guest, please login (you can use your Facebook, Twitter, Google or Microsoft account to login) or register using this link: Log in or Sign Up
Gas-exchange experiments
- Thread starter hax47
- Start date
Interesting how many things can affect the water parameters. I suspected that ambient lighting could affect the plants even when the aquarium lights are off; I did not think that that could be a measurable effect. Also, the one time I measured the CO2 outside before and after sunset, I saw some differences, but that was a very small change.
Nice work there. I can not follow the math part, just the calculation's start and end (the final equation). I guess one could substitute CO2 injection with CO2 production for low-tech aquariums...
I was planning it backwards, once I have the k, I could calculate the CO2 production rate p as k * c, since, in a steady state, the CO2 production rate equals the CO2 loss rate. Determining k from a steady-state situation seems more elegant, though I guess it is hard to calculate how much CO2 is actually dissolved into water. That should be the p, right, not the total CO2 injected? Maybe that uncertainty could be solved with a CO2 bell or measuring cylinder turned upside down. That way, all the disappeared CO2 is dissolved for sure, and the rate could be calculated too. I don't have an aquarium CO2 system, but I have one for brewing; I could do a cylinder experiment and compare the k values with the two methods. The ks calculated with the two methods should be the same, right?
Even though I wrote the article and would love to experiment finding c (calculated CO2 concentration) and p (calculated CO2 per second) for my tank, I think in my case, spending time fiddling with spray bars, bubble counters and drop checkers is time far more productive. After 12 years of CO2 injection, and 35 fire extinguishers usage, I find green drop checker when lights come on is the way to go, to get wonderful plants and complete lack of algae.
I have just changed my bubble counter, cleaned all the piping and thoroughly cleaned my diffuser and of course CO2 injection rate all changed. Just turned down lights 2 tubes only and shorter duration, and kept popping back to tank every couple of hours to increase "bubble rate" (as started at extremely low) and now after a couple of days I get a green heading to yellow drop checker at lights on. Will leave a day or two to confirm bubble rate is stable before increasing light duration and switching on all 4 tubes.
I have just changed my bubble counter, cleaned all the piping and thoroughly cleaned my diffuser and of course CO2 injection rate all changed. Just turned down lights 2 tubes only and shorter duration, and kept popping back to tank every couple of hours to increase "bubble rate" (as started at extremely low) and now after a couple of days I get a green heading to yellow drop checker at lights on. Will leave a day or two to confirm bubble rate is stable before increasing light duration and switching on all 4 tubes.
I think CO2 dosing is quite straightforward as it will directly influence CO2 without depleting O2. I would like to optimize the CO2 in low-tech tanks, where there are a lot of unknown variables. Your equation might still come handy as it can also be used with CO2 production.
I did some calculations based on the idea that the lids would influence the gas concentration in the aquarium. I think I can easily set up experiments to test the CO2 kinetics in this model system. For the O2 part, I would need to convince myself and my wife that I really need that O2 meter...
So here is the model:
Let's take two systems; in one, we have 120 liters of water; in the other, 100 liters of water + 20 l headspace. For the sake of simpler calculation, I did not calculate with any gas exchange, as if the aquariums were sealed air-tight. Theoretically, the k diffusion values could be adjusted in both models to reach the below steady states, in which the produced CO2 would equal the removed CO2. I will not complicate the model with kinetics just yet.
I used the following constants, calculated with 25°C temperature:
Gas solubilities in water:
CO2_solubility = 1449 mg/l at 1 atm
O2_solubility = 40 mg/l at 1 atm
Gas densities in the air:
CO2_density = 1780 mg/l at 1 atm
O2_density = 1.291 mg/l at 1 atm
In an air/water system, the gases are distributed between the two compartments. In equilibrium, the partial pressures of the two compartments should be the same, so the concentration ratios between the air and the water are:
concentration_ratio = gas_density / gas_solubility
The ratios of the gas amounts (since we have different water and gas volumes here) for any gas should be:
mass_ratio = (gas_density * headspace_volume) / (gas_solubility * water_volume)
That is: mass_ratio = concentration_ratio * headspace_volume/water_volume
With all this above, we can calculate the actual concentrations with given gas amounts, both in the water and the headspace. I uploaded the python code for calculations to github if someone is interested, but I think the following is logical even without knowing the exact calculations. So, let's see how the gas concentrations look like if we have the two systems equilibrated with atmospheric air:
As we can see, there is much more O2 in the system with headspace; the headspace functions as a reservoir for O2. Now let's put some biomass into the systems which use O2 and produce CO2. The ratio of produced CO2 and used-up O2 is described with the respiratory quotient (RQ). I don't know what the RQ should be in an aquarium system, but if carbohydrates are burned into energy + CO2, the ratio will be 1. In the case of burning proteins (e.g. from fish food), this will be ~0.8. I don't know the RQ values for plants and microbes, they depend on their energy source and type of metabolism, but I calculated simply with an RQ of 1. However, whatever it is, we should get similar differences between the two systems, except that the absolute slopes would be different (see the slopes later).
So let's assume the systems are closed, and the biomass uses O2 to produce CO2. After we used up 700 mg O2 (and produced the same molar amounts of CO2), this will be the new state of the two systems:
As we can see, the CO2 concentration in the water is slightly lower in the system with headspace, but the O2 concentration is much higher because of the reservoir. The no-lid system has no reservoir; all the used-up O2 is depleted from the water.
We can do the same calculations for a range of used O2 for both systems and check the O2/CO2 levels:
As we can see, we can go much higher with CO2 accumulation with a smaller O2 drop in the headspace system. These closed model systems are different from real-world aquariums in a couple of things, though:
1. Aquariums, even with a lid, are not closed systems; there is always some gas exchange between water and air. Therefore there will be CO2 loss and O2 resupplying. But as I mentioned above, theoretically, one could control the headspace/air gas exchange (ventilation) to adjust the right levels. I have no idea, though, how much the ventilation is realistically restricted with a lid. But I could check that in my aquariums or using model experiments.
2. The CO2 and O2 diffusion rates are different, so the gas exchange rate will not be the same for the two gasses. This would probably slightly modify the CO2/O2 ratios in the water in a dynamic system.
3. There is photosynthesis in planted aquariums, which also removes CO2 while producing O2
4. RQ could be very different depending on O2 availability
Nevertheless, I think that we can still conclude that adding headspace volume while limiting the ventilation (between the headspace and surrounding air) could help accumulate the CO2 in the water with less O2 depletion as in a no-lid system. I'll run some experiments to see how much the lid matters and how much ventilation restriction is needed to have a significantly higher CO2 concentration in a model system.
So here is the model:
Let's take two systems; in one, we have 120 liters of water; in the other, 100 liters of water + 20 l headspace. For the sake of simpler calculation, I did not calculate with any gas exchange, as if the aquariums were sealed air-tight. Theoretically, the k diffusion values could be adjusted in both models to reach the below steady states, in which the produced CO2 would equal the removed CO2. I will not complicate the model with kinetics just yet.
I used the following constants, calculated with 25°C temperature:
Gas solubilities in water:
CO2_solubility = 1449 mg/l at 1 atm
O2_solubility = 40 mg/l at 1 atm
Gas densities in the air:
CO2_density = 1780 mg/l at 1 atm
O2_density = 1.291 mg/l at 1 atm
In an air/water system, the gases are distributed between the two compartments. In equilibrium, the partial pressures of the two compartments should be the same, so the concentration ratios between the air and the water are:
concentration_ratio = gas_density / gas_solubility
The ratios of the gas amounts (since we have different water and gas volumes here) for any gas should be:
mass_ratio = (gas_density * headspace_volume) / (gas_solubility * water_volume)
That is: mass_ratio = concentration_ratio * headspace_volume/water_volume
With all this above, we can calculate the actual concentrations with given gas amounts, both in the water and the headspace. I uploaded the python code for calculations to github if someone is interested, but I think the following is logical even without knowing the exact calculations. So, let's see how the gas concentrations look like if we have the two systems equilibrated with atmospheric air:
As we can see, there is much more O2 in the system with headspace; the headspace functions as a reservoir for O2. Now let's put some biomass into the systems which use O2 and produce CO2. The ratio of produced CO2 and used-up O2 is described with the respiratory quotient (RQ). I don't know what the RQ should be in an aquarium system, but if carbohydrates are burned into energy + CO2, the ratio will be 1. In the case of burning proteins (e.g. from fish food), this will be ~0.8. I don't know the RQ values for plants and microbes, they depend on their energy source and type of metabolism, but I calculated simply with an RQ of 1. However, whatever it is, we should get similar differences between the two systems, except that the absolute slopes would be different (see the slopes later).
So let's assume the systems are closed, and the biomass uses O2 to produce CO2. After we used up 700 mg O2 (and produced the same molar amounts of CO2), this will be the new state of the two systems:
As we can see, the CO2 concentration in the water is slightly lower in the system with headspace, but the O2 concentration is much higher because of the reservoir. The no-lid system has no reservoir; all the used-up O2 is depleted from the water.
We can do the same calculations for a range of used O2 for both systems and check the O2/CO2 levels:
As we can see, we can go much higher with CO2 accumulation with a smaller O2 drop in the headspace system. These closed model systems are different from real-world aquariums in a couple of things, though:
1. Aquariums, even with a lid, are not closed systems; there is always some gas exchange between water and air. Therefore there will be CO2 loss and O2 resupplying. But as I mentioned above, theoretically, one could control the headspace/air gas exchange (ventilation) to adjust the right levels. I have no idea, though, how much the ventilation is realistically restricted with a lid. But I could check that in my aquariums or using model experiments.
2. The CO2 and O2 diffusion rates are different, so the gas exchange rate will not be the same for the two gasses. This would probably slightly modify the CO2/O2 ratios in the water in a dynamic system.
3. There is photosynthesis in planted aquariums, which also removes CO2 while producing O2
4. RQ could be very different depending on O2 availability
Nevertheless, I think that we can still conclude that adding headspace volume while limiting the ventilation (between the headspace and surrounding air) could help accumulate the CO2 in the water with less O2 depletion as in a no-lid system. I'll run some experiments to see how much the lid matters and how much ventilation restriction is needed to have a significantly higher CO2 concentration in a model system.
Most heavily planted CO2 injected tanks will net O2 producers, and will become fully saturated with DO during the photo period so there should be no need to draw air into your headspace as it should be fairly oxygen rich.
That is true, but I am thinking about non-injected low-tech tanks. I wonder if I could prevent the produced CO2 (or introduced during the water change) from escaping while providing enough O2. That would be the purpose of the headspace. If more O2 is produced than CO2 used over days, the better; less aeration would be required for the headspace.
It has been some time since I last posted new experiments. I conducted a set of measurements in August but decided they needed to be repeated. I was unsure if some differences fell within the normal variation range so that they could be merely random. However, I don't see when I could repeat these in the foreseeable future (as one measurement takes almost a full day), so I decided to share the results; some consistent tendencies in the results seem present across different setups (e.g., powerhead vs. airstone).
There were two questions I aimed to answer with these experiments:
This time, I used a 120-liter aquarium with sliding tops to make the experiment more similar to a real aquarium setup. I filled it with 100 liters of water and left about 20 liters of headspace. A hole was drilled through one of the two parts of the sliding lid, and a tube was inserted to allow sampling without opening the top. Opening the cover could lead to significant gas exchange between the air and the headspace, and I wanted to avoid that:
Sampling was done multiple times for at least 400 minutes, and I calculated the diffusion capacity (k) values using the method presented in previous posts. The k value describes how robust the gas exchange is between the aquarium water and the ambient air, i.e., how quickly CO2 dissipates. The larger this value is, the faster the gas exchange occurs. Gas exchange was facilitated in two ways: with a powerhead pump with directed flow toward the water surface or with an airstone at the aquarium's bottom.
The powerhead output was 800 liters/hour, and the air pump output was 480 liters/hour. Although it would be ideal to ensure the same water agitation at the surface for the comparison, this is difficult to control. Judging from the following gifs, my guess is that surface movement was less pronounced in the case of the airstone:
Powerhead:
Airstone:
I filled the aquarium with tap water (13 dKH). Instead of using fresh tap water each time for CO2 supply, I chose to add CO2 with sparkling mineral water, 500 ml each time. After adding 500 mls, the typical pH levels were between 7.30 and 7.40, corresponding to CO2 levels between 33 and 26 ppm. With both aeration methods, I measured the k value in three conditions:
As we can see, with these setups, aeration with airstone was much more effective at removing CO2 than surface agitation with a powerhead, despite water movement being stronger in the latter case. Additionally, closing the lid alone decreased CO2 exchange. When making the seal complete, CO2 dissipation almost came to a halt. Since CO2 dissipation with a given k equals:
I = k * (CO2water - CO2air)
and in a steady state situation, CO2 production (or injection) into the tank equals CO2 dissipation from the tank; there is an inverse linear correlation between the k value and the water-air CO2 difference. This means that if the k value is halved, the CO2 difference must double. So, manipulating the k value, such as using the aquarium lid, should directly influence aquarium CO2 levels.
Interestingly, in the case of airstone aeration, there is no such strong correlation between covering the top of the aquarium and CO2 levels. I think this is expected since the pump draws air from outside the aquarium, so the system is always supplied with fresh air, unlike the system with a powerhead only. This makes CO2 removal much more effective and more or less independent of the cover.
In conclusion, I think the take-home message is that the aquarium lid in this experiment decreased CO2 dissipation. Thus, it should increase the CO2 levels in aquariums, especially if no airstone aeration is used. The actual effectiveness depends on the lid's sealing potential, though. The other conclusion is that airstones promote gas exchange more effectively than surface agitation. Neither of these two conclusions is groundbreaking or probably not even new. Still, I find it challenging to find a lot of reliable information about these online, especially about the effect of the lids. In a low-tech setup, the lid effect could be more prominent, given the relatively low levels of CO2 in these tanks. In discussions of lid vs. no lid aquariums, I usually don't see gas concentration differences as an argument; this effect might be somewhat underestimated.
In these experiments, the O2 levels were not manipulated or measured, but its exchange between water and air should also be affected by aeration to a similar extent as CO2. I assume that O2 supply should be much better with airstones, and if CO2 levels are elevated with lids, O2 levels will be lower in aquariums with living creatures in them.
There were two questions I aimed to answer with these experiments:
- Does the aquarium lid limit gas exchange in the aquarium? The obvious answer to this is that it does limit it since it restricts the airflow between the headspace and the surrounding air. However, I was not sure if this effect was significant or negligible compared to the gas exchange happening between the water and the headspace air. Diffusion in the air could be orders of magnitude faster, which means that a typical aquarium top might not significantly impede dissipation.
- Does an air pump with an airstone facilitate better gas exchange than vigorous surface agitation with a powerhead? For some reason, I thought that the primary gas exchange effect of the airstones is the increase in surface agitation by lifting the water from the deep to the surface. I believed the bubbles' short time in the water column would not allow efficient gas exchange. Is that the case? Also, I was listening to Cory's podcast from Aquarium Co-op the other day, and he said that, based on measuring the O2 levels in the aquarium, airstones do a much better job of increasing O2 than any other method.
This time, I used a 120-liter aquarium with sliding tops to make the experiment more similar to a real aquarium setup. I filled it with 100 liters of water and left about 20 liters of headspace. A hole was drilled through one of the two parts of the sliding lid, and a tube was inserted to allow sampling without opening the top. Opening the cover could lead to significant gas exchange between the air and the headspace, and I wanted to avoid that:
Sampling was done multiple times for at least 400 minutes, and I calculated the diffusion capacity (k) values using the method presented in previous posts. The k value describes how robust the gas exchange is between the aquarium water and the ambient air, i.e., how quickly CO2 dissipates. The larger this value is, the faster the gas exchange occurs. Gas exchange was facilitated in two ways: with a powerhead pump with directed flow toward the water surface or with an airstone at the aquarium's bottom.
The powerhead output was 800 liters/hour, and the air pump output was 480 liters/hour. Although it would be ideal to ensure the same water agitation at the surface for the comparison, this is difficult to control. Judging from the following gifs, my guess is that surface movement was less pronounced in the case of the airstone:
Powerhead:
Airstone:
I filled the aquarium with tap water (13 dKH). Instead of using fresh tap water each time for CO2 supply, I chose to add CO2 with sparkling mineral water, 500 ml each time. After adding 500 mls, the typical pH levels were between 7.30 and 7.40, corresponding to CO2 levels between 33 and 26 ppm. With both aeration methods, I measured the k value in three conditions:
- Open lid
- Closed lid
- Closed lid, edges sealed with duct tape
As we can see, with these setups, aeration with airstone was much more effective at removing CO2 than surface agitation with a powerhead, despite water movement being stronger in the latter case. Additionally, closing the lid alone decreased CO2 exchange. When making the seal complete, CO2 dissipation almost came to a halt. Since CO2 dissipation with a given k equals:
I = k * (CO2water - CO2air)
and in a steady state situation, CO2 production (or injection) into the tank equals CO2 dissipation from the tank; there is an inverse linear correlation between the k value and the water-air CO2 difference. This means that if the k value is halved, the CO2 difference must double. So, manipulating the k value, such as using the aquarium lid, should directly influence aquarium CO2 levels.
Interestingly, in the case of airstone aeration, there is no such strong correlation between covering the top of the aquarium and CO2 levels. I think this is expected since the pump draws air from outside the aquarium, so the system is always supplied with fresh air, unlike the system with a powerhead only. This makes CO2 removal much more effective and more or less independent of the cover.
In conclusion, I think the take-home message is that the aquarium lid in this experiment decreased CO2 dissipation. Thus, it should increase the CO2 levels in aquariums, especially if no airstone aeration is used. The actual effectiveness depends on the lid's sealing potential, though. The other conclusion is that airstones promote gas exchange more effectively than surface agitation. Neither of these two conclusions is groundbreaking or probably not even new. Still, I find it challenging to find a lot of reliable information about these online, especially about the effect of the lids. In a low-tech setup, the lid effect could be more prominent, given the relatively low levels of CO2 in these tanks. In discussions of lid vs. no lid aquariums, I usually don't see gas concentration differences as an argument; this effect might be somewhat underestimated.
In these experiments, the O2 levels were not manipulated or measured, but its exchange between water and air should also be affected by aeration to a similar extent as CO2. I assume that O2 supply should be much better with airstones, and if CO2 levels are elevated with lids, O2 levels will be lower in aquariums with living creatures in them.
I never understood why people consider the air bubbles to be a poor oxygen source, relying only on the surface agitation they promote. We have used similar equipment for CO2 diffusion since the beginning and it is pretty clear that it diffuses more CO2 into the water than it dissipates CO2 through surface agitation, or it just wouldn't work.
Regarding your experiment, I agree that the likely answer is that when injecting fresh air into the partially sealed tank atmosphere, you lose the potential benefits of trying to trap the CO2 in there.
Regarding your experiment, I agree that the likely answer is that when injecting fresh air into the partially sealed tank atmosphere, you lose the potential benefits of trying to trap the CO2 in there.
Hi all,
@hax47 good post.
If I had a fish room etc. I would 100% go for a piston air pump, air main, <"Czech air lifter and HMF type sponge filters">. If I had enough air? I'd add in <"De Bruyn type planted trickle filters"> - <"https://aka.org/!area_Public/publicLibrary/~fishroom/debruynfilter/debruyn_filter.pdf">
One of the real advantages of photosynthesising submerged plants is that they are continually supply a trickle of oxygen and this gets around some of the problems with oxygen solubility.
cheers Darrel
@hax47 good post.
I think a larger gas exchange surface area is beneficial in a low tech for both carbon dioxide (CO2) and oxygen (O) exchange. I like <"a venturi on a powerhead etc">, because I'm not too bothered about CO2 concentration, but I want there always to be <"a lot of dissolved oxygen"> in the water.
If I had a fish room etc. I would 100% go for a piston air pump, air main, <"Czech air lifter and HMF type sponge filters">. If I had enough air? I'd add in <"De Bruyn type planted trickle filters"> - <"https://aka.org/!area_Public/publicLibrary/~fishroom/debruynfilter/debruyn_filter.pdf">
One of the real advantages of photosynthesising submerged plants is that they are continually supply a trickle of oxygen and this gets around some of the problems with oxygen solubility.
There are a couple of provisos to that. One is that air is only 21% oxygen, and the other is that oxygen is a <"much less soluble gas than CO2">.
cheers Darrel
Last edited:
_Maq_
Member
As for comparing solubility of carbon dioxide and oxygen, I've read a paper which recorded some interesting facts. The researchers were examining some springs and the creeks originating in them. Spring water is very often enriched in CO2, for obvious reasons. The researchers noticed, while advancing from the springs down the creeks, that the water gets quickly oxygenated up to full saturation, but CO2 dissipates from the water at much slower pace. They did not provide any exact explanation, they only guessed that it had something to do with much higher solubility of CO2. Something like "CO2 likes being dissolved in water and does not hurry to leave." Unlike oxygen.
From my experience, I can just say that all my tanks are covered by a glass lid, and I often - not always - employ venturi. I've tried to calculate CO2 content in my tanks derived from pH reading plus alkalinity reading (with increased precision). I came to an estimate of 3 to 6 mg/L CO2 being a norm. I.e. several times more than the equilibrium with air. This relatively modest oversaturation seems to be quite stable in spite of employing venturi. I guess that CO2 concentration between the lid and the surface remains elevated, too. Note that even very small increase of CO2 share in the air can move the equilibrium point significantly.
In sum, like Darrel, I don't care for CO2. It's always present, even if only in a tiny amount, but such a situation is fully natural and not dangerous for plants in itself. Unlike oxygen deficiency.
And hats off to @hax47 for his experiments. My deepest respect.
From my experience, I can just say that all my tanks are covered by a glass lid, and I often - not always - employ venturi. I've tried to calculate CO2 content in my tanks derived from pH reading plus alkalinity reading (with increased precision). I came to an estimate of 3 to 6 mg/L CO2 being a norm. I.e. several times more than the equilibrium with air. This relatively modest oversaturation seems to be quite stable in spite of employing venturi. I guess that CO2 concentration between the lid and the surface remains elevated, too. Note that even very small increase of CO2 share in the air can move the equilibrium point significantly.
In sum, like Darrel, I don't care for CO2. It's always present, even if only in a tiny amount, but such a situation is fully natural and not dangerous for plants in itself. Unlike oxygen deficiency.
And hats off to @hax47 for his experiments. My deepest respect.
Good point.
I agree, plants could compensate for the lack of gas exchange during the light period regarding the O2 supply.
The diffusion rate depends on the partial pressure difference, which is probably much bigger for O2 than for CO2 in this case.
Equilibrium CO2 levels
I tried to do a few experiments in the last few weeks that I have been planning for a long time. One thing I couldn't replicate from my experiments this summer: the equilibrium CO2 level. It used to be around 1.4-1.8 back then. Now, I could not get the water CO2 below 3.2 ppm around Christmas, and below 6.8 ppm the last few days. At least until I left all the doors and windows open for an hour to ventilate the house. Finally, I got 0.8 ppm. The 3.2 and 6.8 ppm water levels correspond to air CO2 levels of ~2200 and ~4700 ppm, both way above the supposed healthy levels.
I think the high CO2 levels make sense, the house is well insulated, and the whole family has been home for the holidays, inside the house all day... Even the dogs are inside 24/7. The whole family is feeding the plants.
Also during the winter, it is less likely to leave the doors, and windows open, and the overall ventilation is much worse than during the summer.
I wonder if those reports that claim their low-tech aquariums do better during the winter might actually related to elevated CO2 levels.
I tried to do a few experiments in the last few weeks that I have been planning for a long time. One thing I couldn't replicate from my experiments this summer: the equilibrium CO2 level. It used to be around 1.4-1.8 back then. Now, I could not get the water CO2 below 3.2 ppm around Christmas, and below 6.8 ppm the last few days. At least until I left all the doors and windows open for an hour to ventilate the house. Finally, I got 0.8 ppm. The 3.2 and 6.8 ppm water levels correspond to air CO2 levels of ~2200 and ~4700 ppm, both way above the supposed healthy levels.
I think the high CO2 levels make sense, the house is well insulated, and the whole family has been home for the holidays, inside the house all day... Even the dogs are inside 24/7. The whole family is feeding the plants.
Also during the winter, it is less likely to leave the doors, and windows open, and the overall ventilation is much worse than during the summer.
I wonder if those reports that claim their low-tech aquariums do better during the winter might actually related to elevated CO2 levels.
Parallel Measurements of Air and Water CO2 in an Aquarium
Over the past few years since I started maintaining planted aquariums, I have wondered whether increased aeration in low-tech aquariums adds or removes CO2. I have leaned towards the idea that CO2 levels in aquariums are typically higher than those in equilibrium with indoor air CO2 levels. It's commonly assumed that CO2 levels in low-tech aquariums are around 3 mg/l, which would be in equilibrium with air containing about 2000 ppm CO2. However, 2000 ppm is significantly higher than typical indoor CO2 levels. Furthermore, my pH measurements and CO2 calculations rarely show values below 3 mg/l, and almost never below 2 mg/l in my tanks. This suggests that CO2 production in aquariums exceeds what is consumed by plants, leading to CO2 loss into the air with increased aeration. To investigate this, I wanted to conduct parallel measurements of CO2 in the ambient air and water 24 hours a day, using a method that provides directly comparable CO2 data.
Instrument Design
I designed an instrument consisting of a plastic box with an open bottom that floats on the water surface, containing two CO2 sensors—one inside the box and one outside:
The air inside the box equilibrates with water CO2, functioning similarly to drop checkers. Over time, the partial pressures between water and the box should equalize, allowing the measurement of concentrations corresponding to water CO2 levels. I used affordable MH-Z19b CO2 sensors connected to a NodeMCU development board, taking measurements every 10 seconds for real-time CO2 monitoring. The downside is a detection lag due to slow equilibration, resulting in measurements that are a moving average of the actual values. However, the measurements are convenient and directly comparable, as both air ppm levels and water equilibrium CO2 levels depend on partial pressures.
This is how it looks like from the top and bottom:
I measured CO2 levels for 10 days in a low-tech aquarium without CO2 dosing and performed several interventions. I calibrated the sensors outdoors and began data collection before placing the box in the aquarium to ensure sensor accuracy. The aquarium is 120 liters, filled with 100 liters of tap water, filtered with an undergravel filter plus powerhead, with the powerhead output directed at the surface. It contains some plants, algae, and a few fish fry. The aquarium is lidless, illuminated by a 10W floodlight operating for two 4-hour periods daily with a siesta in between.
I placed a small pump under the box to circulate water and promote air-water equilibration:
Observations and Data
Initially, the water surface was covered with duckweed, except where the powerhead output is directed:
On day three, I turned on an air pump for about 12 hours. On day five, I changed the water and removed most of the duckweed:
I measured the CO2 levels for 10 days:
The plot shows the originally measured air-CO2 levels on the left y-axis and the calculated equilibrium water-CO2 levels on the right y-axis. The blue line represents box (water) CO2 levels, the red line represents air-CO2 levels, and the yellow stripes indicate lighting periods. The aquarium is in a bedroom, so room CO2 elevations occur at night when the room is occupied. Data collection began before placing the box in the aquarium, which is why the two lines start together on the first day.
I think there are a few interesting thing that we can read from the plot:
Over the past few years since I started maintaining planted aquariums, I have wondered whether increased aeration in low-tech aquariums adds or removes CO2. I have leaned towards the idea that CO2 levels in aquariums are typically higher than those in equilibrium with indoor air CO2 levels. It's commonly assumed that CO2 levels in low-tech aquariums are around 3 mg/l, which would be in equilibrium with air containing about 2000 ppm CO2. However, 2000 ppm is significantly higher than typical indoor CO2 levels. Furthermore, my pH measurements and CO2 calculations rarely show values below 3 mg/l, and almost never below 2 mg/l in my tanks. This suggests that CO2 production in aquariums exceeds what is consumed by plants, leading to CO2 loss into the air with increased aeration. To investigate this, I wanted to conduct parallel measurements of CO2 in the ambient air and water 24 hours a day, using a method that provides directly comparable CO2 data.
Instrument Design
I designed an instrument consisting of a plastic box with an open bottom that floats on the water surface, containing two CO2 sensors—one inside the box and one outside:
The air inside the box equilibrates with water CO2, functioning similarly to drop checkers. Over time, the partial pressures between water and the box should equalize, allowing the measurement of concentrations corresponding to water CO2 levels. I used affordable MH-Z19b CO2 sensors connected to a NodeMCU development board, taking measurements every 10 seconds for real-time CO2 monitoring. The downside is a detection lag due to slow equilibration, resulting in measurements that are a moving average of the actual values. However, the measurements are convenient and directly comparable, as both air ppm levels and water equilibrium CO2 levels depend on partial pressures.
This is how it looks like from the top and bottom:
I measured CO2 levels for 10 days in a low-tech aquarium without CO2 dosing and performed several interventions. I calibrated the sensors outdoors and began data collection before placing the box in the aquarium to ensure sensor accuracy. The aquarium is 120 liters, filled with 100 liters of tap water, filtered with an undergravel filter plus powerhead, with the powerhead output directed at the surface. It contains some plants, algae, and a few fish fry. The aquarium is lidless, illuminated by a 10W floodlight operating for two 4-hour periods daily with a siesta in between.
I placed a small pump under the box to circulate water and promote air-water equilibration:
Observations and Data
Initially, the water surface was covered with duckweed, except where the powerhead output is directed:
On day three, I turned on an air pump for about 12 hours. On day five, I changed the water and removed most of the duckweed:
I measured the CO2 levels for 10 days:
The plot shows the originally measured air-CO2 levels on the left y-axis and the calculated equilibrium water-CO2 levels on the right y-axis. The blue line represents box (water) CO2 levels, the red line represents air-CO2 levels, and the yellow stripes indicate lighting periods. The aquarium is in a bedroom, so room CO2 elevations occur at night when the room is occupied. Data collection began before placing the box in the aquarium, which is why the two lines start together on the first day.
I think there are a few interesting thing that we can read from the plot:
- I was surprised how much the duckweed restricted the gas exchange. After I removed it, the CO2 levels became lower than before (compare days 7-8 with days 2, 4 and 5). The CO2 equilibrium levels during the lighting periods averaged for the days 4 and 5 (duckweed), and days 7-8 (no duckweed):
- Similarly to my previous measurements, we can observe that CO2 levels decrease faster during light periods, most likely due to more intensive photosynthesis.
- CO2 levels increase between light sessions when duckweed covers the surface. Diana Walstad recommends a siesta period during the day with her method, and this observation seems to support that idea. However, when the duckweed was removed, there was no more CO2 increase during the siesta, although the CO2 dissipation was slower than during the light periods. I believe the effect of the siesta depends on the overall diffusion capacity of the tank setup (e.g., duckweed, aquarium top, etc.).
- Another factor that could explain the lower CO2 levels after the water surface was cleared of duckweed is that less light reached the plants, reducing CO2 uptake. However, CO2 loss was apparent even during the dark periods on days 6, 7, and 8, compared to accumulation during the same periods on days 2 and 4. This difference is not due to varying lighting intensities.
- Aeration with an air pump and diffuser made the gas exchange more effective, thus decreasing CO2 levels in the water. This is consistent with previous results. I suspect not only was CO2 dissipation accelerated, but probably also O2 uptake from the air.
- A single dose of soda water resulted in elevated CO2 levels, especially during the first lighting period, and even during the second period, compared to the previous day. Unfortunately, the sensor sensitivity range maxes out at 5000 ppm, so the peak of the CO2 curve is capped. Nevertheless, the added single-dose CO2 should result in about 30 mg/l of water CO2. Unlike my other aquariums, in which I regularly use soda water, this tank has no lid, yet CO2 levels still increased. If I used CO2 here, I could adjust the light periods to a continuous 8 hours after the CO2 dose.
- It is interesting to observe how closely the water CO2 levels follow the fluctuations in room CO2. This is most apparent on days 8 and 9, when the room CO2 partial pressure rose above that of the aquarium, causing CO2 to diffuse from the air into the water. However, even when the water CO2 partial pressure remains higher than the air partial pressure, the fluctuations in air CO2 levels are still mirrored in the water. Thus, indoor CO2 levels have a significant impact on the tank's CO2 levels. For me, this influence of ambient CO2 was the most intriguing observation. Room occupancy and ventilation are factors that are rarely discussed in the context of low-tech aquariums, although it seems that the aquarium breaths together with our home, especially when the CO2 diffusion is less restricted.
Awesome effort and really interesting results, I take my hat off to you @hax47 an ingenious way of measuring the tank water CO2 levels without requiring an underwater probe!
Have you ran the air pump with the duck weed removed to see the effect of that without the surface gas exchange restriction?
Have you ran the air pump with the duck weed removed to see the effect of that without the surface gas exchange restriction?
Bradders
Member
I was interested in that too.
I have not, but I would expect the CO2 levels to be even closer to the air CO2 equilibrium. I can try it though in the next few days.
Andy Pierce
Member
Awesome experiment. If you cover the tank with e.g. clingfilm or a lid, does that behave the same as the duckweed? It looks like something in the water is net producer of CO2 for you which is trapped in the water by the duckweed layer, and dissipated when you manually exchange gas under the duckweed with the air pump. Your house must be better insulated than ours - the CO2 in our living room starts in the 500-700 ppm range but rarely goes much over 1000 even when we're in the room for several hours.
I would expect so based on the previous experiments. The difference might be, that if I had a lid and a headspace under it, the oxygen exchange could be better because of the headspace oxygen. With the duckweed instead, there is no headspace as oxygen reservoir.
It is well insulated, and we are running an air-conditioner to cool the whole house, so we probably also ventilate less through the windows/doors than usual. Probably I should control the ventillation of the house based on the CO2 levels, the sensors are connected to home assistant anyways.
This is the first try of that setup but failed because the weather was good and the window was left open during the night before:
Quite interesting though, how much such a thing as an open window during the night pushed down the CO2 starting levels for the next day.
The window was left open (tilted actually), and still, the air CO2 got close to 1000 ppm. That is slightly over twice as high as the outside CO2, which does not seem unrealistic. I am not sure how accurate the sensors are, it is a bit challenging to get calibration CO2 concentrations besides the outdoor CO2 (to which they are calibrated).
I think it might be time to include other sensors for continuous readout in this system. O2, ammonia, and maybe VOC (volatile organic compounds) could be good candidates. What else would be interesting? I wonder if rotting plants, or cutting the plants would influence the VOC levels. I assume some of the dissolved organic compounds (DOC) would be also volatile? I think there are interesting observations to be made.
I have been planning to replace this aquarium with another one for a while. I have had the tank for months, but somehow I did not get to set it up. It would be interesting to watch how the CO2 production builds up during the cycling period. Based on this equation, we could observe the relative CO2 production:
If the DC (aeration, lid, duckweed) is kept constant, and the CO2 is not consumed (e.g. during the dark periods), the J (flux) is proportional to the partial pressure difference between the air and the water. And in equilibrium (constant air and water concentrations), the J equals the CO2 production. CO2 concentration is directly proportional to the partial pressure, so the difference between the two lines during periods, when the concentrations do not change considerably, should reflect the relative CO2 production. If I determine the DC as previously, the absolute CO2 production could be also calculated. It would be interesting to compare different aquariums in terms of CO2 production/liter of water.
