Very old subject that has been debated many times with almost all suggesting that it just works much better on the output. Big reason for me not to even try is that it seems a really good way for any excess cO2 to collect at the top of the canister which is where the impeller sets. An impeller will not move air so I don't want it setting even partly in a bubble. Dry impellers wear very quickly as they normally use water as sort of lube.
I figured I wasn't the first one to come up with this. That kinda makes sense, but I suppose it depends on the canister filter design a little. The newer eheim design seems to suck the water to the bottom and push it through the media, and what remains through the impeller. By my logic if the CO2 is ever going to dissolve, it would have dissolved by then. If it hasn't, it is possible the injection rate is too high. After googling one aquarist said that even if the pump ran empty the overheating is not as much of an issue with modern magnetic impellers. Some users said they had tried it, but none of them had measured the effects. I'm considering doing this with a bigger 400-500 liter tank I am planning in the future. For the 120 liter I have no I don't see much of a benefit with the exception of perhaps better visibility.
Using the canister filter as co2 reactor is good idea. I have experience 5+ year's with dissolving co2 through canister filter. I'm using Fluval 406, Eheim 2075 and Sunsun 302 without any problem, no leaks, no problem with biological filtration, no problem with gasket wear off. Everything works fine. Only one thing, occasionally you can hear undissolved co2 going out through outlet, no big deal.
Thanks for sharing!
Why re-invent the wheel. We know it works at the outflow very well.
In my opinion discussions related to possible improvements should be welcomed, not shut down. While something may work now does not mean it can't be improved upon.
The problem is there are many that do have problems depending on filter and levels of co2. I've never heard of anyone having issues at the outflow.
The "problem" would be low dissolution an atomizer offers and not having to buy an actual reactor. The answer is basically the same as to why someone would want to buy a CO2 reactor. The biggest reason would be more efficient use of CO2 I guess, in the sense that you have to inject less for the same results and it is easier to disperse it around the tank with the water flow. Another one would be less microbubbles = more visibility. My tank for example has lights on an no CO2 injection only 1.5 hours per day.
So far I've found some users having issues with gas building up (which means injection rate was likely too high, and tuned for a setup where a lot of CO2 is lost), and others saying that it wasn't as efficient as they had hoped. Some people raised concerns about bacteria dying because of pH or high CO2 meaning low O2, but some users were quick to point out this not to be the case. If pH was an issue that would be a problem regardless of injection method (or just too much CO2), and CO2 levels are independent of O2 levels - it is possible to have both at high levels at the same time.