Key takeaways
- A reactor dissolves 90 to 100 percent of the gas you feed it, against roughly 50 to 75 percent for an in-tank ceramic disc.
- Reactors cost 10 to 25 percent of your canister throughput, which can matter more than the gas they save on a small filter.
- Ceramic discs need 20 to 30 psi and inline atomisers 40 psi or more, so working pressure has to match the method.
- Swapping diffusion methods at the same bubble rate can lift dissolved CO2 by a third, so halve the rate and ramp again.
- A freshly cleaned diffuser passes far more gas than the clogged one it replaced, which gasses tanks during routine maintenance.
Two 200 litre planted tanks, both holding a steady 30 ppm, both fed from a 5 lb cylinder. One of them gets refilled twice a year. The other needs four fills to do the same job.
The difference is not the regulator and it is not the bubble rate. It is what happens to a bubble in the seconds after it leaves the tubing, and whether that bubble dissolves or simply travels to the surface and escapes into your living room.
Both devices sit at the very end of the line, downstream of the check valve, and everything upstream of them is identical across the run from cylinder to tank. That makes this one of the few CO2 decisions you can change in isolation without rebuilding anything else.
What a ceramic diffuser actually does
A ceramic or glass diffuser forces gas through a fired disc with pores measured in tens of microns. The result is a mist of bubbles small enough that surface area to volume is high, so they dissolve as they drift.
Dissolution happens during the rise. A bubble in a shallow 40 cm tank with weak flow has a couple of seconds before it reaches the surface, and much of it is still gas when it gets there.
Put the same diffuser under a filter outlet and the picture changes completely. Bubbles get caught in the current, pushed sideways and down, and spend far longer in the water. Placement is worth more than the disc itself.
What a reactor does differently
A reactor is a sealed chamber plumbed into a canister filter return. Gas is injected at the top, water is forced through, and bubbles are trapped and tumbled inside the chamber until they are gone.
Nothing escapes to the surface because nothing reaches the surface as gas. Dissolution is close to complete, which is where the two fills a year figure comes from.
The trade is flow. A reactor is a restriction in the return line, and on a canister already loaded with media, expect to lose somewhere between 10 and 25 percent of your throughput.
Which dissolves more, and what it costs you
Efficiency figures here are ranges rather than fixed numbers, because dissolution depends heavily on tank depth and flow pattern. Treat them as the shape of the difference, not laboratory values.
| Method | Typical dissolution | Working pressure | Flow penalty | Mist in tank | Cost |
|---|---|---|---|---|---|
| In-tank ceramic diffuser | 50 to 75 percent | 20 to 30 psi | None | Visible | $10 to $30 |
| Inline atomiser | 70 to 90 percent | 40 psi and up | Small | Fine haze | $25 to $60 |
| Inline reactor | 90 to 100 percent | 15 to 20 psi | 10 to 25 percent | None | $60 to $150 |
| Venturi on a powerhead | 60 to 85 percent | 15 to 25 psi | None | Coarse mist | $20 to $50 |
| Bubble ladder | 30 to 50 percent | Very low | None | Large bubbles | $8 to $20 |
Changing diffusion method at the same bubble rate can raise dissolved CO2 by a third or more overnight. Swapping a ceramic disc for a reactor is not a like for like change. Cut the rate by half and rebuild it using the staged ramp over several days.
Pressure: the requirement people miss
A ceramic disc has real resistance. Below roughly 20 psi it will pass nothing at all, or it will pass gas unevenly in coughs rather than a steady mist, and that unevenness makes bubble counting meaningless.
Inline atomisers ask for more, usually 40 psi or above, because the disc sits inside a pressurised water line. Reactors ask for the least, since gas only has to enter a chamber at slightly above line pressure.
Low pressure DIY yeast systems produce nowhere near 20 psi, which is why they come with a ladder or a bell rather than a disc. Fitting a ceramic diffuser to a yeast reactor produces nothing and wastes an afternoon.
Maintenance is where the honest difference sits
Ceramic discs clog. Bacterial film and mineral deposit close the pores over two to four months, back pressure rises, and output quietly falls while your bubble counter reads the same number.
Clean a disc by soaking it in dilute household bleach at roughly one part in ten for twenty minutes, then rinse thoroughly and soak in dechlorinator before it goes back in the tank. Discs generally want replacing every six to twelve months regardless.
A freshly cleaned diffuser passes far more gas than a clogged one at the same needle valve setting. Cut the rate before you put a cleaned disc back, then work up again over two or three days. This is a common way to gas a tank during routine maintenance.
Reactors need burping. A pocket of undissolved gas accumulates at the top of the chamber, and once it grows large enough it kills the water contact and the pump starts making a churning noise. Some designs purge themselves, most want a manual bleed every few weeks.
Inline atomisers sit somewhere between the two. The disc still clogs on the same schedule as an in-tank unit, but reaching it means shutting the filter down and opening the housing, so most people leave it far longer than they should.
Over five years the running costs land closer together than the purchase prices suggest. A reactor saves perhaps $60 to $120 a year in refills on a large tank, while a diffuser owner spends $20 to $40 a year on replacement discs and an hour or two of cleaning.
The flow question, and when a reactor is the wrong answer
Planted tanks generally want turnover somewhere around eight to ten times the tank volume per hour, and CO2 distribution depends on it more than on injection rate. A reactor that costs you 20 percent of that turnover can leave you with 30 ppm in one corner and 15 ppm in another.
Work out your actual flow after media, hoses, and fittings before adding another restriction, using the sizing approach in how to work out real turnover rather than rated figures. If your canister is already marginal, an inline atomiser gives you most of the efficiency with almost none of the restriction.
A 90 litre nano that should not have had a reactor
A customer fitted a DIY Cerges style reactor, built from a 10 inch inline water filter housing, to a 90 litre tank running an Eheim Classic 2213. On paper the filter moves 440 litres per hour, which sounded ample.
After media, 1.5 metres of hose, lily pipes, and the reactor, real flow was closer to 200 litres per hour. Within three weeks he had black beard algae on every leaf facing away from the outlet and a drop checker that read green at one end of the tank and blue at the other.
We pulled the reactor, fitted a $35 inline atomiser, and flow recovered to roughly 330 litres per hour. Gas consumption rose slightly, the algae receded over six weeks, and the two drop checkers finally agreed with each other. The reactor was the better device and the wrong device for that tank.
Mist: preference, mostly
Some keepers like visible mist and argue that undissolved microbubbles contacting leaf surfaces help directly. The evidence is thin and the effect, if real, is small next to simply having enough dissolved CO2 in circulation.
Others find mist distracting in an aquascape and choose a reactor for that reason alone. That is a legitimate reason to spend the money, and the fuller argument for each placement is set out in the case for keeping gear inside or outside the tank.
Which to buy
Tanks under about 120 litres with a hang on filter or a small canister: in-tank ceramic diffuser, placed under the outlet. Cheap, simple, and the efficiency loss on a small tank amounts to very little gas in absolute terms.
Tanks from 120 to 250 litres with a decent canister: inline atomiser. It hides the equipment, dissolves well, and barely touches your flow.
Tanks above 250 litres, or any tank where cylinder refills are a nuisance: reactor, provided you have flow to spare. Whichever you choose, verify the result with a drop checker at both ends of the tank rather than trusting the specification, and read the colours against the numbers in the shade by shade concentration scale. Gas exchange at the water surface works both ways, as the USGS material on dissolved gases in water lays out.
Questions readers ask about this
There is no benefit and two real drawbacks. You lose the ability to attribute a change to either device, and you carry the flow penalty of the reactor while still venting gas from the diffuser. Pick one, tune it, and keep the second as a spare. If distribution is uneven, fix flow rather than adding a second injection point.
Watch the bubble pattern rather than the calendar. A healthy disc produces an even veil across its whole face. When gas starts emerging from one edge only, or the mist coarsens into visible individual bubbles, the pores are partly blocked. On most tanks that arrives somewhere between two and four months, sooner in harder water.
It can, and the usual cause is a gas pocket in the chamber rather than a fault in the pump. Bleed the reactor and the churning normally stops within a minute. Persistent noise after burping usually means the injection rate exceeds what the chamber can dissolve at that flow, so lower the bubble rate or increase turnover.
The disc material is what matters, and a good disc costs a few dollars regardless of what it is mounted in. Expensive glass units buy appearance and durability, not measurably better dissolution. Spend the money on placement instead: any diffuser sitting under a filter outlet beats a premium one tucked in a still corner.
