Carbonic maceration seals whole, intact coffee cherries inside a tank flooded with carbon dioxide and lets them ferment with almost no oxygen at all. The technique came from winemaking, and the word that gets borrowed along with it is "inside": wine writers describe grapes fermenting from the inside out, cell by cell, before a single yeast is added. Coffee writers reach for the same image when they describe a carbonic-maceration lot. I have used that image myself. Writing this piece sent me looking for the one study that actually measured a coffee cherry treated this way, and it does not let me say that with a straight face anymore.
What actually happens inside a sealed cherry?
The clearest account of the mechanism comes from wine science, where the technique has been studied for nearly a century. The Australian Wine Research Institute describes the setup plainly: whole grapes go into a closed vessel "with absolutely no free grape juice," and the vessel is filled with carbon dioxide, which displaces the oxygen. The berries then absorb that CO₂ until they reach roughly half their own volume in gas. That shift, from an oxygen-rich environment to an oxygen-starved one, pushes the grape's own cells from ordinary aerobic respiration into fermentative anaerobic metabolism, and the AWRI is specific that this first phase runs without any yeast at all. The grape ferments itself, converting a small amount of its own malic acid and sugar into ethanol, before anyone crushes it or adds a culture.
The practice traces to Beaujolais, in France, where producers used it on Gamay grapes from roughly the 1930s through the 1960s, fermenting whole bunches in CO₂-filled tanks to get soft, fruit-forward reds ready to drink young rather than age for years. It stayed a wine technique for most of a century. Coffee's adoption dates to the early 2010s, when producers in Colombia and Costa Rica began experimenting with sealed-tank fermentation generally, and the moment that put the specific, CO₂-flooded version into "coffee consciousness," as the trade press tells it, was Saša Šestić's win at the 2015 World Barista Championship with a carbonic-macerated Sudan Rume grown in Colombia. Producers across Latin America started running their own versions soon after, drawn by the same thing winemakers wanted: a lever on flavor that opens up before fermentation, rather than one that only shows up at the roaster or the cupping table.
On the wine side, the recipe is fairly specific: roughly 30 to 32°C for five to eight days gets the most pronounced character; lower temperatures stretched out to three weeks give something milder. Either way, the first phase stays yeast-free and happens inside the berry's own skin. The AWRI also flags the failure mode, and it is the reason the method is run as a sealed, monitored batch rather than left alone in a corner: if the CO₂ blanket is not maintained, ordinary aerobic microbes get back in, and the higher pH and warmer temperatures the method depends on make acetic acid bacteria and spoilage yeasts like Brettanomyces more likely to take over, not less. The same tank that produces a prized wine, run a little carelessly, produces a ruined one. There is no reason to think coffee is more forgiving of the same chemistry.
Does coffee ferment the same way wine does?
Coffee writing is actually split on this, and the split is worth seeing clearly rather than smoothing over. Barista Magazine's account stays vague about the mechanism: whole cherries go into airtight tanks, carbon dioxide is pumped in to push out the oxygen, and the gas works to "soften the tissue of the coffee fruit" while the seed "sits through fermentation of the fruit's flesh." That reads as ordinary pulp fermentation happening around a passive seed. Ozone Coffee's account goes further and is far more specific: it states plainly that fermentation happens "inside the cherry itself," that "the fruit's own cells switch to anaerobic metabolism," that "enzymes within the cell drive glycolysis," and names the process "intracellular fermentation" outright, adding that "the intact skin is the whole point" because it is what keeps the fermentation inside the cell rather than letting outside microbes take over. That is the precise wine-science claim, carried over term for term, and applied specifically to the fruit's cells rather than the seed.
So one practitioner source hedges and one commits, and between them they cover the honest range of what is plausible. What neither has is a measurement. The most direct evidence I could find is a 2026 peer-reviewed study of ohmic-heated carbonic maceration on Arabica cherries from Tanah Toraja, Indonesia, published in AIMS Agriculture and Food by Mustafa and colleagues. It is exactly the kind of source this question needs, and it is also candid about its own limits: the authors state outright that "microbial community composition was not directly measured" and that the mechanisms behind their own results "were not directly measured in this study." A team that set out to measure exactly this process on exactly this fruit did not test whether the changes they found came from the cherry's own cells switching to anaerobic metabolism, from the microbes already living on its skin working in a low-oxygen environment, or from both at once. Ozone Coffee's specific, confident "intracellular fermentation" claim is a reasonable inference from what wine's cells do under the same conditions. It is still an inference rather than a coffee measurement, and the one group that ran the coffee measurement did not settle it either way.
What does carbonic maceration actually change in the bean?
The numbers that do exist are worth more than the metaphor, because they are specific. Mustafa and colleagues fermented whole cherries in sealed PVC-tube tanks, using an electric current run through the water inside the tank (ohmic heating) to hold the temperature steady at either 31°C or 36°C, for 48, 144, or 240 hours, then depulped, washed, and sun-dried the beans as usual. Their control was cherries that went straight to sun-drying with no fermentation at all, so every comparison below is against doing nothing, not against a washed or natural lot processed normally.
Chlorogenic acids moved, and not all in the same direction. The 3-CQA isomer fell from 127.1 in the control to 88.33 at 36°C for 240 hours; 4-CQA fell from 165.80 to 125.60 under the same conditions. The 5-CQA isomer went the other way, rising from 449.6 to 515.6, the only one of the three that increased under any tested condition. Trigonelline rose too, from 239.60 in the control to 312.20 at the mildest treatment (31°C, 48 hours). Caffeine did something stranger: it rose, from 312.80 to 404.40, at 31°C held for the full 240 hours, but fell to 296.0 at 36°C for the same duration. The same compound moved in opposite directions depending on five degrees of temperature, which is the kind of result that should make anyone cautious about a confident one-line claim for what carbonic maceration "does" to a coffee's chemistry.
The study also ran a full sugar panel (reducing sugar, total sugar, and sucrose) and a volatile-compound screen that picked up 149 separate aroma-relevant compounds across the treatments, mapped out in a heatmap rather than a short table. That is too much detail to responsibly compress into a sentence or two without losing the pattern that actually matters, and the paper itself treats it the same way, presenting it as a reference map rather than a single headline number. The honest summary is narrower than the full dataset: direction and magnitude both depend on exactly how hot and how long the tank runs, and nobody should trust a one-line summary of 149 compounds, including this one.
Does it actually make the coffee taste better?
Practitioner writing promises a lot here. Barista Magazine's own description of the outcome reaches for "intense flavor profiles, with boozy and cooked fruit flavors," while warning that a mishandled batch risks "astringency, dryness, and bitterness" instead. Both can be true, and the study gives a sense of how narrow the gap between them actually is. Mustafa and colleagues had two certified cuppers score every batch. The control scored 84.5. Four of the six fermented treatments scored 85.0 to 85.75, which the authors call "slightly higher." One treatment, held at 31°C for 144 hours, landed right on top of the control at 84.5. And one, the shortest and hottest combination tested, 36°C for 48 hours, scored lower than doing nothing at all: 83.0.
Two cuppers and no reported statistical test is not enough evidence to say carbonic maceration reliably raises or lowers a cup score. A cupping protocol like the SCA's asks tasters to score the same cup on acidity, sweetness, body, and several other axes, and the whole point of running more than one or two cuppers is to see whether their numbers agree with each other before anyone trusts the average. With two people and six treatments, a single cupper's bad morning is enough to move a treatment from "slightly higher" to "about the same." What the numbers do show, reliably, is that the gap between the best and worst outcomes measured here is under three points, and that getting it wrong is not a hypothetical. The same tank, run five degrees hotter for a shorter time, scored below a cherry that was never fermented at all. A producer paying extra for the tanks, the gas, and the labor to watch them is paying for a result that, on this one dataset, is a coin flip between a modest improvement and a worse cup than doing nothing.
How is this different from ordinary anaerobic processing?
We publish an explainer on anaerobic processing generally, and it is worth being precise about where carbonic maceration sits inside that broader category, because the two terms get used almost interchangeably. Anaerobic processing just means fermenting coffee, whole or depulped, in a sealed container that excludes outside air; the carbon dioxide inside builds up gradually as a byproduct of the fermentation the microbes are already running. Carbonic maceration is the narrower, imported version of that idea. The cherries have to go in whole and unbroken, because a split skin lets the surrounding microbes straight at the sugars and defeats the point of the method. And the tank is deliberately flooded with carbon dioxide from an outside source before fermentation gets going, rather than waiting for the process's own gas to accumulate. Every carbonic maceration is anaerobic. Most anaerobic coffee is not carbonic maceration.
That distinction also explains why the fermentation substrate matters here in a way it does not for a depulped lot. A whole cherry still carries its skin, its sugar-rich mucilage, and its parchment and seed, all sealed together, which is a different fermentation environment from a depulped bean sitting in a tank with its mucilage exposed to the open water around it. It is also a close cousin of what happens in a natural-processed lot, where the whole dried cherry ferments in the open air rather than in a sealed, CO₂-flooded tank. Carbonic maceration is natural processing's more controlled, more deliberate cousin, not a wholly separate idea.
"More deliberate" is also where the cost sits. Ozone Coffee's framing of the appeal is control: a producer running carbonic maceration can monitor and adjust temperature, CO₂ level, pH, and duration, and treat fermentation "not just as a step, but as a tool" rather than something that just happens while the cherries sit. That control is not free. It needs sealed, gas-tight tanks, a CO₂ supply, and someone checking the batch on a schedule rather than walking past it once a day, on top of the plain fact that an experimental lot this labor-intensive only pays for itself if the cup ends up good enough, and distinct enough, to sell at a premium over a washed or natural lot from the same farm. That is the commercial logic behind why this shows up on small, award-chasing microlots rather than on the bulk of what any origin produces, and it is also why the study above, with its narrow and uneven gains over doing nothing, is a genuinely useful data point against the marketing version of the story.
Can you buy a carbonic-maceration coffee from us?
Not today. We do not carry a carbonic-maceration lot right now, which means everything above is something I read and reported, not something I cupped myself against a lot we chose. I would rather say that plainly than borrow a flavor description from a producer's sell sheet and let it read as our own tasting note.
If you come across one somewhere else, the honest way to shop it is with the numbers above in mind rather than the marketing copy alone. A carbonic-maceration lot almost always costs more than a washed or natural lot from the same farm, because the tanks, the CO₂, and the labor to monitor the batch are real costs that have to be recovered somewhere. That premium buys you a coin flip, not a guarantee: on the one dataset above, the best fermented batch beat the unfermented control by about a point, and the worst fermented batch lost to it by a point and a half. A seller who frames the method as a reliable flavor upgrade is saying more than the research currently supports. A seller who frames it as an experiment worth tasting for yourself, at a price that reflects the extra work, is closer to what the evidence actually shows.
If you want to taste what an ordinary, well-run fermentation does to a cup, without any of the chemistry above, our washed Kenya Kiambu is the plainer comparison: depulped, fermented, fully washed, and dried, the method most of the world's coffee still uses. It is also where a coffee shop or church consult usually starts, because most buyers are deciding between washed and natural long before an experimental lot like this one is on the table.
