Co-fermentation and infusion

Where it is practiced

Colombia supplies most of the protocols published by producers themselves, and it is a fact that deserves to be explained rather than merely noted. The country has long had a network for training producers, so that a new technique spreads there in one or two harvests instead of a generation. Producers there also suffer rising costs and unstable prices, and look for added value that requires neither a new plot nor a new variety. A fermentation method can be changed in a year; a replanted coffee tree gives its first real harvest after three.

Three names recur in the protocols with figures, and they are names of farms as much as of people. Edwin Noreña runs the finca Campo Hermoso in Circasia, in Quindío. Rodrigo Sánchez Valencia runs Monteblanco, in Huila, and publishes his operating procedure. Diego Bermúdez runs the finca El Paraíso in Piendamó, in Cauca, and his exporter details his lots strain by strain. Panama, Costa Rica and Ecuador, for their part, appear in competition affairs rather than in protocols, which says something about the way the subject arrived in the trade: through scandal before arriving through technique.

The practice is first spotted from 2018. A cinnamon espresso is served at the World Barista Championship in Amsterdam that year. A second cinnamon coffee goes through at the Brewers Cup in Boston in 2019. Both came from Costa Rica, from different farms and varieties, with the same cinnamon profile. It is this identity of taste between two unrelated lots that raised suspicion of an addition, and both competitors were unaware of what they were serving.

Two other cases follow. An Ecuadorian lot flavored with tropical fruits is disqualified at the La Loja competition in 2019. A Colombian lot with rose water and red fruits wins an Italian Brewers Cup, and analysis finds essential oils in it. The trade press traces the appearance of these lots back to around 2018 to 2020.

The same press names five dominant export destinations:

• the Middle East, including the Gulf countries;
• China;
• South Korea;
• Taiwan;
• Japan.

It also reports rising demand in Europe and the United States in the same text, which qualifies the idea of a market confined to East Asia. The volume is described as small compared with washed, natural and honey, with no figure to back it up. No production figure can be verified. See the gray area.

The principle

Two very different gestures circulate under the same banner, and telling them apart is the whole point of this sheet.

Co-fermentation adds a material to the fermenting mass, therefore before drying, therefore while microorganisms are at work. Infusion applies an odorous substance to green coffee, that is, to the bean already dried and stripped of all its coverings. Both produce a coffee that smells of fruit or spice. The path by which the molecule arrives is not the same, and it is this path that decides everything else: admissibility in competitions, the value of the word fermentation, and the question of whether the producer ran a process or perfumed a commodity.

The list of additions reported in the tank is wide:

• fruits, whole, as juice, puree or dehydrated;
• cocoa pulp, molasses, rare sugars such as allulose;
• cascara, which is the dried pulp of the coffee cherry;
• whole spices, flowers, leaves;
• hops, those resinous, bitter cones that flavor beer;
• mosto, which is the juice of the coffee itself and deserves a paragraph of its own.

The mechanism of co-fermentation is first of all indirect, and this point is poorly understood on both sides of the debate. A fruit or a molasses is sugar, organic acids and, in addition, pectin. It is therefore additional substrate for the microorganisms already present, that is, food. The fermentation goes further and drops lower in pH. Yeasts transform these sugars into ethanol, lactic acid bacteria into lactic acid, acetic acid bacteria oxidize ethanol into acetic acid. Ethanol and acids then recombine into esters, that family of fruit-smelling molecules of which ethyl acetate is the best-known representative. In other words, a tank of coffee makes fruity notes by itself, without any fruit being put in. Adding fruit feeds this machine as much as it perfumes the coffee, and probably more.

Mosto is the case that sheds light on all the others. The word, sometimes written mossto, designates the juice drawn from the mucilage, that layer of sweet flesh stuck to the parchment, collected during a first fermentation. This liquid carries sugars, organic acids and a population of yeasts and lactic acid bacteria already established. Reusing it means making a starter, in the exact sense in which a bread starter is made: a fraction of the previous fermentation is used to launch the next. It is not a foreign material introduced into the tank, it is coffee put back on coffee. Edwin Noreña describes his use of mosto as a catalyst of fermentation, and the word is right: mosto does not supply an aroma, it supplies a population and a head start.

There remains the question of direct transfer, the only one that really interests someone who wants to know what they are drinking. The bean is a living seed, wrapped in the parchment, the parchment-like shell, and in the silverskin. During wet fermentation it absorbs water, and small soluble molecules travel with it. That the bean changes in composition during a fermentation is measured and not disputed. Metabolomics, which is the measurement of all the small molecules present in a sample, finds on beans fermented with a chosen starter more than a hundred metabolites different from those of a spontaneous fermentation. Among them are mannitol and methyl phenylacetate, which are microbial products and not products of the fruit.

But that a molecule belonging specifically to the added fruit is found inside the bean has been measured by no trial read for this sheet. Fruit aromas are in large part esters and terpenes poorly soluble in water, which leads one to expect that they stay mostly on the surface and on the parchment. This is solubility reasoning, not a measurement, and it must be said so.

Infusion sidesteps the whole question. Applied to green coffee, it has no covering to cross and no microorganism to mobilize. The dry bean is porous and takes up moisture again: it retains the substance on its outer layers and in its fissures. Calling this a fermentation attributes to microbial work what comes down to a bottle of flavoring.

The sequence

A single word covers practices that have nothing in common. At one end, a bucket of pulp poured onto a mass of cherries, with no measurement and no schedule. At the other, a written protocol, with figures, repeated from one harvest to the next. What follows describes the published protocols, with their numbers, because it is the only way to show what know-how means here.

First protocol, Edwin Noreña’s at Campo Hermoso, published by a green coffee trader on a dragon fruit lot:

• Caturra grown at 1,650 meters, cherries harvested above 20 degrees Brix. Brix measures the sugar concentration of a liquid, and serves here as a ripeness marker;
• twenty-four hours in a low-oxygen system, between 18 and 22 degrees;
• forty-eight hours of carbonic maceration in 2,000-kilo tanks, at the end of which the mosto is collected. It is then at pH 4.2 and 11.5 Brix;
• the mosto is inoculated with a chosen yeast, then fed with a solution of allulose and dehydrated fruit. The result is a starter culture;
• the cherries, brought to the black honey stage, go into 200-kilo reactors. The started mosto goes in at a rate of 10 milliliters per kilo, for 72 hours at 18 degrees;
• drying in a greenhouse under controlled conditions for 10 days, then stabilization with dehydrated pulp for 10 days.

Two points of this protocol matter more than the others. The dose is given, 10 milliliters per kilo, or about one percent of the mass. And the dehydrated fruit does not go into the coffee tank: it feeds the starter, which goes in afterwards. This is not a detail of presentation. A fruit that feeds a starter is consumed by microorganisms before meeting the coffee; a fruit poured onto the mass arrives whole, with its flora and its free sugars.

Second protocol, Rodrigo Sánchez Valencia’s at Monteblanco, which follows the same starter logic and pushes it further. A mother culture of 80 liters brings together lactobacilli and Saccharomyces cerevisiae. It is fed with fresh fruit and a sweetener, molasses for example. The stated aim is to bring the sugar content of the culture to the Brix of the coffee to be processed, and the reason for this match deserves to be understood: a starter suddenly richer than the mass it meets goes off into violent fermentation instead of settling in, and a poorer starter is dominated by the flora of the cherry. This first fermentation lasts 190 hours, the time to reach the target Brix and pH. A passion fruit lot is detailed by a reseller: purple Caturra harvested between 20 and 23 Brix, mother culture with ten yeast strains, lactic acid bacteria and fresh passion fruit, then 150 hours in a closed tank at 25 degrees, pH held above 4.0.

Third protocol, Diego Bermúdez’s at El Paraíso, which is the most instrumented of the three. His exporter describes one strain per target profile: Saccharomyces carlsbergensis, Saccharomyces cerevisiae, or Kluyveromyces lactis combined with lactobacilli in a milk-based medium. This last choice has a precise and verifiable reason: Kluyveromyces lactis consumes lactose, which is milk sugar, something most yeasts cannot do. The culture medium is therefore chosen for the strain, and not the reverse. The rest is figured:

• disinfection of the cherries with ozone, water filtered by reverse osmosis and then by ultraviolet light;
• first fermentation of 48 hours at 18 degrees, second of 48 hours at 21 degrees;
• one lot run under 20 psi of pressure for 36 hours;
• an aerobic phase receiving 250 milliliters of culture medium per kilo;
• thermal shock at the end of the tank stage, water at 40 degrees then water at 12 degrees;
• drying at 35 degrees under 25 percent relative humidity for 34 hours, down to 10 to 11 percent bean moisture.

Thermal shock stops microbial work through an abrupt temperature gap. Drying controlled for temperature and humidity answers the risk specific to these lots, which is the most serious of all: a sweeter, wetter mass dries less well, and drying that drags on over free sugar is the shortest road to mold.

These three protocols share a common architecture, and it is no accident. All are in two stages. Whole cherry first, pulped coffee afterward, the addition coming only in the second phase. The reason is mechanical: the skin of the cherry limits exchanges and protects the mass, while pulping opens the mucilage to everything that comes by. Entering late shortens the time during which a free sugar remains exposed to molds.

The controlled atmosphere is combined with this for the same reason. Carbonic maceration, borrowed from winemaking, places the cherries in an airtight tank saturated with carbon dioxide. The gas drives out the oxygen. Acetic acid bacteria and molds, which need it, retreat; yeasts and lactic acid bacteria hold the ground. Combined with a sweet addition, this practice steers the sugar toward ethanol and lactic acid rather than toward acetic acid, which is the marker of the fermented defect. In other words, the closed tank is not there to look exotic, it is there because a sugar addition makes the tank more fragile and it must be protected.

Infusion, for its part, is run quite differently, and the trade press describes three routes:

• soaking the green bean in water with cinnamon sticks, for three to five days;
• a dry fermentation on the farm, with tartaric acid and cinnamon sticks;
• applying essential oils or commercial flavorings, on green beans or after roasting.

For this third route, no dosage and no duration is published anywhere.

What remains is what, in practice, separates a directed co-fermentation from a simple addition. The answer fits in a list, and it can be checked lot by lot:

• the Brix of the cherry at harvest, which sets the starting point;
• the Brix of the starter, matched to that of the coffee;
• the pH, with a floor held, 4.0 at Sánchez, corrected with citric acid or bicarbonate at Bermúdez;
• the tank temperature, held between 18 and 21 degrees in the protocols read;
• the duration, counted in hours and not in days;
• the dose of the addition relative to the mass, 10 milliliters per kilo for one, 250 milliliters per kilo of medium for the other;
• the sanitary state of the addition and of the cherry, ozone and filtered water;
• the stopping gesture, washing or thermal shock;
• the running of the drying, for temperature and humidity, down to 10 to 11 percent.

A know-how is recognized by what it measures. It is recognized above all by what it stops. Adding sugar to a tank shortens the window during which the fermentation stays in the right place. Without a stopping point decided in advance, the addition directs nothing at all, it accelerates.

What it gives in the cup

The profiles reported by the lot sheets are frank and convergent: exotic fruits, candied red fruits, milky or winey notes, acidity often low, sweetness high. The sellers’ vocabulary leans hard on them, and these descriptors must be read for what they are, that is, a sales argument as much as a description.

A laboratory result usefully complicates the picture here. A fermentation simply inoculated with a yeast, with no fruit added, already gives descriptors of berries, cocoa and honey, and the scores obtained with two native strains were 83.72 and 84.28 points. A co-fermentation of two yeasts assisted by ultrasound, again with no fruit, obtains 82.16 points. In other words, the intense fruity profile is not the signature of an addition: it is reachable through microbial management alone.

This observation cuts both ways, and that is why it is valuable. It takes strength away from the idea that a co-fermented lot could be recognized in the cup, which many tasters claim. It also takes strength away from the idea that a fruit addition would be the only way to obtain these cups, which many sellers imply.

What no published trial provides is the comparison that would settle it: the same coffee, from the same plot and the same harvest, run with and without the addition, tasted blind by a panel that does not know which is which. This experiment is not difficult. It has simply never been published.

What goes wrong

Three things that are constantly mixed together in the case against these coffees must be separated, and they are not equally serious. A disagreement of taste is not an offense. A breach of a competition’s rules is not a deception about the goods. Separating them is the first condition for talking usefully about the subject.

First the technical risks, which are real and concern the producer before the drinker.

• Drying is the breaking point. A sweeter, wetter mass dries more slowly and molds more easily. That is why serious protocols control drying for temperature and humidity instead of leaving it to the patio.
• Acetic acid takes over as soon as oxygen returns. A tank badly closed, opened for a sample then badly reclosed, turns to vinegar faster than a tank without an addition, precisely because there is more to eat.
• The addition itself can be the vector of a contamination. A fresh fruit carries its own flora, and not every species it carries is desirable. This is what the ozone disinfection in the Bermúdez protocol addresses.
• Without a stopping point decided in advance, the window closes without warning. The added sugar shortens the time available.

Next the drift that is a deception, and it is the only one that is. A lot receives a fruit, a spice or a flavoring, then is sold as the expression of a terroir, a variety or an altitude. The buyer pays for agricultural work and receives flavoring work. The grievance is not about the product, it is about what is said of it. And the laboratory result cited above makes this grievance serious: since the cup does not betray the addition, the producer’s declaration remains, in practice, the only information the buyer has.

The flavoring of green coffee sold as a fermentation belongs to this same family, more plainly so. No microorganism works in it. Selling such a lot as an experimental fermentation is not a nuance of vocabulary, it is attributing work to someone who did not do it.

Finally the confusion of vocabulary, which is not an offense but which makes everything else impossible to discuss. Three words serve to avoid saying what was added: natural, experimental, innovative. Two names circulate that are more precisely misleading, “experimental fermentation” and “fruity yeast”.

The word anaerobic belongs to the same family, for another reason and without bad intent. It designates in principle an environment without oxygen. But nobody drove the air out of the tank: it was closed, and the microorganisms consumed the oxygen present. The word is not false by intent, it is used without its meaning.

Where the line runs

The debate on co-fermentation rarely concerns a described practice. It concerns a border. Before saying who is right, one must look at where this border is actually drawn, and by which texts.

The texts of the world championships say the same thing, word for word, and they can be read online. The World Barista Championship rules carry clause 3.1 C, the World Brewers Cup rules clause 3.3 B. Both forbid any addition after the green coffee stage, a stage defined as the seed of the genus Coffea, dried during post-harvest, and stripped of all the layers of the pericarp, that is, of all the coverings of the fruit. Both texts then name the same families of inputs: aromatic substances, flavorings, fragrances, liquids, powders. The 2024 and 2026 versions of the baristas text are identical. The Brewers Cup also closes the entire chain: the grinder must add nothing to the coffee, nor the brewing device, and the serving vessels must transmit no aroma. The rule therefore does not target a product, it protects a chain.

What these texts do not forbid must be stated with the same clarity, because it is almost always passed over in silence. Everything done while the coverings of the fruit are still there escapes the clause. A lot co-fermented with a fruit during fermentation therefore remains perfectly admissible in these two competitions. No competitor served one at the 2024 World Barista Championship, but nothing forbade it.

The Cup of Excellence changed its eligibility rules on October 23, 2023, for the first time since 1999. Three categories replace the single ranking: dry process, wet process, and experimentals. In the experimental category, yeasts are admitted, indigenous or inoculated; no additive other than water is. The principle stated publicly by the management of the Alliance for Coffee Excellence is broader: what does not belong to the coffee chain is not admitted. It is this principle, and not a judgment on the intensity of the result, that explains the fate reserved for mosto. An outside fruit brings molecules that exist nowhere in the plant; mosto concentrates those that are already there. The line is drawn on the origin of the matter.

Best of Panama went further in 2024: the Panamanian specialty coffee association set aside four coffees, its statement saying that these lots had been found altered from their natural expression, with the probable intent of scoring higher, and asking that altered or infused coffees be ranked separately.

So much for the rules. Now comes the argument that puts them in difficulty, and it is a solid one.

Those who reject co-fermentation accept, in their great majority, a series of heavy interventions on fermentation:

• inoculation of a selected yeast, that is, the deliberate addition of a microorganism chosen in a laboratory and bought from a supplier;
• fermentation in a closed tank for ninety hours or more;
• thermal shock, two successive washes at widely separated temperatures, which stops microbial work dead;
• the addition of mosto, sometimes at thirty or fifty percent of the volume.

None of these four practices is closer to non-intervention than a fruit addition. All modify the microbial population, the pH or the duration. Christopher Feran, a consultant and green coffee buyer, made this argument publicly in September 2021: pulping a cherry and then fermenting it for several days in a chamber swept with carbon dioxide is held to be natural, and Saccharomyces cerevisiae, the reference yeast, is native to no coffee-growing terroir. The criterion of naturalness therefore does not stand alone, and the line runs partly between familiar intervention and new intervention, which is not a criterion.

There is, however, a technical version of the line, which holds up better. Diego Robelo, general manager of the finca Aquiares in Costa Rica, puts it this way: inoculating a yeast means introducing a microorganism that will digest the sugars already present in the fruit and the mucilage of the coffee; co-fermenting means introducing in addition a new source of sugar, and with it the microorganisms the fruit carries. The difference is not one of degree, it is one of matter, since part of what ferments is no longer coffee. Marty Pollack, co-founder of Torch Coffee, answers that fermentation happens anyway and that the quarrel over vocabulary matters less than the declaration.

The arguments of the opposing camp are not reducible to a purity reflex either, and four stand up.

• Traceability and the value of origin. Specialty coffee rests on an economic promise: a cup is tied to a plot, an altitude, a variety and a producer, and it is this chain that the buyer pays more for. Kyle Ramage, co-founder of Black & White Coffee Roasters and former US barista champion, describes a shift in which the competition of cultivation becomes a competition of transformation.
• Masking. A powerful aroma can cover a defect, a low altitude or an ordinary variety. This is the reproach of the Panamanian association, which speaks of an artificial homogenization of different terroirs. It is plausible and verifiable in principle, but no published analysis compares a co-fermented lot with the same coffee without an addition.
• The race for intensity, that is, the fear of a shift in the reference taste. The concern is legitimate and is backed by no measurement.
• What goes back to the producer. If the value created comes from the method, it belongs to the producer who found it; if it comes from the added ingredient, the origin premium is no longer paid for the origin. Neither reading is quantified.

One last argument circulates that is worth hearing by those who demand transparency of the producer alone. Chris Kornman, education manager at an American importer, observes that roasters almost never publish the composition of their blends, and that a blend sold under the name of a region may contain only a fraction of it. The demand for transparency is addressed to the one who produces, not to the one who resells.

So where does the line run? On what the label says, and not on what goes into the tank. This position is not a soft compromise, it follows from three facts established above. The first is that microbial management alone already reaches the profiles for which additions are blamed, so intensity proves nothing. The second is that the interventions held acceptable are as heavy as those held unacceptable, so naturalness proves nothing either. The third is that the cup does not betray the addition, so the buyer has only the producer’s word.

What remains is simple to state and demanding to keep: what went into the tank must be written down, with its kind, its timing and its dose. A declared lot is judged on what it is. An undeclared lot is judged on what it hides.

And this must be added, because it is what is read least. The demonization of these coffees has a cost, and this cost is paid by the producers who do the most careful work. The protocols described in the step-by-step account of this sheet are, by their demand for measurement, among the most rigorous in the trade: they include a harvest Brix, a pH floor, a held temperature, a dose relative to the mass, a stopping gesture and a controlled drying. Very few classic washed lots are documented at this level. Filing everything under the word cheating amounts to confusing the person who pours a bucket of syrup onto a patio with the one who matches the Brix of the starter to that of the cherry. They are two different trades, and only one of the two deserves the reproach.

Benchmarks

• World Barista Championship rule, clause 3.1 C, 2024 and 2026 texts identicalno additive after the green coffee stage
• What this clause does not forbidaddition during fermentation, as long as the coverings of the fruit are there
• Cup of Excellence, overhaul of October 23, 2023, the first since 1999three categories, no additive other than water
• Published controlled trials of coffee co-fermented with a fruit or a spicenone among 164 Europe PMC references
• Scores obtained by simple yeast inoculation, with no fruit added83.72 and 84.28 points
• Score obtained by co-fermentation of two yeasts with ultrasound, with no fruit added82.16 points
• Noreña protocol, starter dose relative to the mass10 milliliters per kilo, or about 1 percent
• Noreña protocol, mosto at time of collectionpH 4.2 and 11.5 Brix
• Sánchez protocol, mother culture80 liters, 10 yeast strains, 190 hours, pH held above 4.0
• Bermúdez protocol, tanks48 h at 18 degrees then 48 h at 21 degrees; one lot at 20 psi for 36 hours
• Bermúdez protocol, drying35 degrees at 25 percent relative humidity, 34 hours, down to 10 to 11 percent
• Harvest Brix required in the protocols readabove 20, up to 23
• Best of Panama 20244 coffees set aside for alteration, no analysis report published
• Isotopic detection of an added aroma, carbon 13 on N-methylpyrrole-2-carboxaldehydefrom -35.0 to -31.1 per mil in coffee, from -27.9 to -23.9 per mil when the molecule is added
• Home infusion described on green beanscinnamon sticks, 3 to 5 days