The chaff in your cooling tray is the coffee seed's own last skin, and it is the only part of the coffee fruit that makes the whole trip to your roaster. The cherry's outer skin, the pulp, the sticky mucilage and the parchment shell all come off at a mill in the growing country and stay there. The silverskin does not. It clings to the seed through fermentation, drying, hulling, bagging and the crossing, and it lets go in the drum, when the bean swells past the point the skin can hold. So the flecks drifting in the tray are the last of the fruit, and they are the only piece of it a roastery in the United States ever handles.
What is the chaff that comes off coffee beans?
It is silverskin, dried and shed. Sweet Maria's puts it plainly in their coffee library, and their wording is the one to use: chaff is "paper-like skin that comes off the coffee in the roasting process," and it is "part of the innermost skin (the silverskin) of the coffee fruit that still cling to the beans after processing has been completed." Sweet Maria's was the first import partner Inheritance ever bought from, back when we were ordering green coffee in twenty-pound lots, so we owe them the credit twice over.
The detail their entry adds, which is worth checking against your own beans, is where the silverskin sits: it "lines the crease on the flat side" of the green seed. Pick up a green bean and look at the center cut. That pale line running down it is the skin that ends up in your tray. In the research literature it gets a drier name. Bessada, Alves and Oliveira, in a 2018 review in Cosmetics, call it "a thin tegument that directly covers the coffee seed," and every paper since has used some version of that phrase.
What it is not is burnt bean. New roasters watch the tray fill with brown paper and assume something scorched. Nothing scorched. A layer that was always there came off, on schedule, the way it is supposed to.
Why does chaff come off in the roaster instead of at the mill?
Because heat is what finally separates it, and heat is the one step that happens after the coffee has already crossed an ocean. Every other layer of the cherry is removed wet or removed mechanically at origin. The skin and pulp come off at the wet mill; the mucilage is fermented or scrubbed away; the parchment is hulled off once the seed is dry. We walk through that whole sequence in our piece on natural process coffee, and the thing to notice is where it stops. It stops at the silverskin.
Mills used to go one step further. Sweet Maria's records that dry mills once polished coffee to strip the silverskin off for appearance, and that the practice is now discouraged, because the heat of polishing damages the cup. So the skin stays on, and it arrives in Pensacola still attached.
Then the bean expands. Candela-Salvador and colleagues describe the mechanism in a 2026 paper in Foods in one sentence: "During the roasting process, the coffee bean increases in size, and the silverskin, a thin tegument that covers the green coffee beans, is detached." Bessada's review says the same thing the same way. The seed grows, the skin does not, and airflow through the drum carries the pieces out.
That is why Gottstein and colleagues, writing in Foods in 2021, describe silverskin as a by-product "of the roasting process, which typically no longer takes place in the growing countries" — and why Candela-Salvador's team calls it "the only co-product of the roasting stage." Coffee generates an enormous amount of agricultural waste, and essentially all of it stays where the coffee was grown. Chaff is the exception. It is the one piece of the fruit whose disposal is a problem for the country that drinks the coffee rather than the country that produced it.
How much chaff does a roast make?
Less than it looks like, because chaff is almost weightless and takes up a great deal of room. Both of the careful recent measurements land near four or five percent, and then attach that percentage to different things. The difference matters more than the numbers do.
- Gottstein and colleagues (2021) report that silverskin "represents 4.2% (w/w) of the coffee cherry."
- Candela-Salvador and colleagues (2026) report that it "represents approximately 4–5% of the bean weight."
A whole cherry weighs several times what the dried seed inside it weighs, so those two figures are not the same claim in different clothes. Taken literally, 4.2 percent of a cherry is a great deal more silverskin than 4 to 5 percent of the bean, and the two cannot both be describing the same material. Use the bean-weight version at a roaster, because the bean is what you weigh into the drum, and treat four to five percent of your charge as the working figure.
Globally the total is real enough to have been counted. Boninsegna and colleagues, in Foods in 2025, put annual silverskin production at "between 200,000 and 400,000 tons per year." A figure of two billion tons of coffee by-products a year also circulates, and it does not survive arithmetic. The same paper puts total world coffee production at around eleven million tons. A by-product stream cannot be two hundred times the crop it comes off.
What is chaff actually made of?
Mostly fiber, and more protein than you would guess from handling it. Gottstein's team ran a full characterization on silverskin from several sources and found total dietary fiber between 59.1 and 67.0 grams per hundred grams, of which the great majority was insoluble, 53.2 to 56.0 grams, with 8.8 to 13.0 grams soluble. Protein came in at 17.8 to 22.2 grams per hundred, ash at 8.15 to 11.24, and fat at only 1.50 to 1.82.
Candela-Salvador's 2026 characterization across origins found total fiber higher still, 71.81 to 76.86 grams per hundred, with insoluble fiber at 54.02 to 60.58 and protein lower, 12.37 to 14.21 grams for the arabica samples, with the highest protein readings in Coffea canephora. Fat 1.20 to 2.10, ash 4.23 to 5.34, moisture 4.50 to 6.41.
There is caffeine in it, and chlorogenic acid. Bessada's review gives chlorogenic acids at 1 to 6 percent and caffeine at 0.8 to 1.25 percent. Candela-Salvador measured caffeine at 0.81 to 7.32 milligrams per gram, which is 0.08 to 0.73 percent. Put the two figures side by side in the same units and they do not overlap at all.
The short version, if you want one sentence to carry away: chaff is roughly two-thirds dietary fiber, carries a real amount of protein, and is not caffeine-free.
Why do the published figures disagree?
Because "chaff" is not one substance and the papers are not measuring one thing. Candela-Salvador's study exists precisely because origin changes the answer; their protein range splits along species, with canephora reading higher than arabica. Roast degree moves it too, since the skin comes off a light roast in a different condition than it comes off a dark one. And the denominators are inconsistent across the literature, as the cherry-versus-bean split above shows.
I would rather leave that disagreement standing than average it into a clean number. The honest statement is that silverskin is a fiber-dominant material with meaningful protein and detectable caffeine, and that anyone quoting a single decimal figure for what chaff contains is quoting one paper's samples. Our own approach to roast data is the same, and it is why we publish roast color grading as a practice rather than publishing a number we would then have to defend as universal.
Where does chaff go after the roast, and why is that where fires start?
Into a cyclone, on most commercial machines. Airflow pulls the chaff out of the drum and into a collector that drops it into a bin, which is the piece our roasting guide calls a chaff collector when it lists the gear a new roaster will meet. On a home machine it is a screen or a tray, and you empty it by hand.
Here is the part that turns a housekeeping job into a safety one, and it belongs to Scott Rao, who states it more directly than anyone else in the trade. Writing about plumbing water lines into a roaster, he says the second line "should be plumbed into the chaff collector (cyclone), as that is where most roaster fires occur." Not the drum. The bin of dry paper sitting in the hot exhaust path.
That is the whole argument for emptying it, and it reframes what a chaff clean is for. It is not tidiness and it is not about the next person's cup. It is about the one part of the machine that holds a large volume of extremely dry, extremely light organic material directly downstream of several hundred degrees of air.
Worth saying plainly: our own guide to roaster ventilation requirements covers the approvals, the vent run and the clearances, which is the half a code official cares about. The cyclone is the half that is yours every single roast day, and no permit checks it.
Does chaff in the grinder change how the coffee tastes?
Barely, and mostly it is a static problem rather than a flavor one. Chaff is dry and light enough to pick up a charge, so it clings to the grinder bin and the burr chamber and gets everywhere except where you want it. A small amount reaching the brewer is normal and is not what is wrong with a disappointing cup.
There is a real limit on that reassurance, though, and it is worth stating rather than glossing. Silverskin is not roasted coffee. It is a different tissue with a different composition, and it did not go through the browning that gives roasted coffee its flavor — the melanoidin chemistry we wrote about in what makes roasted coffee brown. It contributes very little because there is very little of it and because what there is has almost nothing to contribute. Those are two different reasons and only the first one is about quantity.
What can you actually do with chaff?
More than throw it away, and the best plain-English survey of the options is MTPak Coffee's write-up on reusing it, which covers compost and fertilizer, fuel briquettes, food uses and even automotive composites. The composition figures above are why any of that works: a material that is two-thirds fiber and carries real nitrogen-bearing protein is a soil amendment before it is anything else.
The caution is scale. Almost all of the published valorization work is written around industrial streams measured in hundreds of thousands of tons, and the chaff sitting in a small roastery's bin at the end of a Wednesday is not that stream. Composting is the one item on that list that scales down to a roastery rather than a factory. Most of the rest of the list is a factory's answer to a factory's problem, and reading it as a business opportunity is how a roaster ends up storing sacks of paper they will never process.
Why is a chaff clean a rule in a shared roastery?
Because a shared machine fails in exactly the places a solo roaster is allowed to be lazy. We run the Inheritance Roastery as Pensacola's only co-roasting space, and the page describing it puts the failure mode in one line: a shared roaster fails the first time somebody runs it hot, skips a chaff clean, or hands it over worse than they found it. That is why every self-roaster here completes a basic roasting class before touching a machine, with no exemption for experience, and the class covers cleaning and maintenance expectations alongside safe operation and roast color grading.
That rule costs us business, and we know what it costs. Somebody who wants to walk in and roast today cannot, and some of them do not come back. We think it is the right trade, and if the class is the part that does not work for you, our wholesale program is the cheaper answer and we would rather say so now than after you have sat through it.
Mason on why the program exists at all: "The co-roasting program has been good for me in many ways, but the one that stands out the most is the systematization of all of the coffee roaster startup fundamentals." He means the things that do not make it into a guide: what it takes to get through inspection and permitting with the city, what it takes to learn the fundamentals, and what it takes to get over the sinking feeling of your first real mistake on a machine. His words for the point of the room: "creating a way for new roasters to learn and experience all of those things in a safe and controlled environment surrounded by supportive people."
The chaff clean is one of those fundamentals. It looks like janitorial work, it takes a few minutes, and it is on the list of house rules for the same reason a fire extinguisher is on the wall. Our rental guide for the Mill City MCR-15 tells you to plan time after your session to cool, vent and remove chaff, then check the drip trays and airways. That sentence has always been correct and it has never explained itself. Now it has: the bin you are emptying is the most likely place in the building for a fire to start, and the paper you are emptying out of it is the last of the coffee fruit, which traveled further than any other part of it to end up in your hand.
Roast with us, or have us roast for you. The Inheritance Roastery rents time on commercial roasters in Pensacola, and every self-roaster completes our basic roasting class first — there is no exemption for experience. If you would rather skip the class entirely, toll roasting carries no class requirement and we run the machine ourselves.
