The Decaf Paradox | Skeleton Brew Coffee
Share
The Decaf Paradox: How Removing Caffeine from Coffee Became a Scientific Race
Ludwig Roselius thought caffeine had killed his father, so in 1903 he paid people to get it out of coffee. They used benzene. It worked, and it started an argument that still isn't settled.
Caffeine is only about one to two percent of a green coffee bean. That's the entire thing you're trying to remove. The other 98 percent, including a few hundred aroma compounds that turn into roughly a thousand more once you roast them, is the part you want to keep.
Which makes decaf sound easy, and it isn't, for one annoying reason. Caffeine dissolves in water, and so does nearly everything that makes coffee taste like coffee. Anything aggressive enough to pull the caffeine out will cheerfully pull the flavor out too. That's the whole problem, and people have been chipping away at it for over a century.

Somebody had to find it first
In October 1819 a young chemist named Friedlieb Ferdinand Runge visited Goethe in Jena and demonstrated how belladonna dilates a pupil, using a cat as the subject. Goethe, charmed, handed him a box of coffee beans someone had sent from Greece and told him to go experiment. He wasn't just being generous. He wanted to know what in those beans was keeping him up at night.
Runge found it, called it Kaffebase, and by 1820 pure caffeine existed as something you could put in a dish and look at. Then about eighty years went by, because knowing what caffeine is turns out to be very little help in getting it out of a bean you still want to drink.
The merchant with a grudge
Ludwig Roselius was born into the Bremen coffee trade in 1874. His father drank a lot of coffee and died at 59, and the doctor blamed the coffee. Roselius took that diagnosis seriously and went looking for a fix.
The story everyone tells is that a shipment of his beans showed up soaked in seawater and he noticed the caffeine was mostly gone while the flavor wasn't. It's a good story. It also never seems to come with a source, and it has the tidy shape of something reconstructed backwards from the result, so treat it as company legend.
What's documented is that in 1903 Roselius and a team of researchers worked out a method that actually scaled: steam the green beans with brine and acids to open the pores, then rinse them repeatedly with benzene. The caffeine leaves with the benzene. The patent came in 1906.
That same year he founded Kaffee Handels-Aktien-Gesellschaft, better known as Kaffee HAG, and gave it a red lifebuoy as a logo. He wasn't selling coffee with a feature removed. He was selling a rescue.
Selling the absence of something
Naming it was awkward from the start. Most of Europe got Kaffee HAG. France got Café Sanka, from sans caféine. The US got it around 1909 and 1910 under the genuinely terrible names Dekafa and Dekofa.
Then the First World War happened, the American arm was seized by the Office of Alien Property Custodian, and it came out the other side as an independent US company. Sanka launched properly in 1923, at first only through two Sanka Coffee Houses in New York. General Foods bought it in 1932 and fixed the which-pot-is-which problem by handing out coffee pots with the handles and lids painted bright orange. That's where the orange decaf pot comes from.
Roselius spent the money on Bremen, rebuilding the Böttcherstraße as an architectural showcase, backing artists including Paula Modersohn-Becker and Bernard Hoetger, and publishing the collectible Kaffee HAG albums. His politics were worse than his taste in art. He was sympathetic to National Socialism and applied to join the Nazi party twice, and was rejected both times because the art on his own street was considered degenerate. Hitler attacked both the street and Roselius's Nordic theorizing at the Nuremberg rally in September 1936. He died in Berlin in 1943.
His process didn't outlast him by much. Benzene is a known human carcinogen, so nobody decaffeinates with it now. The man who invented decaf to protect people from a poison made it using one.
Four ways to be pickier
Everything since 1903 has been an attempt at better aim. Four approaches survived, and all four are in commercial use today.
Better solvents. Methylene chloride (dichloromethane) and ethyl acetate took over from benzene. You can apply them straight to steamed beans, but the gentler and more common version is indirect: soak the beans in hot water, treat the water with the solvent, then give the beans their own water back so they can reabsorb the flavor compounds they lost. The beans never touch the solvent.
Water only. Developed in Switzerland in 1933, then more or less shelved until the Swiss firm Coffex S.A. made it commercially viable around 1980. It works by osmosis. You saturate green coffee extract with everything except caffeine, pass it over fresh beans, and caffeine is the only thing with anywhere to go. The North American operation started in 1988 to supply Nabob and still runs out of Burnaby, BC, which quietly makes Canada the home of solvent-free decaf.
Supercritical CO2. This one came out of coal research. In 1967 Kurt Zosel, at the Max Planck Institute for Coal Research in Mülheim, noticed that caffeine dissolves in carbon dioxide pushed past the point where it behaves as either a liquid or a gas. He patented it in 1970. Moistened beans go in with the CO2, which is very interested in caffeine and mostly indifferent to flavor, and the CO2 is then cooled and run through activated carbon to catch what it picked up. The method now handles something like 100,000 tons of raw material a year.
Sugarcane. At Descafecol in Caldas, the only decaffeination plant in Colombia, sugarcane molasses is fermented into ethanol and acetic acid and combined into ethyl acetate, which does the extracting. Same compound as the solvent method, reached by a route that smells like a distillery. It often gets labeled "natural" because ethyl acetate also shows up on its own in fruit, wine and beer.
Decaf has never meant caffeine-free
It means caffeine-reduced, and the two biggest markets don't define it the same way.
| European Union | A hard ceiling: no more than 0.1% caffeine by weight on a dry basis in roasted coffee, 0.3% in extracts and instant. |
|---|---|
| United States | No federal standard for the word. The industry convention is a claim that at least 97% of the caffeine has been removed. |
In practice, a cup of regular coffee runs 95 to 200 mg of caffeine and a cup of decaf lands in the low single digits to the low tens. It won't keep you up. It also isn't zero, which matters if you've been told to be at zero.
Now the actual paradox
The caffeine doesn't get thrown away. It gets purified into crystalline caffeine and sold to pharmaceutical companies, cosmetics companies, and above all to cola makers, whose caffeine comes mostly, and sometimes entirely, from coffee that was decaffeinated so somebody else could sleep.
Decaf is only about 10 percent of the global coffee market, and even so there isn't enough recovered caffeine to go around, so the shortfall gets made up with synthetic caffeine. One researcher put the obvious question right in a paper title: which is the by-product, the caffeine or the decaf coffee?
A man who was certain caffeine was poison ended up building the supply chain for it. Nobody planned that. It's just what happens when the thing you're removing is worth more than the thing you're keeping.
Still unsettled
In December 2023 a coalition including the Environmental Defense Fund, Breast Cancer Prevention Partners, the Center for Environmental Health and the Environmental Working Group asked the FDA to pull the food-additive approvals for four solvents: benzene, ethylene dichloride, trichloroethylene and methylene chloride. The last one is the one with money attached, because a lot of decaf is still made with it.
Methylene chloride remains legal in US decaf at up to 10 parts per million while the review goes on. The FDA reopened public comment in May 2026, closed it on June 29, 2026, and hasn't ruled. The National Coffee Association opposes the petition, noting that around 10 percent of American adults drink decaf daily, mostly the solvent kind, and arguing it's safe at the current limit. So 120 years in, the live question is still which solvent is acceptable.
Why decaf tastes better than it used to
Decaf earned its reputation, and not only through extraction. For decades it was made from the cheapest green coffee available, on the fairly safe assumption that nobody drinking it was paying close attention. Decaf beans are also harder to roast: the processing leaves them drier, more brittle and structurally different, and they brown faster, so a normal roast profile takes you well past where you meant to stop. That's the scorched cup people remember.
Both problems are getting fixed. Roasters now buy good green coffee specifically to decaffeinate, use the gentler water, CO2 or sugarcane routes, and build roast profiles for the bean in front of them. The old complaint that decaf "isn't real coffee" was never really about the caffeine. It was about how much else came out with it, and that number has been shrinking for 120 years.