Green coffee has almost no flavor: James Hoffmann describes its taste as unpleasant and vegetal. Roasting creates most of the aroma in your cup. Across all coffees it can produce more than 800 volatile compounds, he writes, though any one coffee has only a selection of them. This page covers how that happens: the machines, the stages a roaster watches, the chemistry in brief, and what roast speed and cooling change.
It’s one of three roasting pages. Coffee roast levels explained covers what light, medium and dark mean and how to read a bag, and how to roast coffee at home is the method if you want to try it yourself.
Our 2020 video predates this revision; where the two differ, follow the text.
The machines
Most coffee is roasted in one of two kinds of machine, Hoffmann writes:
- Drum roasters. A metal drum turns over a flame, keeping the beans moving so they roast evenly. The roaster controls the flame, and the airflow through the drum, which sets how quickly the heat reaches the coffee. Drums can roast at slower speeds, which makes them popular with craft roasters. The largest hold around 1,100 lb (500 kg) per batch.
- Fluid-bed, or hot-air, roasters. Invented by Michael Sivetz in the 1970s, they tumble and heat the beans with jets of hot air. The larger volume of air moves heat into the coffee faster than a drum does, so roasts are shorter, and the beans tend to swell a little more.
Big industrial plants also use tangential roasters, large drums with internal shovels to mix the beans, and centrifugal roasters, which can roast very large batches in as little as 90 seconds. Hoffmann notes that roasting that fast keeps weight loss down and leaves more to extract from the beans, which matters for instant coffee, but it isn’t designed to make the best cup. Home roasters are small versions of the drum and the hot-air roaster.
The stages of a roast
A roast passes through the same stages whatever the machine. How fast a coffee moves through them is its roast profile, and many roasters log it so they can repeat a roast closely. The stages below follow Hoffmann’s account in The World Atlas of Coffee.
Hoffmann’s shorthand is that coffee won’t turn brown while its water is still there. In practice, browning speeds up as the beans dry, and he stresses how much the first two stages matter. If the coffee isn’t dried properly, it won’t roast evenly afterwards: the outside of each bean ends up well roasted while the inside is underdone, and the cup tastes bitter and sour and grassy at once. Slowing the roast later can’t fix that, he writes, because different parts of the bean are already moving at different rates.
At first crack, a roaster also sees the bean temperature climb more slowly, even with the same heat going in. Hoffmann warns that not adding enough heat at that point can stall the roast and “bake” the coffee.
Temperatures: every thermometer reads its own
Jessica Easto’s Craft Coffee gives the usual textbook figures: the Maillard reaction between 302 and 392°F (150–200°C), caramelization from 338 to 392°F (170–200°C), first crack at about 385°F (196°C), and second crack around 414 to 424°F (212–218°C). Use them as a rough map, not as settings. A roaster’s display shows what its probe measures, and probes differ:
- The air. Rob Hoos, a production roaster, notes that a drum roaster’s charge temperature, before the beans go in, is read by the probe in the exhaust air. He calls it an approximation of the air’s temperature after it has passed through the drum.
- The beans. The bean probe sits in the tumbling beans. Hoos writes that its readings vary with the thermocouple’s type, thickness and placement, and advises noting the temperature at which the beans turn yellow on your own roaster and using that to standardize your notes.
How far apart two machines can read is clear from one study. Christina Rune and colleagues roasted five specialty coffees on a roaster that has both a bean probe and an infrared sensor, and ended their lightest roasts on the bean probe’s reading of 336°F (169°C), between 0 and 28 seconds after first crack. That’s about 49°F (27°C) below Easto’s figure for first crack (calculated). A temperature from one roaster, or one book, won’t transfer to another, which is why Hoos’s advice is to learn your own roaster’s numbers.
The chemistry, in brief
Most of coffee’s aroma is made by three groups of reactions, Hoffmann writes: the Maillard reaction, caramelization and Strecker degradation.
- The Maillard reaction. Sugars react with amino acids in a cascade of steps that makes much of coffee’s aroma and, at the end, the brown polymers called melanoidins. It’s the same family of browning reactions that happens when meat is roasted, and Hoffmann notes that it happens when cocoa is roasted too. Our sister site covers how cacao beans are roasted.
- Strecker degradation. A side branch of the Maillard reaction that turns amino acids into aldehydes, several of them among the odorants that matter most in a roasted coffee’s aroma.
- Caramelization. Sugar breaking down on its own. The sugars that caramelize become less sweet and eventually start to add bitterness, Hoffmann writes, and by the end of first crack few or no simple sugars are left.
Our research note on Maillard, caramelization and Strecker maps all three in detail. Two more threads run alongside:
- Chlorogenic acids. Roasting breaks most of them down, and what they become accounts for much of a coffee’s bitterness. Farah and colleagues (2005) found that the main bitter lactones peaked at a light-medium roast of about 14% weight loss. Thomas Hofmann’s group found that harsher phenylindanes dominate in darker roasts. The chlorogenic acids note follows both.
- Melanoidins. In one study, the amount in the brew rose from a light to a medium to a dark roast of the same coffee (Borrelli and colleagues, 2002). The melanoidins note covers what they do for color, crema and body.
With so little sugar left, why can a roasted coffee taste sweet? The sweetness note looks at the evidence.
Because the bitter compounds are made in the roaster, a bitter cup can come from the roast rather than the brew. Timing the brew is the quickest way to tell them apart.
Fast or slow
Hoffmann describes a roast as the product of two things: the final color of the beans, and the time it took to get there. Calling a coffee a light roast isn’t enough, because two coffees roasted to the same color, one fast and one slow, taste different.
- Weight. Roasting drives off water and other compounds. Hoffmann writes that slow roasting, over 14 to 20 minutes, loses more weight, about 16 to 18%, than fast roasting does, and the very fastest industrial roasts lose the least. He also judges slow roasting the better, if more expensive, cup.
- Acidity, sweetness and bitterness. It’s generally agreed, Hoffmann writes, that the longer a coffee is roasted, the less acidity it ends up with, while bitterness rises slowly with roast time and definitely with darkness. Sweetness peaks in between. Hoos puts the balance of acids mainly down to the time after first crack, and how far caramelization goes mainly down to the end temperature.
- Too fast and too slow. The Specialty Coffee Association’s cupping guidance describes the two common roasting faults. Roasted too fast, the center of the bean doesn’t develop as far as the surface, which looks much darker than the inside. Roasted too slowly, the bean “bakes”: it turns brittle, and the flavor and aroma are subdued, often with grain notes.
Hoos’s own roasts give a sense of the proportions. On his drum roasters, the drying phase tends to take 43–51% of the roast, the stretch from yellowing to first crack 32–44%, and development 12–20%, though he adds that yours needn’t fall in those ranges. In his cupping experiments, a longer stretch from yellowing to first crack gave a heavier body in most of the coffees he tried, with exceptions he reports himself.
Cooling
A roast doesn’t stop when the heat does. Hoffmann writes that the beans must be cooled quickly to prevent over-roasting or baked flavors. Small batches usually go into a cooling tray that draws air rapidly through the beans. With large batches, air alone isn’t enough, so a mist of water is sprayed on and pulls heat out as it evaporates.
Done correctly, he writes, misting doesn’t hurt the cup, though the coffee ages a little faster. But some companies add more water than they need, to make the beans weigh more, which he calls unethical and bad for cup quality.
After the roast
Freshly roasted beans are full of carbon dioxide, which comes out over the following days, and their aroma fades over the following weeks. Hoffmann writes that coffee is at its best when used within a month of roasting, which is one reason it’s usually roasted close to where it will be drunk: green coffee keeps far longer. The freshness guide covers the timeline, and the Freshness tool works out where your own bag is.
Sources
- James Hoffmann, The World Atlas of Coffee, 2nd ed. (Mitchell Beazley, 2018): “Coffee Roasting,” p. 54 (“Fast or slow, light or dark?”), pp. 56–57 (“The stages of roasting”), p. 59 (“Sugars in coffee”; “Aromatic compounds in coffee”) and “Quenching”; “Types of Coffee Roasters” (drum, fluid-bed, tangential and centrifugal roasters)
- Jessica Easto, Craft Coffee: A Manual (Agate Surrey, 2017), ch. 3, “Roast,” pp. 146–147 (reaction temperatures, first and second crack)
- Rob Hoos, Modulating the Flavor Profile of Coffee: One Roaster’s Manifesto (self-published, 2015), “Charging Toward the Beginning of Chemical Reaction,” pp. 16–17 of the PDF edition (exhaust and bean probes, the yellowing temperature), and “Maillard Reaction,” pp. 20–26 (the cupping experiments)
- Rob Hoos, Cultivar: A Practical Guide for Coffee Roasters (self-published, 2025), p. 53 (his phase shares)
- Christina J. Birke Rune and colleagues, “Acids in brewed coffees: Chemical composition and sensory threshold,” Current Research in Food Science 6 (2023): 100485, section 2.1 and Table 1 (roaster, bean-probe end temperatures, development times): doi.org/10.1016/j.crfs.2023.100485
- Specialty Coffee Association, A System to Assess Coffee Value: Understanding the SCA’s Coffee Value Assessment (June 2024), §5.1.1, “Roasting,” p. 21 (“Avoiding Roasting Problems”)
- Adriana Farah, Tomas de Paulis, Luiz C. Trugo and Peter R. Martin, “Effect of roasting on the formation of chlorogenic acid lactones in coffee,” Journal of Agricultural and Food Chemistry 53, no. 5 (2005): 1505–1513, abstract: doi.org/10.1021/jf048701t
- American Chemical Society, “Battling Bitter Coffee: Chemists Identify Roasting As The Main Culprit,” news release on Thomas Hofmann’s presentation at the 234th ACS National Meeting, as published by ScienceDaily, August 22, 2007: sciencedaily.com
- Rosa Cinzia Borrelli, Attilio Visconti, Carmela Mennella, Monica Anese and Vincenzo Fogliano, “Chemical characterization and antioxidant properties of coffee melanoidins,” Journal of Agricultural and Food Chemistry 50, no. 22 (2002): 6527–6533, abstract: doi.org/10.1021/jf025686o