Where Does Sweetness Come From? Part Three: Roasting

Coffee should be sweet. Ok, so you have a wonderful, naturally sweet, perfectly harvested green coffee, that was then carefully processed so as to preserve that beautiful sweetness. Now what?

Well, now it's time to roast that coffee. Roasting is a very complicated process. There are hundreds of different organic compounds within a single coffee bean, and every single one of them reacts differently to the process of roasting. The most important process to understand when it comes to the formation of sweetness and other flavors in coffee is what is called a Maillard Reaction. This is the process of amino acids interacting with sugars to form new compounds in the course of applying heat to a food.

One way to think about how roasting affects sweetness is by the analogy of caramelizing sugars. If you have ever sauteed onions, then you know this process. As the onions are cooked further, they become sweeter and sweeter due to the caramelization of the sugars in the onion. This is actually another form of the Maillard Reaction.

Generally speaking, as you roast coffee it gets sweeter and sweeter, up to a point. Behold my mighty MS Paint skills! (click it to enlarge.) So in a very light roast (or cinnamon, or "cupping" roast), fewer of the sugars have caramelized, leading to a brighter-toned, more sour cup. As you proceed darker and darker, the Maillard Reactions continue, producing more and more sweetness in the cup.

However, there are diminishing returns on this once you begin to carbonize some of the cellulose and other starches in the bean. Sometime after "second crack" in the roasting process, dried-out portions of the bean begin to carbonize (turn to char and ash). This process can go on even while Maillard Reactions continue. This is why the best dark roasts can be very sweet without being too ashy. But it must be done with extreme care and skill. Very quickly the ashy bitterness can take over the cup, and carbonization can begin to kill those sugars the roaster was trying to cultivate. To go back to our onion analogy, this is like leaving the onions on the pan until they are black and smoking and dry.

As you can see from the rough schematic graph above, roasting is a balancing act. The graph I drew is extremely basic, and doesn't necessarily apply to all types of beans or machines. But it gives you an idea of how different factors are changing at different rates over time. A skilled roaster has to choose which spot along the curve he or she wants to arrive at (something that is best determined on the cupping table, of course, with many many samples of roasted coffee).

Not only that, but the roaster must choose how to arrive at that point. One cannot simply pick a time and temperature and then walk away. Because different chemical reactions are happening at different times during the roasting process, the roaster must also pick the "curve" he or she will use to reach the desired roast degree. Will it be a fast-ramping curve, a slow-building one, or a complicated S-curve that brings out just the right sweetness while preserving acidity and the more delicate aromatics?

The answer to all these questions would take (and has taken) an entire book to investigate fully. But I hope this little primer helps explain where sweetness comes from in the roasting process.

Our little bean has gone a long way since being planted in the ground at the farm, but it still has a long way to go before it's a perfect cup of sweet coffee...

Comments

[Youri Vlag]

Hi Daniel,

I just want to say what a great blog and you must be keeping yourself busy with it lately!

Cheers,
Yorui

[Daniel Humphries]

Thanks Youri!

[Colin O'Callaghan]

Hi Dan,
Loved the article, very informative! I just have one thing I would change on the graph and it's about the Time/temperature label on the x-axis. Time and temperature are not proportional like that in the coffee roasting process, sometimes as the time goes on the temperature stays the same or rises more slowly or more quickly. A graph with an x, y and z axis would have been a more accurate way to go but that would have been an extremely complicated graph, as it is I think the x-axis should only have the temperature label, so as not to confuse the reader. As you pointed out, "one cannot simply pick a time and temperature and then walk away." Anyway, I did really like your article that was just what I was thinking as I read it. What are your thoughts?

Colin O'Callaghan