Coffee Extraction

Coffee brewing advice is often hard to trust, due to much of the industry's lack of scientific knowledge, and the fact that most of the people giving the advice are trying to sell some fancy new brewing device. For exactly this reason, the most important coffee-related invention of the past few decades isn't a coffee brewer, grinder, or filter. It's a refractometer.

Refractometers, as the name suggests, measure the refractive index of a material (i.e., the amount that light bends when passing through said material). If that material is a liquid, a refractometer can be used to measure how much stuff is dissolved in the liquid—the total dissolved solids (TDS)—by measuring the difference between the refractive index of the solvent (e.g., water) and the solution (e.g., brewed coffee). Of course, refractometers have been around since the 19th century, but it wasn't until 2008, when VST Technologies released a refractometer specifically designed for coffee, that specialty coffee shops started using them to measure coffee strength and extraction yields. (I'm sure the fact that VST released an iPhone app to go with the refractometer helped to encourage adoption.)

So why is this important? Don't we already know how to make coffee stronger or weaker? Just use more or less coffee! Well, yes, but there is a crucial distinction between strength and extraction. One might take a very small amount of coffee and boil it for an hour, or take a large amount of coffee and steep it in cold water for a few seconds. The resulting solutions might be the same strength, but would have very different levels of extraction. The boiled coffee would contain nearly all the soluble solids from the coffee grounds, while the coffee steeped for a few seconds would contain only a small fraction of those soluble solids.

Both of these extremes are undesirable. While consumer preferences regarding coffee strength vary significantly, preferences regarding level of extraction are remarkably consistent. Coffee strength is expressed as the percentage of the brewed coffee that consists of dissolved coffee solids (% TDS). The Specialty Coffee Association of America recommends that brewed coffee consist of 1.15% - 1.35% TDS, while the Specialty Coffee Association of Europe recommends 1.2% - 1.5% TDS. These numbers are flexible; a ristretto shot of espresso can be upwards of 10% TDS and delicious.

Extraction yield is far less flexible. Extraction is expressed as the percentage of the ground coffee that is dissolved into the brewed coffee. The SCAA and SCAE recommend 18% - 22% extraction for both brewed coffee and espresso. This means that about one-fifth of the coffee in your brewer should end up in your cup, regardless of how strong you like your coffee.

Where did these numbers come from? The Coffee Brewing Institute (long since defunct) first studied consumer preferences for extraction levels in the 1950s, and their research has since been replicated by multiple organizations in several countries. The CBI originally came up with 17.5% - 21.2%, but more recent studies have settled on 18% - 22%. More information on these studies can be found in a post by Vince Fedele, head of VST. The point is, this number is based on empirical evidence regarding the preferences of the general public.

Overextraction results in intensely bitter coffee, while underextracted coffee is usually sour and lacking in complexity. The distinction between sourness and acidity is, as far as I can tell, unique to coffee. Acidity is generally a desirable characteristic in coffee; sourness refers to an acidic coffee that lacks the sweetness to balance its acidity. Since the caramelized sugars that give coffee its sweetness are less soluble than acids (the main acids in coffee are citric and malic), underextracted coffee is often sour in this sense.

The ultimate goal of any coffee brewer, therefore, is to achieve extraction yields of 18% - 22%. Extraction yield is affected by a number of variables, including: water temperature, grind size, brew time, agitation, coffee/water ratio, water profile, and coffee density and roast level. The problem is that the only way to measure extraction directly is to dry the spent grounds in an oven and weigh them, a process that takes hours and careful attention to detail. Because of the difficulty of measuring extraction, for decades the only people doing so were at large research institutes. However, if one could accurately measure the strength of the brewed coffee and the mass of the ground coffee and brew water, it would be possible to calculate extraction yield by the following formula: Extraction[%] = BrewedCoffee[g] * TDS[%] / CoffeeGrounds[g]. The mass of coffee grounds and brewed coffee can be easily measured with a scale, but measuring TDS is more difficult.

An affordable option is to buy a cheap TDS meter, such as those used to measure water quality. I bought one of these, but found that my readings varied wildly. The refinement provided by VST's coffee refractometer was a major improvement. Unfortunately, the device costs $400, which is more than I could justify to myself; but for a café that has already spent thousands upon thousands of dollars on espresso machines and grinders, spending $400 to ensure consistent quality is a small price.

The emergence of an objective measure of quality is powerful. 18%-22% extraction does not guarantee a great cup of coffee; the coffee itself still has to be good. But extraction yield is an objective evaluation of the brewing method, allowing for further empirical research into coffee brewing. Two significant results were presented at this year's Nordic Barista Cup. The first was a study by David Walsh reconfirming the validity of the 18%-22% extraction preference, while also suggesting a secondary preference for extractions at 12.5%-14.5%.

The second was an analysis by Vince Fedele of the liquid retained by the coffee grounds after brewing. The standard equation for calculating extraction yield from brew strength disregards any coffee extracted by the liquid retained by the spent coffee grounds. This assumption turns out to be roughly true of percolation brewing methods (drip, espresso), but not of infusion methods (e.g., French press, syphon/vacuum, Aeropress). When brewing by percolation, the liquid retained by the coffee has only been in contact with the coffee for a short time, and hence has extracted very little coffee. When brewing by infusion, on the other hand, the retained liquid has been in contact with the coffee for the entire brewing process. Therefore, the liquid retained by an infusion brewer has a significant quantity of coffee solids in solution; the retained liquid is coffee of essentially the same strength as the cup you just brewed. The standard extraction formula doesn't take this extra extraction into account. Practically, this means that to brew coffee to the same strength and extraction by infusion requires ~10% more ground coffee, to account for the brewed coffee that is retained by the grounds.

In addition to allowing cafés and consumers to improve their brewing practices, fast and easy measurement of extraction is a valuable tool for engineers developing coffee equipment. For example, soon after developing the VST refractometer, Vince Fedele used his invention to identify a major weakness in commercial espresso machines—the portafilter basket—and develop an improvement.

But more than anything else, I appreciate the development of the VST refractometer, and the renewed focus on extraction it brought with it, for the simplicity it provides. Though extraction is a complex concept, it is a single, measurable number which tracks quality of brewing methods.

There's one major caveat to these extraction numbers, which is that they're based on what our grinders actually produce. Even the best burr grinders on the market produce a significant quantity of grinds that are finer than desired. These fines extract more quickly than the rest of the coffee, which puts a limit on how high we can push our extraction yield without overextracting. If you have a low-quality grinder (e.g., a blade grinder), even an 18% extraction is going to taste over extracted, due to the wide variation in grind size. Experiments with sifting ground coffee, which removes the fines, suggest that a perfectly even grind would allow extraction up to 23% or 24%. (Matt Perger won this year's World Brewer's Cup with sifted coffee at 23% extraction.) But sifting is impractical in most settings, so until the next grinder revolution, we're stuck with a 22% ceiling.