Coffee aroma chemistry: why compounds are not flavour notes

A thousand compounds, a few dozen that matter

Sourced

A review of espresso aroma gives the count of volatile compounds identified in coffee as more than a thousand, and then reports the view of the field that only twenty to thirty of them are important to the aroma of any given coffee. The rest are present and can be measured, and contribute little or nothing that a person can smell.

A study that roasted one coffee to four degrees shows the gap in a single experiment. Assessors smelled each compound as it left the instrument, and 25 had a detectable smell. Several of the largest peaks in the measurement, acetic acid and pyridine among them, had none. The review makes the general point: the standard instrument for separating and identifying volatile compounds gives no description of how any of them smells, and cannot tell the compounds responsible for an aroma from the bystanders.

So the first thing to know about a chart of a coffee's volatile compounds is that the height of a peak is not the size of a contribution. Abundance and importance are different properties.

The classes of compound, and what is said of each

Sourced

The classes below are the ones the review lists. Where a smell is given, it is how a compound of that class has been described when smelled alone by trained assessors, as the sources report it. It is not what the class does to a cup.

Furans

With the pyrazines, the most abundant class by quantity, on the review's account, and the major class detected in espresso. They form from the breakdown of sugars and from the Maillard reactions, and the review adds the oxidation of lipids as a source. One sulphur-bearing member, 2-furfurylthiol, is described in a study of coffee with milk as one of the most important potent odorants in brewed coffee.

Pyrazines

Nitrogen-containing rings formed where sugars and amino acids react; a study of aroma during roasting lists alkylpyrazines among the products of the Strecker degradation. The review, citing an earlier study, reports pyrazines described as earthy, musty, roasty or burnt, and woody. It also ranks them, with the sulphur compounds, as the most significant class for coffee's character.

Aldehydes and ketones

Small molecules, many from the Strecker degradation of amino acids and some from the oxidation of lipids. The review reports certain aldehydes described as fruity and the two-ketone compounds called diones as buttery. In a study of single beans, ketones were the most uniform class from bean to bean and one aldehyde, hexanal, among the most variable.

Sulphur compounds

Present in very small amounts and, by the review's account, among the most significant for character. Small amount and large effect is the pattern that makes peak height misleading. In the roast-degree study the very dark roast was marked by compounds assessors described as burnt and sulphurous.

Phenolic compounds

Formed as the chlorogenic acids break down; a review of coffee phenolics refers to phenol derivatives made in roasting. One laboratory that produced roast faults on purpose found a phenol raised in its dark roast and a related compound in its scorched one.

Pyrroles and pyridines

Nitrogen compounds; a review of trigonelline reports volatile pyrroles and pyridines among the products of its breakdown in roasting. Pyridine itself was one of the large peaks with no detectable smell in the roast-degree study, and one pyridine was among the most variable compounds between single beans.

Esters, alcohols, acids and others

Listed by the review among the classes present. CoffeeHQ has read nothing that establishes what esters contribute to roasted coffee, though they are the class most often named when a coffee is described as fruity. One study of aroma during roasting notes that low-molecular-weight alcohols do not, in general, tend to be potent odorants in coffee.

How the compounds that smell are found

Sourced

The method that separates odorants from bystanders uses a person. The stream of separated compounds leaving the instrument is split, part going to the detector and part to a port where a trained assessor sniffs and reports when something smells, and of what. The review describes several ways of scoring the result, including diluting the sample step by step to see which smells persist longest.

The strength of the method is that it tests each compound on a human nose. Its limit is the same fact. Each compound arrives alone, in air, at whatever concentration the instrument delivers. The description it earns there, fruity or buttery or earthy, is the description of that compound smelled alone. A cup of coffee never presents one in that way.

From molecule to word: a model of the steps between

CoffeeHQ explanation

A model of what has to happen for a compound in the cup to become a word in a tasting note. Each step is a place where chemistry and vocabulary can part company, and the sections that follow give the evidence for four of them.

Mixtures: where the parts stop being audible

Sourced

A review of research on how odour mixtures are perceived describes three things that can happen when odorants meet. One can cover another, completely or in part, which is masking. A mixture can smell stronger than its parts would suggest, which is synergy. Or the components can merge into a single smell that is perceived as one object and is not the sum of its parts.

It also reports a limit on analysis by nose. In the experiments it reviews, people could rarely identify more than three components of a mixture, four at most, and training or expertise did not raise the limit. A trained taster is better at naming what they perceive and no better at hearing the individual instruments in a large orchestra.

For coffee, with its twenty or thirty significant odorants, this means the aroma is perceived as blended impressions and not as a list. A note such as blackcurrant or toasted hazelnut is a name for an impression produced by many compounds together. It is not a claim that the coffee contains a blackcurrant compound, and the review's account of mixtures is a reason to think that removing or adding one compound could change the impression in ways its own smell would not predict. That last step is CoffeeHQ's inference: the review is about smell in general and not about coffee.

The liquid matters: milk as a worked example

Sourced

Two Japanese studies, both by authors working for food companies, show the liquid changing what reaches the nose. In one, a device that imitates swallowing and breathing out was fed a coffee-flavoured drink with increasing amounts of milk fat and of the non-fat solids of milk. As the fat rose, the release of the compounds that dissolve most readily in fat fell significantly; the non-fat solids had a smaller effect. The fat was holding them back. Most of the drinks were mixtures of ten aroma compounds, made up at room temperature and buffered to neutral, and stirred in a vessel held at body temperature; a smaller set was made with real espresso. In one of those, with 2 per cent milk fat and 4 per cent non-fat solids, a sulphur compound and a second roast-smelling compound were released at about a quarter and a sixth of the rate measured without milk.

In the other, three trained panellists drank black coffee and coffee with milk while the air they breathed out through the nose was collected. Most odorants, including the potent ones, were breathed out in similar amounts either way. Three were significantly lower with milk for every panellist, one of them a sulphur compound related to 2-furfurylthiol, and the same compound the first study found held back. The authors then added that compound and its relative to the milk coffee until the amount breathed out matched black coffee, and 31 trained colleagues rated the adjusted drink as smelling more of coffee than the unadjusted one. The coffee was drunk at room temperature.

So the same coffee, with the same compounds in it, delivers a different mixture to the nose when milk is added, and the change is selective. It is three panellists and a company's own panel in one study and an instrument in the other, both on drinks at room temperature. As a demonstration that composition in the cup is not the same as composition at the nose, it is clear enough.

The coffee is not uniform either

Sourced

A study that roasted beans one at a time, ten from each of 25 lots, under identical conditions found the amounts of volatile compounds varying widely between beans of the same lot. For some compounds the spread between beans was larger than the average. Phenols and nitrogen-containing ring compounds varied most within a lot and ketones least, and the most variable compounds included ones the authors call potent odorants.

A cup of coffee is an average over some hundreds of beans, which smooths this out. But it is a reminder that the composition of a coffee is a distribution, and that a published figure for a compound in a coffee describes one sample of it.

Three kinds of sentence to distrust

CoffeeHQ explanation

A compound named as the cause of a note

Of the form: this coffee tastes of a fruit because it contains a particular ester. No source read for this page demonstrates a link of that kind for any coffee, and the evidence on mixtures is a reason to doubt that one could be read off an analysis.

A class named as the cause of a character

Of the form: pyrazines make it nutty, furans make it sweet. The descriptions attached to classes come from compounds smelled alone. The classes are large, their members smell different from one another, and in a cup they are perceived together.

More of a compound read as more of a flavour

Abundance is not contribution. Some of the most abundant volatile compounds in roasted coffee had no detectable smell in the one study read that tested them by nose, and the compounds ranked most significant are present in the smallest amounts.

What is left unsaid about aroma

CoffeeHQ explanation

It gives no table matching compounds to flavours, and no threshold or concentration for any compound. It does not say which odorants are the key ones for coffee in general: the studies read disagree on the list, and each tested particular coffees. It says nothing about taste in the narrow sense, sourness, bitterness and sweetness, which are covered on the pages about tasting and chemistry.

The account of the classes leans on one review, whose statements about what a class smells of are its reports of other studies that were not opened. The account of mixtures comes from a general review of smell. Both studies of milk were made by companies with an interest in flavour. Nothing here was smelled or measured by CoffeeHQ.

What to do next

• Coffee flavour and tasting — The same question from the taster's side: the routes of smell, the tasting words and how far each is pinned down.
• Coffee roasting chemistry — Where the volatile compounds come from, and what happens to them at the dark end of a roast.
• Why coffee goes stale — How the mixture drifts after roasting as some compounds escape and others react.
• Milk science — What milk's proteins and fat do, including to the coffee they are mixed with.
• Tasting Lab — A worksheet for recording what you perceive in your own words before reaching for anyone else's.