Green coffee chemistry: what the raw seed is made of

What the seed is made of, and whose figures these are

Sourced

Shares of the dry weight of green coffee, each from the review named. The reviews gather other authors' analyses made by different methods over several decades, so the rows do not add up to a whole and should not be made to.

No figure is given here for caffeine, trigonelline, minerals or free amino acids, because no source giving their levels in green coffee was read in full for this page. Water is left out because it is not part of the dry weight; the figures for it, and whose they are, are in the guide to buying and judging green coffee.

Half the seed is cell wall

Sourced

The feature that most sets a coffee seed apart from other seeds is how much of it is wall. The 2006 carbohydrate review puts polysaccharides at about half the dry weight and names three main kinds: mannans, some of them carrying side branches of galactose and so called galactomannans; arabinogalactans bound to protein; and cellulose. This is why green coffee is so hard. The review explains that a pure mannan forms a hard, insoluble structure much like cellulose, because its chains bond to one another, and that each galactose branch interrupts that bonding and makes the polymer easier to dissolve.

The walls are the seed's food store as well as its skeleton. The review of seed physiology described in the guide to the coffee plant sets out how, during germination, enzymes that break down mannan soften the part of the endosperm in front of the root tip so that the embryo can push through. The same polymers that resist a grinder are what the seedling lives on.

For a drinker the walls matter because they resist water. The review describes cellulose and the mannans as the polymers most resistant to being dissolved, and calls the solubility of the galactomannans a critical factor in the yield of soluble coffee in manufacture. The two species differ here. The review reports that robusta contains an arabinogalactan that is more branched than arabica's, and offers that as the reason robusta's arabinogalactans dissolve more easily. What that difference does to the cup is not stated in the review, and this page does not guess.

Almost all the sugar is sucrose

Sourced

Early in the growth of the fruit the seed's free sugars are glucose and fructose. The carbohydrate review reports work in which they were the main sugars up to the halfway stage of maturation, at levels much higher in arabica than in robusta, and found mostly in the perisperm, the temporary tissue that fills the young fruit before the endosperm replaces it. By the end both had fallen to a few hundredths of one per cent of the dry weight, and sucrose, at 5 to 12 per cent, was essentially all of the free sugar in the mature seed.

Two things follow that bear on common claims. First, sucrose becomes the seed's sugar only late in development, which is consistent with the ripeness of the fruit at picking mattering to the composition of the seed. Second, the species differ: a 2025 metabolomic comparison found sucrose significantly lower in robusta than in arabica, and similar in arabica and stenophylla.

What does not follow is that a seed with more sucrose makes a sweeter cup. The review says that carbohydrates undergo complex changes in the roaster and that those changes contribute to the appeal of the beverage: sugar in the green seed is a starting material. No study read for this page measured sucrose in green coffee and sweetness in the cup from the same lots.

The oil, and what it carries

Sourced

Coffee is an oily seed. The 2006 lipid review gives the fat content of the two farmed species as between 7 and 17 per cent, averaging about 15 per cent in arabica and about 10 in robusta. Most of it is in the endosperm. A thin wax, about 0.2 to 0.3 per cent of the bean's weight, coats the surface.

The oil itself is ordinary: mainly triacylglycerols, with fatty acids in proportions like those of common edible vegetable oils. The review reports analyses that found no significant differences in the triacylglycerols between types, origins or drying methods. What is unusual is the rest. Up to a fifth of the total lipid is made of diterpenes, compounds of a kind found in few other foods, chiefly cafestol and kahweol, most of them bound to fatty acids.

One of them is a reliable marker of species. A diterpene called 16-O-methylcafestol was isolated from robusta in 1989, and the review records that its absence from arabica has been confirmed by several groups since. Because it survives roasting unchanged, it is used to detect robusta in a coffee sold as pure arabica. This is a rare case in coffee of a chemical test that answers a question a label raises.

The oil also explains a difference between brewing methods that has nothing to do with taste. According to the review, less than 0.2 per cent of the coffee's lipids pass into filtered coffee made in a household machine, against 1 to 2 per cent into an espresso, and the diterpenes travel with them.

Protein and free amino acids

Sourced

Less was read on this fraction than on any other, and the section is short for that reason. A 2022 study of coffee proteins gives the protein content of green arabica as 8.5 to 12.7 per cent, citing earlier work, and describes one protein, a storage protein of the class called 11S, as the major one. Seeds of many plants store nitrogen in proteins of this class.

The same study looked at what processing does to that protein. In fourteen arabica samples from different countries and processes, it found significantly more chemically modified fragments of the storage protein in wet-processed beans than in dry-processed ones, the modification being the attachment of phenolic compounds. That is a detected difference in the green bean. The authors present it as a contribution to understanding cup quality; they did not roast and taste the samples.

Free amino acids, the small molecules that react with sugars in roasting, are a separate and much smaller pool. The trial of wet processing in Yunnan described in the fermentation guide found their concentrations in the seed changing from stage to stage, which its authors attribute to the seed's own metabolism. No figure for them is given here.

Chlorogenic acids: a family, not a compound

Sourced

Chlorogenic acids are esters of quinic acid with one or two phenolic acids, mostly caffeic and ferulic. The 2006 review of coffee phenolics lists the main groups, each with at least three forms differing in where on the quinic acid the bond is made, and reports that one of them, 5-caffeoylquinic acid, accounts on its own for about 56 to 62 per cent of the total. Chlorogenic acid in the singular usually means that one.

They are among the most abundant small molecules in the seed, and one of the clearest chemical differences between the species: 4 to 8.4 per cent of dry matter in arabica against 7 to 14.4 in robusta, on the review's summary. They also fall as the fruit ripens. The review cites one study that found 8.7 per cent in arabica seeds from dark green fruit and as little as 1.3 per cent in seeds from over-ripe fruit.

Roasting takes them apart. The review describes them as isomerised, hydrolysed or degraded during roasting, almost completely under intense conditions, with part converted to lactones. It attributes to them a share of the acidity, astringency and bitterness of the drink, each on the authority of earlier authors, and says that bitterness rises in roasting through the release of caffeic acid and the formation of lactones and other derivatives.

On quality the review opens candidly: the relationship between chlorogenic acids and cup quality is still unclear and somewhat controversial. The studies it then cites lean one way. Several found higher levels in green coffee going with lower cup quality, one found a disagreeable flavour from adding one group of these acids that disappeared when another was added, and the review offers, as a hypothesis, that there is a level above which quality falls, connecting it with the products of their oxidation in the seed. That is an association and a hypothesis. A page that says a coffee is harsh because it is high in chlorogenic acids states as settled what the review puts forward as unsettled.

Caffeine and trigonelline

Sourced

Both are nitrogen-containing compounds made by the plant, and both are treated at more length in the coffee chemistry guide and in the methodology behind the caffeine estimator. What was read for this page is a comparison and not a set of levels. The 2025 metabolomic study found caffeine significantly higher in green robusta than in arabica or stenophylla, and similar in those two.

Trigonelline breaks down on heating; a review registered for the chemistry guide summarises earlier studies of its decomposition, none of which was opened. What roasting does to caffeine, and how much reaches a cup, is set out on the pages that carry the evidence for it, and no figure is repeated here.

What calling something a precursor claims

CoffeeHQ explanation

Green coffee is often described by its flavour precursors. The word is a claim about roasting, made of a compound that is measured before roasting. CoffeeHQ's reading of how far each claim goes in the sources read:

Sucrose

Established as a starting material: it is abundant in the green seed, and the carbohydrate review describes complex changes to carbohydrates in roasting. Not established here: that more of it in a given lot gives a sweeter or better cup.

Free amino acids and protein

Named in the sources as the partners of sugars in the browning reactions of roasting. The one study read on coffee protein measured modification in the green bean and did not follow it into the cup.

Chlorogenic acids

Established as broken down in roasting into smaller acids, lactones and phenols, and credited by the review with part of the acidity, astringency and bitterness. The review itself calls the link to quality unclear.

Lipids

Not a precursor in the same sense. The lipid review reports that in roasted coffee most of the diterpenes are still bound to fatty acids, that the esters of cafestol and kahweol fall as the roasting temperature rises, and that those of 16-O-methylcafestol are stable through roasting. What the oil contributes to the cup is not quantified in the sources read.

Polysaccharides

Partly broken down in roasting, and the review says those changes contribute to the appeal of the beverage. How much any one polymer contributes to body or to flavour is not settled in what was read.

What changes while green coffee waits

Sourced

A green seed is not chemically at rest. The lipid review reports that an enzyme that splits fats, a lipase, was active in every green coffee its authors examined, including Brazilian coffee ten years old, and it describes an experiment of their own that shows what follows. A Colombian coffee was stored for eighteen months at three temperatures and three moisture contents, and the free fatty acids released from its oil were measured.

Moisture and temperature both mattered, and the atmosphere in the store did not. At 25 degrees Celsius the coffee at its original moisture of 11.8 per cent roughly doubled its free fatty acids over the eighteen months; the same coffee dried to 6.2 per cent hardly changed; coffee moistened to 13.5 per cent rose most. At 12 degrees, which the authors describe as typical of warehouses in Hamburg, even the moistened coffee rose only a little. At 40 degrees pronounced changes appeared within three months in all but the dried coffee.

That is a measurement of a chemical change, not of taste. Free fatty acids are a common index of deterioration in oils. Whether they are what a taster means by past crop is not shown in the review, which reports no tasting scores. Its one remark on the cup is that the storage at 40 degrees was stopped after a year because the brews made from that coffee were, in its authors' words, simply not worth discussing. The guide to storing and shipping green coffee sets out what three packaging studies with tasters found and where they disagree.

What this page gives no account of

CoffeeHQ explanation

Minerals, organic acids other than the chlorogenic family, and the hundreds of minor compounds are not covered, because nothing was read on them for this page. No level is given for caffeine or trigonelline in green coffee. The three reviews that carry most of this page date from 2006 and gather analyses older still; methods have changed, and a newer survey might move several of the ranges. Nothing here is a statement about health, and nothing here predicts how a particular coffee will taste.

What to do next

• Coffee chemistry — What roasting and brewing make of these compounds, and which of them reach the cup.
• Buying and judging green coffee — What a buyer measures on the raw seed, and which of those measurements predict anything.
• Storing and shipping green coffee — Packaging, warehouses and containers, and what tasters found after months in each.
• The coffee plant — How the seed is built during the months the fruit is on the tree, and how it germinates.
• Coffee fermentation — What processing changes in the seed before it is dried.