Coffee oil
The fat of the coffee bean – what it is made of, where it sits in the cell, why it reaches the surface of dark and ageing beans, how much of it ends up in each brew, and how its oxidation makes coffee stale.
Across the chain: Science
Contents
In brief
- Green arabica coffee averages about 15% lipids and robusta about 10%; most of this coffee oil is triacylglycerols, stored in oil bodies in the endosperm.1,2
- Coffee oil carries diterpenes – cafestol and kahweol in arabica; robusta also contains 16-O-methylcafestol, which survives roasting and is used to detect robusta in blends.1
- Gas pressure pushes freed oil through the cell walls to the bean surface; this runs faster in dark roasts and continues during storage.3,4
- Paper filters retain most of the oil; in one 2025 study, boiled coffee held 939 mg/L cafestol and paper-filtered home brews about 12 mg/L.8,9
- Oxidation of unsaturated fatty acids such as linoleic acid forms aldehydes, including hexanal, which are linked to the rancid flavour of stale coffee.5,6
What coffee oil is#
Coffee oil is the fat of the coffee bean. Arabica and robusta green coffee contain between 7 and 17% fat.1 Green arabica averages about 15% lipids, robusta about 10%.1
Most of the oil is triacylglycerols, also called triglycerides: three fatty acids bound to glycerol.1,5 Linoleic and palmitic acid are the main fatty acids.1,5 Linoleic acid is polyunsaturated, which makes it prone to oxidation.6
The oil also carries smaller groups of compounds. These are diterpenes (cafestol and kahweol), sterols and tocopherols (vitamin E compounds).1 A thin wax covers the surface of the green bean, about 0.2–0.3% of its weight.1
The oil matters at three points in the chain. It shapes how roasted beans look and age. It decides part of what reaches the cup. And its oxidation is one cause of stale, rancid flavour.3,6,11
Where the oil sits in the bean#
Almost all of the oil sits in the endosperm, the tissue that makes up the green bean. Only the wax lies on the outside.1 The plant stores it as an energy reserve for germination.2
Inside the cells, the oil is packed into oil bodies, also called oleosomes. These are tiny spheres of triglycerides, each wrapped in a single protein membrane.2 In a 2013 microscopy study, they measured typically 0.2–2.5 µm across.2 According to a roasting textbook chapter, they lie in the cytoplasm along the cell walls.3
Oil on the surface of roasted beans#
Roasting destroys the cell organisation of the bean, breaks up the oil bodies and frees the oil.3 Gas pressure inside the bean then pushes the oil through tiny channels in the cell walls to the surface.3 Small droplets appear first. They can join into a shiny film over the whole bean.3
Dark roasts turn oily sooner. Oil migration runs faster in them, because the driving force is stronger.3,4 In extremely dark roasts, oil may already appear in the last stages of roasting.3 During storage, oil migration accompanies the release of gas from the bean.4
Darker roasts also release more gas, and release it faster.14 The second crack entry shows what breaks in the bean late in a dark roast. Porosity and cell structure explains the channels the oil travels through.
Coffee oil in the cup#
Hot water extracts the diterpenes, but a paper filter retains them.8 According to a brewing textbook chapter, paper absorbs oil very efficiently, so paper-filtered brews are low in fat.11 A metal mesh, as in a French press, holds back less.9,11
In espresso, pressure forces oil droplets out of the coffee bed into the cup, according to the same chapter.11 There they form an emulsion: tiny droplets spread through the liquid.12,13 Hotter water also extracts more lipids, which the chapter links to body.11
How much oil adds to body is not settled. The body and mouthfeel entry covers that question, and the espresso entry covers crema.
Oil, staling and rancidity#
Coffee oil affects how roasted coffee ages.6 Oxidation of unsaturated fatty acids forms aldehydes such as hexanal, with fatty, soapy or green notes.5 Hexanal is used as a marker of lipid oxidation in stored coffee.5,6
The oil on the surface is most exposed. According to the Specialty Coffee Association (SCA) freshness handbook, an oily surface exposed to oxygen forms compounds that contribute to stale and rancid flavours.6 Even coffee stored under vacuum or low oxygen can show some lipid oxidation, driven by free radicals formed in roasting.6
The freshness and storage entry describes how packaging and temperature slow this down.
Reference values#
Values reported in the sources. They describe particular coffees and methods; they are not targets.
| Parameter | Value reported | Basis and conditions | Source |
|---|---|---|---|
| Total lipids, green coffee | 7–17% fat; arabica averages ≈15%, robusta ≈10% | Review of earlier analyses | Speer & Kölling-Speer1 |
| Triacylglycerols | 75.2% of green coffee lipids | One 1981 analysis, as tabulated in the review | Speer & Kölling-Speer1 |
| Diterpene esters (fatty acid esters of diterpene alcohols) | 18.5% of lipids; free diterpene alcohols 0.4% | Same analysis | Speer & Kölling-Speer1 |
| Sterols | Sterol esters 3.2%, free sterols 2.2% of lipids; β-sitosterol ≈50% of the sterols | Same review | Speer & Kölling-Speer1 |
| Tocopherols | 0.04–0.06% of lipids; α-, β-, γ-tocopherol ≈2:4:0.1; total ≈5.5–6.9 mg/100 g | Same review | Speer & Kölling-Speer1 |
| Surface wax | 0.2–0.3% of bean weight | Green coffee | Speer & Kölling-Speer1 |
| Carboxylic acid 5-hydroxytryptamides (C-5HT) | Arabica 500–2,370 mg/kg; robusta 565–1,120 mg/kg; normal roasts 500–1,000 mg/kg | Green and roasted coffee | Speer & Kölling-Speer1 |
| Cafestol and kahweol, beans | ≈1% of bean weight | Review | Urgert & Katan8 |
| Cafestol and kahweol, one green arabica | 0.8% w/w | One lot of 100% arabica, bought in Rio de Janeiro | Novaes et al.7 |
| Cafestol esters, green coffee | Arabica 9.4–21.2 g/kg; robusta 2.2–7.6 g/kg | Dry weight | Speer & Kölling-Speer1 |
| Oil bodies | 0.2–2.5 µm | Endosperm, five Coffea species, light and electron microscopy | Crisafulli et al.2 |
| Lipids reaching the brew | Household filter coffee < 0.2%; espresso 1–2%; Scandinavian boiled up to 22% | Espresso 1–2% of the coffee’s lipids, Scandinavian style up to 22% of the coffee fat; no basis stated for filter coffee | Speer & Kölling-Speer1 |
| Cafestol, boiled / fabric-filtered | 939.2 / 28.0 mg/L | Mean of duplicates | Orrje et al.9 |
| Cafestol, French press / percolator | 86.8 / 91.2 mg/L | Mean of duplicates | Orrje et al.9 |
| Cafestol, paper-filtered home brew | Median 11.5 mg/L (4.2–23.8) | Five coffees, two kinds of paper filter | Orrje et al.9 |
| Cafestol, espresso | Median 1,059.3 mg/L (35.6–2,446.7), n = 4 | Three cafeterias and one laboratory machine, Gothenburg | Orrje et al.9 |
| Cafestol, workplace brewing machines | Median 175.7 mg/L (24.4–444.0), n = 11 | Machines brewing from ground coffee | Orrje et al.9 |
Lipid classes and fatty acids#
Lipid classes. In the analysis tabulated by a 2006 review, triacylglycerols made up three quarters of green coffee lipids.1 Esters of diterpene alcohols with fatty acids came second, followed by sterol esters and free sterols.1 The review notes that diterpenes of the kaurene family can reach up to 20% of the total lipids.1 Tocopherols, phosphatides and tryptamine derivatives are minor classes.1
Data table
| % of lipids | Source | |
|---|---|---|
| Triacylglycerols | 75.2% | 1 |
| Diterpene esters | 18.5% | 1 |
| Sterol esters | 3.2% | 1 |
| Free sterols | 2.2% | 1 |
| Free diterpene alcohols | 0.4% | 1 |
Fatty acids. The review’s table gives the fatty acids of the triacylglycerols for four arabica and nine robusta samples.1 Linoleic acid (C18:2) leads in both, followed by palmitic acid (C16:0).1 Arabica had more linoleic acid; robusta more oleic acid (C18:1).1 In robusta, stearic acid (C18:0) is clearly lower than oleic acid; in arabica the two are almost equal.1 The review suggests the stearic to oleic ratio as a first indication of robusta in blends.1
Data table
| % of fatty acids | Source | |
|---|---|---|
| Linoleic C18:2 · arabica | 52.2–54.3% | 1 |
| Linoleic C18:2 · robusta | 43.9–49.3% | 1 |
| Palmitic C16:0 · arabica | 26.6–27.8% | 1 |
| Palmitic C16:0 · robusta | 27.2–32.1% | 1 |
| Oleic C18:1 · arabica | 6.7–8.2% | 1 |
| Oleic C18:1 · robusta | 9.7–14.2% | 1 |
| Stearic C18:0 · arabica | 5.6–6.3% | 1 |
| Stearic C18:0 · robusta | 5.8–7.2% | 1 |
Sterols and tocopherols. β-Sitosterol makes up about half of the sterols, followed by stigmasterol and campesterol.1 Two sterols, 24-methylenecholesterol and Δ5-avenasterol, occur in much higher amounts in robusta than in arabica.1 In one data set cited by the review, α-, β- and γ-tocopherol occurred in a ratio of about 2:4:0.1.1
The wax layer. A thin waxy layer covers the green bean.1 Its main constituents, in the fraction insoluble in petroleum ether, are carboxylic acid 5-hydroxytryptamides (C-5HT).1 Polishing, dewaxing and steaming remove the wax.1
Diterpenes and the robusta marker#
Cafestol and kahweol. Arabica contains cafestol and kahweol. Robusta contains cafestol, small amounts of kahweol and, in addition, 16-O-methylcafestol (16-OMC).1 Most of the diterpenes are esterified with fatty acids; the free forms are usually less than 3.5% of the total.1 Cafestol esters ranged from 9.4–21.2 g/kg in arabica and 2.2–7.6 g/kg in robusta, on a dry-weight basis.1
16-OMC as an authenticity marker. 16-OMC was not detectable in arabica beans and stays stable during roasting.1 According to the review, this makes it the ideal marker for detecting robusta in arabica blends. The German standard method DIN 10779 detects robusta shares below 2%.1 DIN 10779 has since been withdrawn and replaced by DIN EN 18003 (see Arabica and robusta compared). A 2023 study notes that 16-OMC had long been described as unique to green robusta. It adds that a later study identified 16-OMC in 100% arabica beans, using more concentrated extracts.7
What roasting does to the lipids#
Bulk lipids. According to a roasting chemistry chapter, most lipids survive roasting.5 The fatty acid composition changes little, while trans fatty acids increase.1 Part of the lipids breaks down to aldehydes, which can react further with other coffee constituents.5 Roasting retained 83–99% of total tocopherols in the data cited by the review.1
Diterpenes. Cafestol and kahweol esters decrease depending on the roasting temperature; 16-OMC esters are stable.1 Dehydrocafestol and dehydrokahweol form, and their amounts rise with roasting temperature.1 According to the review, the ratio of cafestol to dehydrocafestol indicates roast degree: 25–40 for a well-roasted coffee, up to 15 for a strongly roasted one.1
One roasting series. A 2023 Brazilian study roasted one arabica lot to light, medium and dark.7 Degradation products of cafestol and kahweol appeared from the light roast on; the study identified 16, ten from kahweol and six from cafestol.7 Total diterpenes dropped sharply only in the dark roast, and kahweol proved less heat-stable than cafestol.7 How this fits the review is shown under Contested.
Oil migration. The mechanism is set out above and in the second crack entry: freed oil is driven to the surface by gas pressure, faster in dark roasts.3,4 In cryo-scanning electron micrographs reproduced in the roasting textbook chapter, a bean surface was smooth just after roasting and covered with small droplets after one day.3 The SCA handbook notes that carbon dioxide can dissolve in coffee oil at high concentrations, one of the ways gas is held in the bean.6 According to a roasting chemistry chapter, fast-roasted coffee is presumed more affected by lipid oxidation, because more oil migrates to the surface.5
Diterpenes in the brew#
Filter matters most. A 1997 review states that hot water extracts diterpenes and a paper filter retains them.8 Boiled coffee, such as Scandinavian boiled and Turkish coffee, carries them into the cup.8 The 2006 review reports less than 0.2% lipids for household filter coffee, 1–2% of the lipids passing into espresso and up to 22% of the coffee fat in Scandinavian-style coffee.1
Measured in 2025. A Swedish study compared cafestol across brews.9 Boiled coffee held 939.2 mg/L; poured through a fabric filter, 28.0 mg/L.9 French press and percolator coffee had intermediate levels, and paper-filtered home brews a median of 11.5 mg/L.9 The authors name the lack of a fine filter as one possible reason for elevated levels in workplace brewing machines.9
Data table
| mg/L | Source | |
|---|---|---|
| Espresso (median, n = 4) | 1,059.3 mg/L | 9 |
| Boiled, unfiltered | 939.2 mg/L | 9 |
| Workplace brewing machines (median, n = 11) | 175.7 mg/L | 9 |
| Percolator | 91.2 mg/L | 9 |
| French press | 86.8 mg/L | 9 |
| Boiled, through fabric filter | 28.0 mg/L | 9 |
| Paper filter, home (median) | 11.5 mg/L | 9 |
Espresso. The four espressos in the 2025 study varied from 35.6 to 2,446.7 mg/L, a variation the authors could not explain.9 A 2020 Brazilian study found similar cafestol and kahweol per 50 mL in espresso and paper-filtered coffee.10 Small cups mean that concentrations per litre do not translate directly into amounts per cup (see Contested).9,10
Roast degree and the brew. In the 2023 roasting series, boiled coffee extracted the most diterpenes, followed by French press, moka pot and Turkish coffee; filtered coffee extracted the least.7 In boiled coffee from the dark roast, the diterpene degradation products added up to more than 1 g/L.7
Lipids, crema and body#
Emulsion. A crema chapter describes espresso as an emulsion of oil droplets, a suspension of solid particles and a foam, all in one cup.12 A 2011 review reports that 90% of the oil droplets in espresso are smaller than 10 µm.13 The espresso entry gives the details.
Foam stability. According to the crema chapter, lipids can destabilise espresso foam.12 Pure arabica espresso contains more lipids than robusta espresso on average, so lipid-induced foam collapse is more likely.12 The chapter adds that oil spreading at the surface of the cup may also destabilise the foam.12
Body. A brewing chapter calls the role of extracted oil in body and mouthfeel “not fully clear”.11 Espressos are higher in oil than unpressurised brews, but also stronger; French press coffee is high in oil but not generally perceived as fuller.11 The body and mouthfeel entry discusses the evidence.
Oxidation chemistry#
Pathway. Linoleic acid and other polyunsaturated fatty acids are abundant in coffee and oxidise to hexanal and other aldehydes.6 These aldehydes, nonenal and other enals and dienals among them, are potent odorants but not key compounds of fresh coffee aroma.5 Secondary oxidation products contribute to the rancid flavour of stale coffee.6
Drivers. According to the SCA handbook, oxygen, moisture and temperature accelerate lipid oxidation, and its extent correlates with the surface area of ground coffee.6 Free radicals formed in roasting can drive oxidation even under vacuum or low oxygen.6
Cafestol and cholesterol#
This section describes what reviews and studies report. It gives no health advice.
A 1997 review states that some brewing methods raise the serum concentration of total and low-density-lipoprotein (LDL) cholesterol in humans, and that cafestol and kahweol are the responsible factors.8 The 2006 lipid review also attributes the effect of some coffee preparations on serum cholesterol to these diterpenes.1 Because paper retains diterpenes, the 1997 review links the effect to Scandinavian boiled, cafetière and Turkish coffee; filtered coffee does not affect cholesterol.8
Risks and defects#
- Oily beans in storage. Oil on the surface exposed to oxygen forms stale and rancid off-flavours.6 Dark roasts reach this state sooner.3,4
- Rancidity in ground coffee. Lipid oxidation correlates with the surface area of ground coffee.6
- Fast roasting. A textbook chapter presumes fast-roasted coffee more affected by lipid oxidation, through stronger oil migration.5
- Diterpenes in unfiltered brews. Boiled and other unfiltered brews carry far more cafestol than paper-filtered ones.8,9
Open questions#
- Diterpene loss in roasting. How much cafestol and kahweol are lost at each roast degree differs between the review and a single-lot study (see Contested).1,7
- Oil and body. Whether extracted oil itself adds body, separate from strength, is not established.11
- Espresso diterpenes. Values vary widely and two studies disagree on how espresso compares with paper-filtered coffee.9,10
- Rate of oil migration. No study opened for this entry measured how fast oil reaches the surface across roast degrees and storage temperatures; the sources describe it qualitatively.3,4
Key studies#
Speer & Kölling-Speer, 2006 – The lipid fraction of the coffee bean. Reviews the lipid classes of green and roasted coffee: triacylglycerols, diterpenes and their esters, sterols, tocopherols and the C-5HT of the wax layer. It sets out how roasting changes them and why 16-O-methylcafestol serves as the marker for robusta in blends.
Crisafulli et al., 2013 – Oil bodies in Coffea. Examined the endosperm of five Coffea species by light and transmission electron microscopy. Coffee oil is stored in oil bodies of typically 0.2–2.5 µm, each bounded by a single protein membrane.
Urgert & Katan, 1997 – The cholesterol-raising factor. Reviews the evidence that cafestol and kahweol, about 1% of the bean, raise serum total and LDL cholesterol. Hot water extracts them, paper filters retain them, which is why boiled but not filtered coffee has the effect.
Novaes et al., 2023 – Diterpenes through roasting and brewing. Roasted one arabica lot to three degrees and brewed it in several ways. Total diterpenes fell sharply only in the dark roast, kahweol degraded more readily than cafestol, and 16 degradation products were identified.
Orrje et al., 2025 – Cafestol in workplace machines. Measured cafestol and kahweol in coffee from 14 workplace machines and compared it with home brews. Boiled coffee and espresso were highest, paper-filtered coffee and most liquid-model machines lowest; many brewing machines gave intermediate values.
- Diterpenes in espresso: four espressos sampled in Gothenburg had a median of 1,059.3 mg/L cafestol (range 35.6–2,446.7 mg/L), against 11.5 mg/L for paper-filtered home brews9; a 2020 Brazilian study found similar cafestol and kahweol per 50 mL serving in espresso and paper-filtered coffee.10 Coffees, machines and sampling differ.
- Diterpenes and roast degree: a 2006 review reports that cafestol and kahweol esters decrease depending on roasting temperature1; in a 2023 study of one arabica lot, measured total diterpenes did not fall at light and medium roast and dropped only at dark roast.7
Coffearium (2026). Coffee oil. Coffearium – The open coffee compendium. Retrieved 22 October 2026, from https://coffearium.org/entries/coffee-oil/
@misc{coffearium2026-coffee-oil,
author = {{Coffearium}},
title = {{Coffee oil}},
year = {2026},
howpublished = {Coffearium -- The open coffee compendium},
url = {https://coffearium.org/entries/coffee-oil/},
urldate = {2026-10-22},
note = {Retrieved 22 October 2026}
}Entries are updated when the evidence changes; the retrieval date tells your readers which version you used. Changes are listed in the Change log.
Change log
- – Removed the roasted-arabica lipid value (about 15% of dry weight) from the reference table; reworded the tocopherol sentence to the reported ratio; added that a later study identified 16-O-methylcafestol in arabica and softened the quick fact "in robusta only"; replaced the claim that oil reaches the surface "in lighter roasts" during storage with the thesis statement that oil migration accompanies gas release in storage; removed a health statement taken from the background of a primary study; corrected the basis of the lipids-in-brew row. Reason: Verification against sources (number rule, Tier B only; Speer & Kölling-Speer 2006 states α-tocopherol predominance but gives a 2:4:0.1 α:β:γ ratio; Novaes et al. 2023 discussion; Schenker 2000 §2.2.4; health claims only from Tier A reviews; Speer & Kölling-Speer 2006 does not state a basis for the filter-coffee value).
Sources
- ASpeer, K.; Kölling-Speer, I. (2006). The lipid fraction of the coffee bean. Brazilian Journal of Plant Physiology 18(1), 201–216. doi:10.1590/S1677-04202006000100014 · other copies
- ACrisafulli, P.; Navarini, L.; Silizio, F.; Pallavicini, A.; Illy, A. (2013). Ultrastructural Characterization of Oil Bodies in Different Coffea Species. Tropical Plant Biology 7, 1–12. doi:10.1007/s12042-013-9132-2 · other copies
- BSchenker, S.; Rothgeb, T. (2017). The Roast – Creating the Beans' Signature. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 11, pp. 245–271. doi:10.1016/B978-0-12-803520-7.00011-6 · other copies
- BSchenker, S. (2000). Investigations on the Hot Air Roasting of Coffee Beans (Diss. ETH No. 13620). ETH Zürich (doctoral thesis).
- BPoisson, L.; Blank, I.; Dunkel, A.; Hofmann, T. (2017). The Chemistry of Roasting – Decoding Flavor Formation. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 12, pp. 273–309. doi:10.1016/B978-0-12-803520-7.00012-8 · other copies
- BSmrke, S.; Sage, E.; Wellinger, M.; Yeretzian, C. (2018). The Coffee Freshness Handbook (ISBN 978-0-9995807-2-1). Specialty Coffee Association, Santa Ana.
- ANovaes, F. J. M.; da Silva, M. A. E.; Silva, D. C.; de Aquino Neto, F. R.; Rezende, C. M. (2023). Extraction of Diterpene-Phytochemicals in Raw and Roasted Coffee Beans and Beverage Preparations and Their Relationship. Plants 12(8), 1580. doi:10.3390/plants12081580 · other copies
- AUrgert, R.; Katan, M. B. (1997). The cholesterol-raising factor from coffee beans. Annual Review of Nutrition 17, 305–324. doi:10.1146/annurev.nutr.17.1.305 · other copies
- AOrrje, E.; Fristedt, R.; Rosqvist, F.; Landberg, R.; Iggman, D. (2025). Cafestol and kahweol concentrations in workplace machine coffee compared with conventional brewing methods. Nutrition, Metabolism and Cardiovascular Diseases 35, 103933. doi:10.1016/j.numecd.2025.103933 · other copies
- AWuerges, K. L.; Dias, R. C. E.; Viegas, M. C.; Benassi, M. T. (2020). Kahweol and cafestol in coffee brews: comparison of preparation methods. Revista Ciência Agronômica 51(1). doi:10.5935/1806-6690.20200005 · other copies
- BMestdagh, F.; Glabasnia, A.; Giuliano, P. (2017). The Brew – Extracting for Excellence. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 15, pp. 355–380. doi:10.1016/B978-0-12-803520-7.00015-3 · other copies
- BFolmer, B.; Blank, I.; Hofmann, T. (2017). Crema – Formation, Stabilization, and Sensation. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 17, pp. 399–417. doi:10.1016/B978-0-12-803520-7.00017-7 · other copies
- AIlly, E.; Navarini, L. (2011). Neglected Food Bubbles: The Espresso Coffee Foam. Food Biophysics 6(3), 335–348. doi:10.1007/s11483-011-9220-5 · other copies
- ASmrke, S.; Wellinger, M.; Suzuki, T.; Balsiger, F.; Opitz, S. E. W.; Yeretzian, C. (2018). Time-Resolved Gravimetric Method To Assess Degassing of Roasted Coffee. Journal of Agricultural and Food Chemistry 66(21), 5293–5300. doi:10.1021/acs.jafc.7b03310 · other copies
Tier A · authoritative · B · trade and expert · C · general. Supporting quotes for every source are in the data version of this entry. If a publisher’s site refuses access, try the doi link or “other copies”.
Discussion
A space for readers to share experience from the farm, the roastery or the bar, ask each other and offer other views. The Coffearium team does not reply here. Please keep to the community rules; to change the entry itself, use Suggest a change.