Coffee Seed Oil

UNSAPONIFIABLE MATTER 8.09 +/- 0.39 % IS A DIRECT MEASUREMENT, not a quoted figure: Table 1 of Raba et al., PLoS ONE 10(9):e0138080 (2015), column GCO - crude green coffee oil, Soxhlet-extracted with petroleum ether from Arabica Typica beans grown at Guaxupe, Brazil. The same table and column give saponification value 185.86 +/- 3.1 mg KOH/g, iodine value 88.91 +/- 2.47 g I2/100 g, free fatty acids 1.91 % as oleic and peroxide value 3.20 meq O2/kg, all by standard oil-analysis methods. THIS IS WHY THE OIL BEHAVES ODDLY IN A RECIPE: about an twelfth of the mass is unsaponifiable - diterpene alcohols of the kaurene family (cafestol, kahweol, 16-O-methylcafestol) and their fatty esters, plus sterols, tocopherols and waxes - and unsaponifiable matter does not react with lye. The measured 185.86 sits about 5.5 % below the 196.7 mg KOH/g that the recorded fatty-acid profile implies by stoichiometry, and the unsaponifiable fraction accounts for that gap almost exactly. A calculator that ignores it and saponifies coffee oil as if it were pure triglyceride will leave the batch lye-heavy. Note also that whatever is not saponified stays in the bar as free oil, so coffee seed oil superfats harder than its usage rate suggests, and the diterpenes are heat- and light-sensitive. THE FATTY-ACID PROFILE comes from a different sample: Table 2 of Dong et al., Ultrasonics Sonochemistry 74:105578 (2021), Coffea arabica cv. Catimor 7963 grown in Yunnan, China, wet-processed, extracted four ways (ultrasonic/microwave-assisted, ultrasound-assisted, microwave-assisted, pressurised liquid). The min/max here span all four columns: myristic 0.25-0.29, palmitic 36.57-38.71, palmitoleic 0.22-0.26, stearic 6.70-7.11, oleic 8.54-9.94, linoleic 40.67-43.77, arachidic 2.77-2.90. Coffee oil is unusual in being roughly half saturated (SFA 46.7-48.9 %) and yet soft, because the saturated half is almost all palmitic and the unsaturated half is almost all linoleic - there is very little oleic. Recorded ranges sum to about 99.4 %; the paper also reports C15:0 0.10-0.12, C17:0 0.32-0.43 and C21:0 0.07-0.11, which have no key in this schema, and it did not measure alpha-linolenic, behenic or lignoceric, which other coffee-lipid work reports at roughly 1-2 %, 0.6-0.9 % and 0.2-0.4 %. TWO DISAGREEMENTS WERE JUDGED. First: Dong et al. also print saponification values, but they range from 88.49 to 230.04 mg KOH/g across four extractions of the same beans, a spread no oil can have, so none of those figures is used and the Raba value stands alone. Second: reviews of coffee lipids describe the unsaponifiable fraction as much larger than 8 % - the composition table reproduced from Speer & Kolling-Speer in Molecules 30(14):2951 (2025) puts diterpene esters alone at 18.5 % of the lipid, with free diterpene alcohols 0.4 %, sterol esters 3.2 %, sterols 2.2 % and tocopherols 0.04-0.06 %, and the Foods 2024 review cites 12.8 g/100 g for a screw-pressed oil. Those are compositional accountings, not gravimetric unsaponifiable-matter determinations on one sample, and they include ester-bound diterpenes whose fatty-acid half does saponify. 8.09 % is recorded because it is the measured number, on a named sample, by the standard method; a soapmaker should treat 8 % as a floor and expect a pressed oil to be less predictable. Melting point not measured (liquid at room temperature). INCI not verified on CosIng in this session, left null.

This record is part of the open ingredient database (ODbL). Found an error or a better primary source? The dataset is on GitHub; corrections with a document link are welcome.