Antioxidants and chelators: keeping soap from going off

A bar of handmade soap carries its oils’ weak points with it. The double bonds that keep olive and sunflower oil liquid survive saponification, and over months they react with oxygen. What comes of it is the stale smell of old soap and the yellow-orange blotches soapmakers call dreaded orange spots. This guide explains the chemistry, then takes each antioxidant and chelator in turn: what it does, the dose our additive database carries and where that dose comes from, and what testing stands behind it. The orange spots guide covers diagnosing spotted bars and what to do with them.

How soap goes off

Oxidation starts where a fatty-acid chain has a double bond, and above all at a carbon that sits between two of them. Linoleic acid has one such carbon and linolenic acid has two. A hydrogen atom there comes away far more easily than anywhere on a saturated chain, so the rate climbs steeply with each double bond: a 2025 review of oil oxidation puts a chain with two double bonds at 10 to 40 times the rate of one with a single bond (Gharby et al.). Oleic acid, the main acid of olive oil, has one double bond. Lauric and stearic acid have none.

That makes the oil list a fair first guide to how long a soap will keep.

Ranges from the Codex standards and pharmacopoeia monographs on each oil’s page. The iodine value counts all double bonds, so it tracks the same thing more roughly; the iodine value guide explains it.

The oils soapmakers most often blame for orange spots sit in the lower half of that table, which is what the chemistry predicts. No controlled soap test we know of has ranked them against each other, so treat the table as a reason for care rather than a scoreboard.

The soap goes off too, not only the superfat

It is easy to assume that only the unsaponified oil can go rancid. Saponification does not touch the double bonds, though: sodium oleate and sodium linoleate carry the same chains as the oil they came from. Kevin Dunn tested the point directly. Olive oil soaps with no superfat and with 10 % turned noticeably orange at about the same time, some 75 hours into his accelerated test. The 10 % soap ended up more orange, so superfat adds fuel, but taking it away did not stop the spots. The superfat guide covers the rest of that trade-off.

Iron and copper speed it up

Traces of iron and copper make oxidation far faster, and the amounts that matter are tiny. A Lever Brothers patent on superfatted soap blames metals for “deterioration of base odour, spotting and discoloration” of bars, and reports that “as little as 0.2 part per million copper, in the absence of preservatives, has been shown to have a significant effect on colour and odour deterioration”. Its test soap held 0.3 ppm of copper and 4.6 ppm of iron before any was added, and it notes that up to 1 ppm of copper and 10 ppm of iron “are frequently found in practice”.

Such traces are hard to avoid. The Codex standard for vegetable oils allows up to 1.5 mg of iron and 0.1 mg of copper per kilogram of refined oil, and 5.0 mg and 0.4 mg in virgin oils. Water can add more; the hard water guide works through how much. Dunn’s experiment makes the point without setting out to: his soaps were made with distilled water, and EDTA, whose job is to bind metal ions, still more than quadrupled the time they took to turn.

Two defences, and why neither is a preservative

The additives in this guide fall into those two groups. Antioxidants (rosemary extract, BHT, tocopherols) break the chain. Chelators (EDTA, citrate, GLDA, gluconate, phytate) take the metals out of play.

Neither makes soap keep forever. An antioxidant is consumed as it works, so what it buys is time: a longer delay, the induction period, before oxidation takes off. A chelator is not consumed, and even a small dose is large next to the metals: 0.1 % of a 1.4 kg batch is 1.4 g of tetrasodium EDTA, or 3.7 mmol, while a kilogram of virgin oil at the Codex maxima carries about 0.1 mmol of iron and copper together. But a chelator does nothing about oxygen, heat or light, which drive the reaction without any metal’s help.

Neither is a preservative either. In EU cosmetics law a preservative is a substance “exclusively or mainly intended to inhibit the development of micro-organisms in the cosmetic product”. Vitamin E and rosemary extract slow the oxidation of oils. They do nothing against mould or bacteria, so they do not make a lotion or a liquid soap safe to keep.

The one controlled test

Almost everything measured about these additives in soap comes from one study. From the autumn of 2004, Dunn and his students at Hampden-Sydney College made olive oil soap from 100.0 g of olive oil, 13.0 g of sodium hydroxide and 26.0 g of distilled water, with 0.1 g of an additive, 0.1 % of the oil weight. They held each soap at 60 °C (140 °F) for 4 hours to finish saponifying, sealed 1 g samples in vials of oxygen or of nitrogen, kept them at 60 °C and scanned them daily, measuring colour saturation. Above about 30 % saturation the soap looks noticeably orange.

Dunn, “The Dreaded Orange Spot”, The Handcrafted SoapMaker, winter 2005/06. The test stopped at about 300 hours, so “beyond 300” means it never took off while they watched.

Dunn’s own suggestion was 0.1 % of the oil weight each of BHT and sodium citrate for soap that might avoid spots altogether, or 0.1 % rosemary extract for makers whose customers want natural ingredients. In Scientific Soapmaking he writes that “the most potent preservative was edta”, effective “when added to the lye at a rate of as little as 0.3 ppt”.

Read the result for what it is. It is one recipe, a single-oil olive soap, at one dose. It is an accelerated test, 60 °C in pure oxygen, and nobody has worked out how many months on a shelf 300 hours stands for. It measures overall colour rather than smell or spots, and spots are the less understood of the two: Lever’s patent calls oxidation “the most probable cause of rancidity” in bars but says that “the development of spotting or overall discoloration is less clearly understood. It often accompanies deterioration of base odour.” Each additive appears as one sample in each gas, without repeats or error bars, and the article does not give the strength of the rosemary extract or the form of the EDTA. It was published in a guild journal, not a peer-reviewed one. It is also the only controlled test of these additives in soap that we have found. Some of its findings fit the chemistry, such as the strong chelator working where the weak one did not. Others are unexplained: vitamin E is a phenolic antioxidant too, and it failed where BHT and rosemary extract succeeded.

The antioxidants

Our additive database carries four, with doses traced to their sources. The calculator fills in the suggested dose, and its additive guide lists the others with their sources.

Rosemary extract

Rosemary oleoresin extract, INCI Rosmarinus Officinalis Leaf Extract, is an oil-soluble concentrate whose active part is a group of phenolic compounds, chiefly carnosic acid and carnosol. It is sold at different strengths, commonly 2, 5 or about 7 % carnosic acid, and the same INCI name also covers water and glycerin extracts of rosemary, which are a different product. Rosemary essential oil is a different product again and no substitute.

The suggested dose is Dunn’s, 0.1 % of the oils: 1 g for 1000 g of oils, which is roughly 30 to 50 drops by the drop weights in our data. Craft sources use less, 0.02 to 0.08 % of the oils or no more than 0.05 %. It goes into the oils while they are fresh, before they meet the lye. That timing follows from the chemistry: an antioxidant delays the chain reaction but cannot undo hydroperoxides that are already there, so it does most for oil that has not started to turn.

More is not better. In light-exposed soybean oil, Hall and Cuppett found rosemary extract worked best between 0.02 and 0.05 % and sped oxidation up at both 0.01 % and 0.5 %. That was oil under light, not soap, but it is a reason to weigh ROE rather than pour it.

Vitamin E

Tocopherols occur naturally in vegetable oils (the Codex standard tabulates them oil by oil) and are sold as antioxidants, so vitamin E is the one most soapmakers reach for. In Dunn’s test at 0.1 % of the oils it did nothing: the soap turned orange as fast as plain soap. Why it failed where rosemary extract worked is not known. Our database gives Cavitch’s dose, about 0.06 % of the total ingredients, in the oils before the lye.

Check which form you have. “Vitamin E oil” sold for skin is often tocopheryl acetate, an ester, which lye can split like any other ester. And vitamin E is sold to soapmakers as a preservative far more often than the evidence allows: by the definition above it is not one.

BHT

Butylated hydroxytoluene, a synthetic phenol, doubled the induction period in Dunn’s test on its own, to about 150 hours, and with sodium citrate gave the best result of anything he tried. The suggested dose is his, 0.1 % of the oils. It melts at about 70 °C, so it goes into the warm oils to dissolve. In the EU it has been limited since 2023 to 0.8 % in rinse-off products such as soap, after an assessment of possible hormone disruption. It is listed on a label as BHT.

Grapefruit seed extract

It did nothing in Dunn’s test. It also has a history. Von Woedtke and colleagues found the preservative benzethonium chloride in every antimicrobially active grapefruit seed extract they tested, and Takeoka and colleagues measured it at 8 % of a liquid extract. Whatever activity a bottle has may come from something that is not on its label. It is not an antioxidant to rely on.

The chelators

From our additive database. “Counted” means the calculator adds the lye the additive consumes. The first figure is the one it fills in.

EDTA

Tetrasodium EDTA is the chelator with the best record in soap. Dow sells it for bar soap at 0.1 to 0.2 %, where it “helps prevent softening, brown-spotting, rancidity, cracking and discoloration due to metal ions”. Lever’s patent paired it with etidronic acid at about 0.02 % each and found the two together worked better than either alone. It needs no extra lye.

Put it in with the water, not the oils: Dow’s chelators are “only soluble in aqueous solutions”, and its guidance is to add one “as soon as practical in the process, for example, during the water phase”. Dunn adds it to the lye. One supplier guide warns against putting it into strong lye solution, but Nouryon’s data show EDTA still dissolving to about 6 % by weight (counted as the acid) in 28 % caustic soda, far above the few tenths of a percent a soap dose makes. Dissolving it in the water before the lye goes in sidesteps the question.

Disodium EDTA is the form sold for neutral cosmetics. In lye it takes two sodium hydroxides and becomes the tetrasodium form, and the calculator adds the 0.215 g of lye per gram that this takes.

Sodium citrate, and citric acid that becomes it

Sodium citrate is a weaker chelator than EDTA. Alone, at 0.1 % of the oils, it made no difference in Dunn’s test; with BHT it gave the best result of all, which suggests it works as a partner rather than on its own. Craft doses are much higher, 1.3 to 3.9 % of the oils, and are sold mainly as a hard-water remedy, though what a bar releases in one wash is far too little to soften the water (the hard water guide does the sums). Too much leaves white crystals on the bars.

Citric acid in the lye water turns into sodium citrate, using lye as it does. In 1000 g of olive oil soap at a 5 % superfat, 10 g of citric acid makes the calculator raise the sodium hydroxide from 128.7 g to 134.9 g, so the superfat stays where you set it, and the bar ends up with the same citrate as about 15 g of sodium citrate dihydrate. Leave the lye unadjusted and the acid simply raises the superfat. Add the lye slowly: the acid fizzes and heats where the two meet.

GLDA, gluconate and phytate

These three are sold as gentler or greener alternatives to EDTA, and none has been tested against rancidity in soap that we could find. Their doses in our database come from cosmetics in general, or for gluconate from one supplier guide.

GLDA is made from monosodium glutamate, and its maker sells it as readily biodegradable, setting it against classic chelators such as EDTA on exactly that point. COSMOS, the organic cosmetics standard, allows it in soap. Its maker’s figures are worth reading closely. GLDA holds copper up to about pH 12 but holds iron only up to about pH 8, above which iron settles out as hydroxide. Soap sits well above pH 8, so GLDA should keep copper in check and cannot keep iron in solution. Whether iron that has settled out still speeds rancidity inside a bar has not been tested, as far as we know.

Sodium gluconate is the one sold on its strength in alkali: its maker says that in strongly alkaline solutions it “surpasses all other chelating agents, such as EDTA”. Dissolved, it spoils within days, so it should not go into a lye solution kept for later. Sodium phytate grades differ: check that yours is fully neutralised, at about pH 12, or it will use lye that the calculator does not count.

Putting it together

For a recipe heavy in the oils at the bottom of the table above, the best-supported combination is EDTA in the water and rosemary extract in fresh oils, each at the doses Dunn tested. If synthetic ingredients suit your customers, BHT with sodium citrate did better still. The budget bar in our recipe collection, a quarter of it canola oil, shows what the first combination looks like: 1.5 g of tetrasodium EDTA is 0.1 % of its 1433 g batch, and 1 g of rosemary extract is 0.1 % of its oils. Neither changes the 142.1 g of sodium hydroxide it needs. Our Aleppo-style laurel bar uses the rosemary half on its own: 1 g, 0.1 % of its oils, stirred into the fresh oils to see it through a six-month cure.

The additives are a second line. The first is the formula: Dunn’s coconut oil soap, made from a fat with almost no double bonds, never turned orange at all. After that come fresh oils, since an antioxidant cannot rescue oil that has already started to turn, and bars cured and stored away from heat and light, both of which speed the reaction.

Whatever you add, label it. Dunn’s closing advice was to tell customers exactly what is in the bar, because people choose soap for reasons of their own, and “natural” rosemary extract and “synthetic” BHT each matter to someone.