Iron oxides, titanium dioxide and the other mineral pigments in balms

CI 77491, 77492, 77499 and 77891 explained: opacity, tinting strength, use at 0.1 to 10 percent, dispersion in castor oil and where each is banned for lip use.

Four CI numbers do almost all the work in a tinted balm: 77491, 77492 and 77499 for the three iron oxides, and 77891 for titanium dioxide. Everything else in the mineral pigment box is either an effect pigment or a colour you cannot legally put on lips in the United States. This page covers what each material is, how strongly it tints, how much you need, how to grind it so it leaves no specks, and the purity documentation to insist on.

• Iron oxides: 21 CFR 73.2250
• Lead limit 20 ppm in the oxide
• TiO2 opacity peaks near 250 nm
• Pigment density 4 to 5 g/cm3
• Must be synthetically prepared

The three iron oxides and what they physically are

Iron oxide is a family name for three minerals with three crystal habits, and the differences show up in the pot.

CI 77491, red. Anhydrous iron(III) oxide, the same compound as the mineral haematite, in roughly nodular particles of 0.1 to 0.6 microns. The whole earth-tone range is built on it, and the same chemical runs from bright orange-red at fine particle sizes to deep maroon-brown at coarse ones.

CI 77492, yellow. Hydrated iron(III) oxide, chemically goethite, FeO(OH), in needles 0.1 to 0.8 microns long. It is the weakest tinter and the one that most needs grinding, because needles interlock into agglomerates that resist wetting. It is a hydrate, so hard heating dehydrates it to red, but that threshold sits far above 200 C and is a manufacturing concern rather than a balm one.

CI 77499, black. Magnetite, a mixed iron(II) and iron(III) oxide, and the strongest tinter in the set by a wide margin. It is also magnetic, which is worth knowing because a stainless spatula will not attract it but a magnetic stirrer bar will drag it out of a melt into a black collar.

Brown iron oxides sold as a single powder are blends of the three, not a fourth species. Buying red, yellow and black and mixing your own is cheaper, more flexible and easier to reproduce than stocking eight pre-blended browns.

Tinting strength, and mixing a shade you can hit twice

Tinting strength is a pigment's ability to change the colour of a mixture, measured in the coatings trade by mulling a fixed pigment-to-white ratio against a reference standard (ASTM D387). It has nothing to do with how dark the powder looks in the jar. Black iron oxide is roughly an order of magnitude stronger than yellow, which is why "a pinch of black" is not a recipe.

Two rules follow. Build a shade from the weakest pigment upward, weighing yellow and red first and bringing in black as a fraction of the total pigment load. And write shade recipes as ratios of the pigment blend rather than percentages of the batch, so a 20 g test and a 2 kg run use the same numbers. A workable warm nude, as a ratio of total pigment: 55 red, 25 yellow, 3 black, 17 titanium dioxide. A cool berry: 70 red, 5 yellow, 8 black, 17 titanium dioxide. Both are starting points for a skin swatch.

Titanium dioxide: opacity, particle size and the nano question

Titanium dioxide is a white pigment because its refractive index is around 2.6 to 2.7, higher than almost any other material a formulator can use, and scattering rises with the mismatch between pigment and the medium around it. Scattering peaks when the particle diameter is roughly half the wavelength of the light being scattered, which puts the visible optimum at about 200 to 300 nm. Pigment grade titanium dioxide is manufactured to sit exactly there.

That explains the most common purchasing error in tinted balms. A nano grade, sold for sunscreens at 15 to 50 nm, sits far below the scattering optimum for visible light and is close to transparent on skin, working in the ultraviolet instead. Buy it expecting a white and you get a barely lightened balm carrying an unintended UV filter, with the regulatory consequences described in SPF in lip balm. Pigment grade is a different product.

The EU defines a nanomaterial in Article 2 of Regulation 1223/2009 as an insoluble or biopersistent, intentionally manufactured material with one or more external dimensions, or an internal structure, on the scale of 1 to 100 nm. Nanomaterials must be notified to the Commission six months before market and declared on the ingredient list with "nano" in brackets after the name. Titanium dioxide is also in Annex III with a condition restricting uses that can lead to inhalation exposure of the user's lungs, following its classification as a suspected carcinogen by inhalation in powder form. A poured balm is a solid, so the restriction does not bite. The person weighing the loose powder is a different matter.

Buy a surface treated grade. Uncoated titanium dioxide, particularly anatase, is photocatalytic: under ultraviolet light it generates radicals that attack the oils around it through the mechanism in rancidity and oxidation. Cosmetic grades are normally coated with alumina, silica or a hydrophobic treatment, and a coated grade also disperses better and feels less chalky. Above about 4 percent of the formula it reads as dry and powdery whatever the grade.

The pigments that are not permitted on lips in the US

This is where makers get caught, because these materials are sold openly as cosmetic pigments and are legal in other cosmetic uses.

Under 21 CFR Part 73 Subpart C, ultramarines (CI 77007), chromium oxide greens (CI 77288), chromium hydroxide green (CI 77289), manganese violet (CI 77742) and ferric ferrocyanide (CI 77510) are listed for colouring externally applied cosmetics, including the eye area. A lip product is not externally applied, because part of it is ingested, so none of them is permitted in a lip balm, lipstick or lip scrub sold in the United States. Iron oxides and titanium dioxide are listed for cosmetics generally, which includes lips.

The EU and UK are more permissive. Annex IV of Regulation 1223/2009 lists permitted colourants with a field of application column separating products that may contact mucous membranes from those that may not, and these blue, green and violet minerals are not held away from lip use the way the US list holds them. The same recipe can be legal in London and not in Boston.

Two things catch people out. A green or blue mica gets its colour from something, and if that something is chromium oxide or ultramarine the composite pigment is not a US lip colour whatever the base mica's status, so check every component, as mica and pearl pigments sets out. And selling internationally means meeting the stricter list, not the one where you live. The duties are in labelling cosmetics in the US and selling balms in the UK and EU.

Use levels and the loading arithmetic

Loading goes wrong in both directions: too little to see, or so much that the balm becomes a crayon. The bands below are weight percent of the finished formula, counting all pigments together.

The arithmetic, worked. A 60 g batch of medium coverage warm nude at 3.5 percent total pigment needs 2.1 g of pigment. The 55:25:3:17 ratio gives 1.155 g red, 0.525 g yellow, 0.063 g black and 0.357 g titanium dioxide. The black is the problem: 63 mg is at the edge of what a 0.01 g scale weighs honestly, and a 20 mg error there is a visible shift in the finished stick. That is the real argument for a pre-milled paste, and for the discipline in weighing and calibration.

Make the paste at 30 percent pigment and you need 7 g of paste to deliver 2.1 g of pigment, of which 4.9 g is castor oil. Since the base formula probably already carries 10 to 15 percent castor oil, subtract that 4.9 g from the castor oil line rather than adding it on top, or the totals stop adding to 100. The batch calculator will hold the totals for you while you move the numbers around, and the full percentage layout of a finished tinted stick is in tinted lip balm.

Dispersion: castor oil, a muller and the speck failure

Pigment powder is not single particles. It is agglomerates, clusters held together by surface forces with air trapped between them, and colour development means wetting each particle, breaking the clusters and keeping them apart. Stirring powder into a molten balm does none of that, which is why the result is speckled, weak and draggy.

Castor oil is the standard grinding medium because ricinoleic acid carries a hydroxyl group on the fatty chain. That makes the oil unusually polar for a triglyceride, so it wets a mineral surface and displaces air, and unusually viscous, so shear from the tool reaches the agglomerates instead of flowing around them. The chemistry is in castor oil.

How the grinding gets done matters more than makers expect:

1. Mortar and pestle. Adequate for red and yellow, poor for black and titanium dioxide, because a pestle applies compression rather than the shear that splits agglomerates. Unglazed porcelain also holds pigment in its pores and contaminates the next shade.
2. Glass muller on a ground glass plate. The right small-scale tool. Flat-on-flat contact generates high shear in a thin film, which is what the job needs, and glass cleans completely. Five to ten minutes of real work per shade.
3. Three roll mill. The industrial answer, and the reason commercial lipstick has no specks. Three rollers turn at increasing speeds with gaps from roughly 25 microns down to 5, so the paste crosses two shear zones per pass and is usually run three times or more. Nothing hand-held reaches the same fineness.
4. Check the result. Draw the paste down thinly on glass with a straight edge. Specks, streaks or gritty lines mean it is not finished. The coatings trade does the same check with a grind gauge (ASTM D1210), reading the depth at which particles first break the surface of a drawn film.

Name the failure precisely. Skin reads particles above roughly 20 to 25 microns as grit and the eye sees a dark agglomerate as a speck at about the same size. Primary pigment particles are well under one micron, so every speck in a finished balm is an agglomerate you did not break. Unlike a wax speck from an incomplete melt, it will not remelt away. Poor dispersion also causes most of the patchiness in streaky or mottled colour, and hard particles trapped between balm and lip are a direct cause of drag.

One more physical point. Mineral pigments are four to five times denser than the melt around them, so a thin, hot balm lets them fall. Stir the paste in, keep stirring while the melt cools and thickens, and pour promptly rather than dribbling out the last of the jug. The settling mechanism is in pigment sinking to the bottom and the windows in pour temperatures.

Heavy metals and the certificate to demand

Mineral pigments are made from mined precursors, so trace heavy metals are inherent rather than accidental, and for lip products it matters more than anywhere else in cosmetics because a fraction of the product is swallowed.

In the EU, lead, arsenic, mercury and cadmium are prohibited under Annex II of Regulation 1223/2009, and Article 17 permits only traces that are technically unavoidable and leave the product safe. There is no single published number, so the burden falls on the safety assessment rather than a table, and your assessor will want the pigment data. Health Canada publishes impurity guidance with limits of the same order.

Demand five things per batch and keep them with your batch records: the CI number and full composition; a statement that the pigment meets the relevant colour additive specification for your market; heavy metal results by ICP-MS covering at least lead, arsenic, mercury, cadmium and antimony; the surface treatment if any; and, for titanium dioxide, whether the grade is nano under the EU definition. A supplier who cannot produce that is selling craft colourant, as sourcing ingredients argues. In the EU and UK the same paperwork feeds the product information file behind your safety assessment.

Why cosmetic iron oxide has to be synthetic

Natural ochres have coloured things for forty thousand years, which makes the marketing appeal obvious and the regulatory answer unhelpful. The US listing at 21 CFR 73.2250 defines the colour additive as synthetically prepared iron oxides, free from admixture with other substances. Mined ochre does not qualify, however pure the deposit.

The reasoning is what comes out of the ground alongside iron oxide: variable lead, manganese, nickel, chromium and crystalline silica, in proportions that change with the seam. Synthetic routes, whether precipitation from iron salts or the Penniman and Laux processes, give a defined composition, a controlled particle size distribution and a reproducible shade. This is a rare case where the manufactured material is unambiguously better than the mined one.

None of this changes what you write on the label. The INCI name is Iron Oxides, or CI 77491 and its siblings under the EU convention, with no distinction drawn between manufacturing routes. Nor does it license a claim: a mineral pigment is a colour additive, not an active, so describing it as skin-nourishing crosses into cosmetic and drug claims territory.

Shade drift between batches, and how to stop it

Mineral pigments do not fade. They are fully oxidised, thermally stable well beyond any balm temperature and genuinely lightfast, so a stick tinted with iron oxides looks the same after a summer on a desk. When a tinted line drifts, the pigment is almost never degrading. Three causes account for most of it:

• Supply

  A new pigment lot

  Shade is controlled against a master standard within a tolerance, not to a point, and lot tolerances can exceed the difference an eye can see. Buy one large lot per product line.

• Process

  Uneven dispersion

  An underground paste reads lighter at the same loading, because part of the pigment surface never meets light. Two grinding times are two shades.

• Weighing

  The black pigment

  At tens of milligrams, ordinary scale error is a visible shade change. Dosing a pre-milled paste removes the problem.

The controls are dull and they work: one pigment lot per product line, a milled masterbatch at a recorded ratio, a retained reference sample kept in the dark, and a swatch of every batch compared against it on the same substrate under the same light. Judge on skin from a cooled sample, never from the pot, because molten balm always looks darker than the set stick. If drift runs across your whole range and not just the colour, the systematic approach is in batch to batch inconsistency. Colour that seems to weaken on the shelf is a separate question, covered in tinted balm colour fading.

The decision rule, and where mineral pigments stop

The rule is short. Buy synthetic red, yellow and black iron oxide plus a coated pigment grade titanium dioxide, all four with certificates, and mix your own shades. Those four cover every nude, brick, berry, brown and earth tone, in both major markets, at 0.1 to 10 percent, with no lightfastness worry and no approval ambiguity.

What they cannot do is worth stating plainly. Minerals cannot make a clean bright pink, a true blue-toned crimson or any vivid non-earth shade; those need certified organic lakes or carmine, with the trade-offs in colourants and tints. They cannot give shimmer, which is a mica job. They add drag and cost gloss in proportion to loading, so an opaque mineral balm needs more castor oil and a more careful base than a clear one, for the reasons in lip colour and gloss physics. And no hand grinding matches a three roll mill, so at real coverage levels a hand-ground balm feels slightly grainier than a manufactured lipstick. That is a limit of the equipment, not the formula.

Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.