Colour that fades in the tube, and the pigments that cannot take light or heat

Natural colourants photodegrade in weeks while iron oxides hold for years. A window sill test, the packaging change and the pigment swaps that stop fading.

A tinted balm that looked right on the bench and looks washed out six weeks later has not settled, absorbed or sunk. A chromophore has been destroyed, and the pigment family you chose decides whether that takes three weeks or outlasts the oil around it. Fade is not reversible, so everything useful on this page is either diagnosis or prevention.

• Iron oxides: no measurable fade
• Plant colours: 2 to 8 weeks in light
• 40 C roughly halves time to visible fade
• Foil-wrapped twin is the control

Four different faults all get called fading

Photodegradation, poor dispersion, settling and oxidation of the base all produce a stick that is the wrong colour, and only one of them is fixed by changing pigment. The distinguishing evidence is geography: real photofade is worst where the light fell.

What makes a colourant fugitive

Colour comes from a chromophore, a run of alternating single and double bonds long enough to absorb visible light. The part of the molecule that gives the colour is also the part that absorbs the energy. A photon lands on the chromophore, lifts it to an excited state, and the excited molecule either reacts directly with dissolved oxygen or hands its energy to an oxygen molecule to make singlet oxygen, which then attacks the double bonds. Every bond that breaks shortens the conjugated run and shifts the absorption toward the ultraviolet, so the colour first changes hue, then goes.

Three chromophore classes cover almost every plant colour a balm maker meets. Annatto (bixin and norbixin) and paprika oleoresin (capsanthin and capsorubin) are carotenoids, long polyene chains that are close relatives of the pigments in carrot and tomato, and their photo-oxidation and cis-trans isomerisation are among the best characterised instabilities in food colour chemistry. Beetroot colour is betanin, a betalain, which is water soluble, degrades readily with light, heat and oxygen, and barely transfers into an oil phase in the first place. Alkanet gives alkannin and shikonin, naphthoquinones that are genuinely oil soluble, genuinely red-purple, and both light sensitive and pH sensitive, so the colour drifts on the lip as well as on the shelf. The regulatory position on all of these is covered under colourants and tints, and it is the shorter argument: in the US and the EU a colour used to colour a cosmetic has to appear on a positive list, and these do not.

Two consequences follow that makers usually get wrong. Fading is a surface phenomenon first, because the pigment nearest the light absorbs it, which is why a faded stick often has a pale skin over stronger colour. And an antioxidant slows the process without stopping it, because the primary event is absorption of a photon, not a radical chain that a tocopherol can interrupt.

Why iron oxides do not fade

Iron oxides are not molecules with a fragile conjugated chain. They are inorganic solids, CI 77491 red, CI 77492 yellow and CI 77499 black, in which colour comes from electronic transitions inside a metal oxide lattice. There is nothing left to oxidise, no double bond to cleave, and their thermal decomposition sits hundreds of degrees above anything a balm will see. A stick tinted with iron oxides looks the same after a summer on a desk as it did on the day it was poured, and people reach past them for the shade range, not for the durability. The CI numbers, tinting strengths and grinding method are set out under iron oxides and mineral pigments.

Two caveats belong here rather than in a footnote. Uncoated anatase titanium dioxide is photocatalytic: under ultraviolet it generates reactive oxygen species at its surface, which will not fade the pigment but can accelerate oxidation of the oils around it. Cosmetic grades are normally surface treated with alumina or silica precisely to suppress that, and it is worth asking a supplier which grade you are buying. Ultramarines are stable to light and unstable to acid, releasing hydrogen sulphide and losing colour if a formula carries an acidic active. The effect pigments described under mica and pearl pigments do not fade either, though a mica-only tint can look duller over time because the base under it has darkened.

Lakes are the interesting middle case

An aluminium lake is a soluble dye precipitated onto an alumina substrate so that it behaves as an insoluble pigment. That construction gives it better lightfastness than the free dye and a cleaner, brighter shade than any mineral can reach, at the cost of two failure modes minerals do not have.

The first is pH and metal sensitivity. The dye is held on the substrate by an interaction that depends on pH, so an acidic or alkaline component can strip it off, and free polyvalent metal ions shift or dull the shade. An anhydrous balm carries less of this risk than a water-containing product, because there is no aqueous phase to carry protons around, but it is not zero: honey, an acidic active or a poorly rinsed steel vessel supplies what the reaction needs.

The second is bleed. A well-made lake keeps its dye; a poorly made one, or one pushed outside its pH range, slowly releases dye into the oil phase. The result is a balm whose colour looks less crisp month by month, stains a tissue, and transfers more heavily than it did when new. The test is trivial: press a fresh smear and an aged smear onto white blotting paper, leave them an hour, and compare the haloes of colour that spread beyond the smear. A growing halo is dye migrating out of the pigment, and no amount of grinding will fix it.

How long each family actually holds

Read the last two rows as the argument. The gap between a mineral pigment and a plant colour is not a matter of degree that better packaging closes. It is the difference between a material that has already reacted with all the oxygen it is going to and one whose colour is a reaction waiting to happen.

The window sill test

This is the cheapest stability experiment in balm making, and it works because the sample acts as its own control, so nothing rests on your memory of the original shade.

1. Fill four identical tubes from one melt. Same batch, same fill temperature, same pack. Two go on the test, one is wrapped tightly in aluminium foil, one goes straight into a freezer bag in the freezer as the reference.
2. Put one test tube and the foil-wrapped tube on the same window sill. Both experience the same temperature. Only one sees light, so any difference between them is photochemical rather than thermal. South facing in the northern hemisphere, or the brightest sill you have.
3. Put the second test tube in a 40 C oven in the dark. That is the standard accelerated arm described under accelerated ageing, and it isolates the heat pathway from the light pathway.
4. Photograph all four at day zero. Same light, same white balance locked manually, a neutral grey card in frame, sticks wound out to the same height. Phone auto white balance will silently correct the very change you are trying to measure.
5. Assess at 2, 4 and 8 weeks. Draw a swatch of each onto white paper side by side, photograph the set, and only then look at the tubes. Swatches show hue shifts that a stick hides.
6. Write down what you see, not what you conclude. "Window tube two shades lighter and more orange, foil tube unchanged, oven tube slightly duller" is data. File it with the batch record so the next reformulation starts from evidence.

Two weeks separates the hopeless colourants from the plausible ones, four ranks the plausible ones, and eight tells you whether an opaque pack is enough on its own. The formal version of the experiment is ICH Q1B photostability testing, which specifies not less than 1.2 million lux hours of visible light and not less than 200 watt hours per square metre of near ultraviolet, measured rather than guessed. A window sill delivers an unknown dose, so it ranks materials against each other honestly and cannot produce a shelf life. That job belongs to the retained samples described under shelf life testing.

Heat is the second driver, and it is separable

Light gets the blame because the evidence is visible, but temperature multiplies whatever the light starts. Carotenoid and betalain degradation accelerates strongly with heat, and the working convention in cosmetic stability testing is a Q10 of about 2, so the rate roughly doubles per 10 C, putting 40 C at four times the rate of a 20 C reference. On a lit shelf the observed effect of heat is smaller than that arithmetic implies, closer to halving the time to a visible change, because the photochemical step that dominates in light is barely temperature dependent. Heat accelerates the thermal pathway hard and the photochemical one hardly at all.

The practical version is that a tinted balm in a car in July is running two clocks at once, which is why the same product survives a winter and fails a summer. If your stock ships in warm weather, the packaging and the pigment choice both have to hold up under the conditions in shipping in hot weather, and colour drift belongs on the checklist alongside the softening covered under balm melting in the heat.

Why doubling the pigment does not work

The instinct when a colour fades is to load more of it, and it fails for a reason worth understanding. Fading happens where the light is absorbed, in the outer fraction of a millimetre. Doubling the concentration darkens the whole stick but does nothing for that surface layer, so you get a deeper colour that develops a pale skin at the same rate, then a visibly two-tone product as the customer winds it past the bleached zone.

Worse, fugitive colourants do not fade neutrally. A carotenoid loses the long-wavelength end of its absorption first, so an annatto orange goes yellow before it goes pale. Overdosing therefore buys you a stronger version of the wrong hue rather than more time at the right one. And there is a hard ceiling on loading anyway: past roughly 5% of most pigment systems in an anhydrous base a lip balm goes draggy and chalky, which is a different complaint entirely, covered under drag and poor glide.

Reformulating on pigment family is the move that works. Match the shade you wanted with iron oxides plus titanium dioxide for opacity, or with an approved lake if you need brightness minerals cannot reach, grind the pigment into castor oil as a paste before it meets the batch, and accept that a mineral red is a different red. The percentages and the grinding method are worked through in the tinted lip balm formula, and the reason the shade reads differently on the lip is optical rather than chemical, as set out under lip colour and gloss physics.

Can you rescue the batch in front of you?

Not the colour. Photodegradation destroys the chromophore, and there is no workshop route back. Remelting a faded batch gives you smooth faded balm, and re-tinting means adding fresh colourant to a base that has already spent weeks accumulating light and oxygen, so the new colour starts in worse conditions than the first lot did.

What you can do is decide what the base is worth. If it smells clean against a frozen retain, the fats are fine and it is a cosmetic defect only, so remelt it as an untinted or minimally tinted product and sell it as that, or keep it for your own use. If it smells at all sharp or waxy, the fade was a symptom rather than the fault, and the batch goes in the bin. Never re-tint stock you intend to sell without restarting the shelf life clock and the stability check, because the date on the label is only as good as the history of the material under it.

Fading and rancidity usually arrive together

They share all three drivers. Light, heat and oxygen degrade a fugitive chromophore and initiate lipid oxidation by mechanisms that are chemically distinct but environmentally identical, so a clear tube on a sunny shelf is running both reactions in parallel. That is why "the colour went and then it smelled odd" is such a common sequence: the colour is simply the earlier indicator, because your eye detects a 20% loss of pigment long before your nose detects the first aldehydes. The underlying chain reaction is set out under rancidity and oxidation.

Working out which came first is worth doing, because it points at different fixes. Compare the faded tube with the foil-wrapped twin from the same batch. If the foil tube is full strength and smells clean, the storage conditions did both, and the answer is packaging. If the foil tube also darkened or smells off, light was not the main event and you have an oxidising base, which points at oil selection, an antioxidant and a shorter date. A carotenoid tint complicates the reading, because carotenoids are themselves radical scavengers and are consumed as they protect the oil, so a bleaching annatto can be evidence of oxidation being fought rather than of light alone.

The decision rule, and what it cannot tell you

If the product has to hold its shade for a stated best-before period in a pack a customer will leave on a window sill, use mineral pigments and stop trying to make plant colour work. If you want a shade minerals cannot reach, use an approved lake, put it in an opaque pack, and shorten the date to something you have actually tested. If you want alkanet specifically, understand that you are choosing a colour with a fade you can measure in weeks and a regulatory status that does not support selling it.

What this page cannot settle is the dose. No controlled study of cosmetic colourant fading in an anhydrous lip base has been published, so the times in the table are working expectations rather than measurements, and they move with your oil blend, your pigment supplier and the light in your window. The rankings are robust because they follow from chromophore chemistry. The numbers are not, and the only ones worth quoting on your own product come from a foil-wrapped twin and a photograph with a grey card in it.

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