Layered and banded set balm, where the melt stratified before the wax could hold it
Slow cooling lets wax crystallise first and dense material settle before the network forms. How to read the cut face, when to remelt, and the cooling rate.
A balm that has set in bands is not one fault but a family of them, and a knife separates them in about a minute. The common mechanism is that the melt stayed fluid long enough for its components to sort themselves by density, and the crystal network then froze that arrangement in place. Which component moved decides whether you remelt, reformulate, or accept that the batch was never going to work.
Layers form in the window between wax crystallising near 60 C and the butters crystallising at 35 to 25 C, when the melt is fluid and has no yield stress but already carries dense solids. Solid beeswax is about 0.96 g per cm3 against 0.91 to 0.93 for the oils, so it sinks. Remelt to 75 C, stir without pause until the melt visibly thickens, then pour.
What a layered balm actually is
Nothing in an ordinary anhydrous balm is immiscible. Wax, hard butter and carrier oil dissolve in one another in the melt and stay dissolved for as long as they are all liquid, which is why the jug looks uniform at 75 C. Separation here is not a liquid splitting into two liquids. It is solids appearing in a liquid and then travelling through it.
Those solids do not appear all at once. Every structurant has its own crystallisation temperature, so a blend of carnauba, beeswax and shea does not set once, it sets three times. Between the first crystallisation event and the last there is a period in which the contents of the tin are a suspension: dense crystals in a thin, warm oil with essentially no resistance to flow. A suspension with no yield stress does what physics says it will do, which is settle. Once the solid fraction grows high enough to percolate into a continuous network, everything stops moving and whatever arrangement had been reached becomes permanent. The reason a set balm holds a dense particle indefinitely while a warm melt cannot is yield stress, and the network that supplies it is a wax oleogel.
Layering is therefore a race between settling and setting. Everything below is about shortening the time the suspension exists, or raising the resistance it meets while it does.
Read the cut face before you do anything
Take one tin from the middle of the batch, warm a thin blade under a tap, and cut straight down through the puck to the base. Lift one half out. The vertical section carries almost the whole diagnosis, and five patterns cover nearly every case.
- A sharp horizontal line. A discrete phase reached a level and stopped. The boundary is flat, it sits at the same height all the way round the tin, and the material either side differs in colour, hardness or both. This is stratification proper.
- A gradient with no boundary you could point at. Hardness or colour changes continuously from base to surface. Nothing settled as a distinct phase; the melt simply cooled slowly enough for the crystal population to differ top to bottom. The fix for this one is entirely thermal.
- A hard crust two to three millimetres deep on the top only. The surface lost heat fastest and crystallised first. That is a skin rather than a layer, and it usually comes with a dip or a hollow underneath, which is tunnelling rather than separation.
- Beads or a slick on an otherwise uniform cut face. Not layering. Oil has migrated out of a network that was already set, which is sweating and a different fault with a different fix.
- A thin, darker, sticky film at the very bottom. A minor component that was never really in solution: an emulsifier, honey, glycerin, or a powder that was never wetted.
Cut two tins, one from the first container you filled and one from the last. If only the last is banded, the melt was stratifying in the jug while you poured, and the fix is stirring during filling rather than anything in the formula. This is the most common finding once a batch passes a litre or so, and it is one of the standard traps in batch scaling.
Layering, sweating and a failed emulsifier look alike
Three faults produce a two-tone tin and only one of them is fixed by stirring. Settle which you have before you reprocess anything, because remelting the wrong one wastes a batch twice.
| Fault | What the cut face shows | Test that settles it | Does remelting fix it |
|---|---|---|---|
| Melt-phase stratification | Sharp flat line, or a continuous gradient, present the moment the balm set | Scrape a little from each side and warm both on a spoon: they melt at visibly different temperatures | Yes, with agitation on the way down |
| Post-set sweating | Uniform body with free oil on the surface or under the lid, appearing days later | Wipe clean, hold at 30 C for 24 hours, see whether it returns | Only briefly. The oil load exceeds what the network can hold, so it needs formulating out |
| Cleansing balm emulsifier drop-out | A clear, tacky, amber film at the base, often only 1 to 2 mm deep | Rub the base film with a wet fingertip: it turns milky white while the body above stays clear | No, not on its own. The level or the mixing temperature has to change |
The third case is worth a moment because it is under-diagnosed. Polysorbate 80 has a density around 1.06 to 1.09 g per cm3 and only limited compatibility with a fatty base at the 8 to 15% levels a rinse-off cleansing balm needs. Stirred in too cool, or stirred in for ten seconds instead of a full minute, it never disperses and simply goes to the bottom. The visible result is a layered tin; the functional result is a balm whose top two thirds will not rinse.
The densities that drive it
The direction of travel is not a matter of opinion. Once you know the density of the solid and the density of the surrounding melt, you know which way it goes.
| Material | Density (g/cm3) | Behaviour in a warm melt |
|---|---|---|
| Carrier oil, liquid at 60 C | about 0.88 | This is the fluid everything else moves through |
| Carrier oil, liquid at 20 C | 0.91-0.93 | Reference band across the named vegetable oils |
| Beeswax, solid | 0.95-0.97 | Sinks slowly as crystals |
| Shea stearin crystals | about 0.95 | Sinks slowly, and forms very late |
| Candelilla wax, solid | about 0.98 | Sinks |
| Carnauba wax, solid | about 1.00 | Sinks, and crystallises first of all |
| Polysorbate 80 | 1.06-1.09 | Sinks as droplets if not fully dispersed |
| Glycerin | 1.26 | Not soluble in the base. Sinks as droplets |
| Honey | about 1.4 | Sinks immediately unless emulsified |
| Iron oxide pigment | about 5.0 | Sinks fast |
| Zinc oxide | about 5.6 | Sinks fastest of anything in a normal formula |
Two conclusions follow. First, everything sinks. The only thing in a balm with a density below the oil is air, which is why air bubbles rise and nothing else does, and why a hard top and a soft base is almost never a settling fault. Second, the gap between wax and oil is small. Fifty-five kilograms per cubic metre is a weak driving force, and it only produces a visible band when the melt sits fluid for a long time. The powders are a different problem in scale: zinc oxide has roughly eighty times that density difference, and a poorly dispersed pigment gives a hard grey cake at the base rather than a soft band, which is covered in pigment sinking to the bottom. Sugars and polyols are a third case again: honey, glycerin and the other humectants are not soluble in a fatty base at all, so a layer of them is not settling but simple phase separation.
The gap between the waxes and the butters
The reason a slow cool is so much worse than a fast one is that the two structurant families crystallise a long way apart in temperature. Nothing bridges the middle.
| Component | Melts at (C) | Clouds on cooling near (C) | What it means for layering |
|---|---|---|---|
| Carnauba wax | 82-86 | 78 | First solid in the pot, with the longest fall ahead of it |
| Sunflower wax | 74-80 | 70 | Efficient gellant, so the network arrives early |
| Candelilla wax | 68-73 | 64 | Similar, denser crystals |
| Stearic acid | 67-72 | 65 | Crystallises above beeswax and closes the window early |
| Beeswax | 61-65 | 58 | The usual first event in a plain formula |
| Cocoa butter | 33-36 | 26-28 | Undercools heavily, so it sets very late |
| Shea stearin | 33-36 | 30 | Adds nothing to structure until the melt is nearly cold |
| Coconut oil | 24-26 | 22 | Effectively a liquid throughout the risky period |
Read the middle of that table. In an ordinary beeswax and shea balm, the wax has finished crystallising by about 55 C and the shea has not started until about 30 C. Between those two points nothing new appears, the oil is thin, and the crystals that already exist are free to move. That is the settling window, and it is where the layers are made. A formula built from carnauba and shea alone has a window of over forty degrees, which is why hard vegan formulas band more readily than beeswax ones at the same total structurant. Choosing waxes whose crystallisation temperatures are close together is a real design decision, and the ranges are tabulated in waxes compared.
How fast things sink, with the arithmetic
Stokes' law gives the terminal velocity of a sphere falling through a viscous fluid: velocity equals two thirds of the density difference times gravity times the radius squared, divided by nine times the viscosity. The important part is the radius squared. Doubling the size of a settling unit quadruples its speed, which is why the size of the crystals you produce matters far more than anything about the formula.
| Settling unit | Velocity | Time to fall 15 mm |
|---|---|---|
| 10 um wax crystal, from a stirred cool | 1.1 mm/h | About 14 hours |
| 20 um crystal | 4.3 mm/h | About 3.5 hours |
| 50 um aggregate, from a still cool | 27 mm/h | About 33 minutes |
| 100 um aggregate | 108 mm/h | About 8 minutes |
| 5 um zinc oxide agglomerate | 23 mm/h | About 39 minutes |
A tin that takes twenty minutes to set is safe from ten micrometre crystals and defenceless against hundred micrometre ones. Agitation is what decides which you get: stirring multiplies nucleation sites, so the same mass of wax comes out as a large number of small crystals rather than a small number of large ones, and the small ones barely move in the time available. That is the same lever that controls smoothness, described in crystallisation and cooling rate. Settling is also self-limiting, because viscosity climbs steeply as the solid fraction builds, which is why a banded balm is usually part-separated rather than cleanly split in two.
Fix this batch
For plain wax, butter and oil formulas, layering is completely reversible. Nothing has oxidised, nothing has degraded, and a rescued batch is a normal batch.
- Reprocess the whole lot together. Scrape every tin back into one vessel, including the scrapings from the walls. Rescuing pot by pot guarantees that no two tins match, which is its own fault in batch to batch consistency.
- Melt to 75 C in a water bath and hold ten minutes. That clears every wax in a normal formula and comfortably clears the 41 to 43 C point at which surviving shea seed crystals would otherwise template grain. Do not go above 85 C, which darkens beeswax and drives off volatiles.
- Fix the dispersion now if a powder or an emulsifier was the culprit. Grinding a pigment into a little oil, or stirring an emulsifier for a full minute at 65 to 70 C, has to happen in the melt. Cooling faster will not rescue a material that was never dispersed.
- Stand the vessel in a shallow cold water bath and stir without stopping. Scrape the base and the walls, because that is where the melt is coldest and where the first crystals form. Do not stop to label, answer the door, or line up tins. Everything else should already be ready.
- Pour the moment the melt thickens and coats the spatula. In a beeswax formula that point falls around 60 to 63 C, at the upper end of the tin window in pour temperatures. Waiting for it to thicken further gives you drag marks and a rough surface instead.
- Cool the filled containers fast, uncovered and spaced apart. Get from 40 C down to 25 C in under thirty minutes. A lid traps heat over exactly the part of the puck you are trying to freeze quickly.
- Cut one tin at 24 hours and another at seven days. A same-day check will not catch a slow segregation that only becomes visible once the late-crystallising butter has finished.
Remelting is not always the right call. Every reheat costs volatile top notes and reheats material you have already heated once, so a heavily fragranced or actives-bearing balm degrades a little each time. If the balm smells sharp, waxy or like old crayons, stop: remelting a rancid batch gives you smooth rancid balm. And if a dose-regulated active has settled, the top of the tin is under-dosed and the bottom is over-dosed, which is a labelling problem before it is a texture one.
Prevent it: the cooling schedule
- Melt and hold: 75 C, ten minutes. Ceiling 85 C.
- Stir continuously from 70 C to the pour, scraping base and walls.
- Pour above 60 C, so the melt arrives with few crystals already in it.
- Cross 40 C to 25 C in under thirty minutes.
- Uncovered, spaced and unstacked until cold to the touch.
- Workshop above 24 C: use a cool tray or a fan, not the fridge.
Pouring hot and cooling fast sounds contradictory, and it is worth being explicit about why it is not. Pouring above 60 C means the melt reaches the container before beeswax has crystallised, so nothing is in suspension when it arrives and nothing has had time to sort itself in the jug. What happens next is a separate decision: heat has to leave the container quickly, which is why the tins want space, air and a cool surface. Pouring hot into a tray of tins that then cool slowly on a warm bench is the worst of both, and it is the usual recipe for a gradient rather than a sharp line. Lids belong on cold balm only, and stacked warm tins insulate one another for hours.
Formula levers that shut the window
Process fixes the batch in front of you. If the same fault keeps returning, the formula is leaving the window open too long, and there are four levers in rough order of effect.
Raising the wax by one to two points is the blunt one. It brings the solid fraction to the percolation threshold earlier and at a higher temperature, so the melt develops a yield stress before there is time to settle. It also hardens the balm, so check it against the texture you wanted using the trade-offs in balm texture science.
Adding 0.5% stearic acid is the precise one. It crystallises around 65 C, above beeswax, so it seeds a weak network several degrees before the wax arrives and holds the wax crystals where they form. Half a point is enough to change the behaviour and small enough not to make the balm feel soapy or draggy.
One to three percent castor wax, or two to five percent ozokerite, does the same job harder. Both gel well above the beeswax crystallisation temperature, and both bind oil better than beeswax does, so they address sweating at the same time. For a pigmented balm, 0.5 to 2% fumed silica builds a yield stress at a temperature where nothing else is doing anything.
The fourth lever is geometry. Depth is the distance a crystal has to fall, so a 15 mm tin bands far less readily than a 40 mm jar of the same formula, and a lip tube barely bands at all. If a formula only misbehaves in the deep container, do not reformulate; change the container or the pour.
When to stop trying to save it
The decision rule is short. If the cut face shows a sharp line or a gradient in a plain wax, butter and oil balm, remelt it, stir it down, and it will come back right, because nothing has changed chemically. If the line is an emulsifier, a humectant or a sugar, remelting alone will not fix it, because the material is not soluble in that base at that level, and the answer is a lower level, a different material, or a solubiliser. If it is a settled powder that was never wetted, remelting works only when you redo the dispersion properly first.
Two honest limits. The settling figures on this page are calculated from Stokes' law with stated assumptions, not measured on balm melts: nobody publishes settling data for craft-scale anhydrous batches, and a real crystal is a plate rather than a sphere. And the crystallisation onsets are workshop observations of the point at which a stirred melt clouds, several degrees below the published melting ranges because waxes and fats undercool by an amount that depends on the cooling rate itself. Use them to rank your options and to decide where the risky window sits in your own formula, then confirm on a 30 g trial batch rather than on the production run.
Frequently asked questions
Why is the bottom of my balm harder than the top?
Because solid wax is denser than liquid oil, roughly 0.96 against 0.91 to 0.93 g per cm3, so wax crystals sink through a melt that stays fluid. If the balm cooled slowly the crystals had time to fall before the network formed, leaving a wax-rich base and an oil-rich top. Remelt to 75 C, stir continuously as it cools, and pour above 60 C.
Can I just remelt a balm that has separated into layers?
For a plain wax, butter and oil balm, yes, and the result is a completely normal batch, because nothing has degraded. Melt the whole lot together to 75 C, hold ten minutes, then stir without stopping as it cools and pour the moment it thickens. Remelting will not fix a separated emulsifier, an undissolved humectant, or an unwetted powder.
Is a layered balm safe to use?
Layering is a physical fault, not spoilage, so an ordinary layered balm is safe if it smells normal. Two exceptions matter. A balm carrying a dose-regulated active is unevenly dosed once that active has settled, so the base of the tin carries more than the label says. And a layer of honey or glycerin at the base introduces water into an anhydrous product, which is a microbiological question rather than a cosmetic one.
Why does my cleansing balm have a sticky layer at the bottom?
Almost always polysorbate 80 or another oil-soluble emulsifier that never dispersed. It is denser than the fat phase at around 1.06 to 1.09 g per cm3, and if it goes in below about 65 C or is stirred for only a few seconds it sinks as droplets. The test is to rub the base film with a wet fingertip: the emulsifier layer turns milky white while the balm above stays clear.
Does putting the tins in the fridge stop layering?
It shortens the settling window, so it does help, but it is a crude tool. Fridge cooling chills the outside of the tin first and can produce a hard rim with a softer, still-settling middle, and a cold container in a warm room collects condensation, which is unwanted in an anhydrous product. Stirring through the cooling range and pouring into spaced, uncovered tins on a cool tray is the better lever.
Why did only the last few tins of a batch come out layered?
Because the jug was stratifying while you filled. The melt cools from the moment it leaves the heat, so the last pour is several degrees cooler than the first and has been standing longest. Stir the jug between every few containers, work faster, or split a large melt into two vessels. This is one of the standard failure modes when a batch size increases.
Is stirring while it cools really necessary, or is pouring cooler enough?
They do different jobs. Pouring cooler shortens the time the balm spends fluid, but it also means crystals have already formed in the jug and it risks drag marks and a rough surface. Stirring changes the crystals themselves, producing many small ones instead of a few large ones, and since settling velocity scales with the square of the radius that is the larger effect by far.
Sources and further reading
- FAO and WHO Codex Alimentarius Commission, CODEX STAN 210-1999, Standard for Named Vegetable Oils, relative density and refractive index tables.
- European Directorate for the Quality of Medicines, European Pharmacopoeia monographs: Cera alba and Cera flava, Strasbourg, for beeswax identity and physical constants.
- National Library of Medicine, PubChem compound summary: zinc oxide, NCBI, for density and particle data.
- Rhodes, M., Introduction to Particle Technology, 2nd edition, Wiley, 2008, chapter on single particles in a fluid, for Stokes' law and its validity limits.
- Timms, R. E., Phase behaviour of fats and their mixtures, Progress in Lipid Research, 23(1):1-38, 1984, for sequential crystallisation and undercooling in mixed fat systems.
- Blake, A. I. and Marangoni, A. G., Structure and physical properties of plant wax crystal networks and their relationship to oil binding capacity, Journal of the American Oil Chemists' Society, 91(6):885-903, 2014.
- Doan, C. D. and colleagues, Evaluating the oil-gelling properties of natural waxes in rice bran oil, Journal of the American Oil Chemists' Society, 92(6):801-811, 2015, for wax gelation onset temperatures.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.