What happens in the ten minutes after you pour, and why it sets the texture

Supercooling, nucleation rate and crystal growth with numbers: how fast cooling gives small smooth crystals, and how slow cooling gives grain, weeping and dull tops.

Two tins filled from the same jug in the same minute can set into two different products. Stand one on a cold metal tray and the other on a sunny windowsill and you get one smooth, glossy and firm and one grainy, dull and slightly soft, from percentages that are identical on the label. Cooling is not a wait at the end of the process. It is the step that decides the microstructure, and so the texture.

• Bench, 30 g tin: 0.5-1 C/min
• Fridge: 2-4 C/min
• Water bath: 5-10 C/min
• Full melt hold: 75-80 C
• Judge on day 3

Fats crystallise below their melting point, not at it

Water at atmospheric pressure sits at 0 C and turns to ice. Fats do not oblige. A triglyceride or a wax has to be cooled some distance below its melting point before crystals appear at any useful rate, and that distance, the supercooling, is the driving force for the whole process. It is written as the melting point minus the actual temperature: a beeswax phase whose crystals melt out around 63 C and which sits at 55 C is supercooled by 8 C.

A crystal has to start somewhere. A handful of molecules briefly lining up into an ordered cluster gains energy by bonding inside the cluster and loses it by creating a surface against the liquid, and below a critical cluster size the surface term wins and the cluster falls apart again. The height of that barrier drops roughly with the inverse square of the supercooling, so the rate at which viable nuclei appear is nearly nothing for the first few degrees and then climbs very fast.

Growth behaves differently. It needs molecules to diffuse to the crystal face and dock in the right orientation, and diffusion slows as the melt gets colder and more viscous, so growth rate rises with supercooling, peaks, then falls away. Those two curves are what a maker is really controlling:

• Shallow supercooling (a few degrees, slow cooling): very few nuclei form, each one has plenty of time and mobile liquid around it, so each grows large. Few big crystals.
• Deep supercooling (fast cooling to well below the setting range): nuclei appear in enormous numbers at once and the melt is used up before any of them can grow far. Many small crystals.

This is why a melt sits apparently stable at 60 C for ten minutes and then sets solid in ninety seconds once it reaches 50 C. It also explains the pour window: pouring a few degrees below the melting range is safe because nucleation there is slow, which is what the windows in pour temperatures exploit. Leave that jug for twenty minutes at the same temperature, though, and slow is not zero. It will skin over and set, coarsely.

Three ways a crystal starts, and only one of them is rare

Textbooks separate nucleation into types, and the distinction matters at the bench because two of the three are things you control by accident.

Primary homogeneous nucleation is a crystal forming from the melt alone, with nothing to grow on. It needs supercooling of the order of 20 to 30 C and clean conditions, and essentially never happens in a balm.

Primary heterogeneous nucleation is a crystal forming on a foreign surface, which supplies part of the ordering for free and lowers the barrier substantially. This is the normal route in real products, because a balm is full of surfaces: the tin wall, a cold jug lip, dust, and any mineral powder you added, from iron oxides to zinc oxide. That is one reason a pigmented balm and its uncoloured control do not set identically.

Secondary nucleation is existing crystals shedding fragments under shear, each fragment becoming a new nucleus. Gentle stirring through the setting range multiplies crystal number and refines the structure. Vigorous whisking does the same but entrains air that is then trapped as the viscosity climbs, which is the trade-off.

A fourth route is not really nucleation at all: crystals that never melted. Fat has what the trade calls crystal memory. Shea warmed to 60 C rather than melted at 75 to 80 C still holds intact high-melting crystals, invisible in a melt that looks clear, and each is a ready-made template that grows in its own form on cooling. That is the mechanism behind grainy shea butter, and behind hard flecks of wax that never dissolved. Erasing crystal memory is the whole purpose of the full melt hold in melting methods, and no amount of clever cooling recovers from skipping it.

How fast a tin of balm actually cools

Cooling rate is discussed far more often than it is measured, and the measured figures are less flattering than most people assume: a tin left on a wooden board cools at a rate that sits squarely in the grain-forming region.

Three things fall out of that table. Cooling rate is set by surface area divided by mass, so the container decides more than the fridge does, which is why filling lip balm tubes rarely produces grain while jars of the same formula do. The rate is not constant either: heat loss is proportional to the temperature difference with the room, so the first ten degrees go fastest and the last ten crawl. And scale is brutal, because multiplying the batch by fifty cuts the cooling rate by roughly a factor of ten, which is most of the story in formula fails after scale-up.

Why the cooling curve flattens instead of falling smoothly

Log a tin's temperature every thirty seconds and the trace does not fall smoothly. It drops, flattens into a plateau of several minutes somewhere in the setting range, sometimes ticking up by a degree, then resumes falling. Nothing is wrong with the thermometer. That plateau is the latent heat of fusion released as the crystals form.

The arithmetic makes the size of the effect obvious. Triglyceride and wax crystals give up roughly 120 to 200 joules per gram on crystallising, depending on the fat and the crystal form, which is what a differential scanning calorimeter integrates when it measures a fat. Liquid vegetable oil has a specific heat capacity of about 2 joules per gram per degree. So for a 30 g tin that ends up 25 percent solid:

• Crystallising 7.5 g of fat releases about 7.5 x 165 = 1,240 J.
• Cooling the whole 30 g by one degree requires removing about 30 x 2 = 60 J.
• The crystallisation heat alone is therefore worth about 20 degrees of cooling, on top of the sensible heat you were already removing.

That heat leaves through the same small surface, so the balm stalls at its setting temperature until it is gone. Two things follow. The real cooling rate through the range where crystal size is decided is much slower than the averages in the table above, so what separates the methods is how fast they carry away the plateau. And in a large batch the plateau lasts hours while the centre holds itself warm, which is why bulk cooling grains when the same formula in tins does not.

Crystal size, and the four things it changes

Cooling rate matters because crystal size matters, and crystal size changes four separate properties that makers usually treat as unrelated complaints.

Smoothness. A crystal has to reach a certain size before skin registers it as a particle rather than as texture. The best-characterised threshold is the chocolate trade's: particles above roughly 25 to 30 micrometres are reliably detected on the palate, which is why chocolate is refined below that. No equivalent figure has been established for balm on skin, but it is the right order of magnitude. Slowly crystallised shea forms spherulites tens to a few hundred micrometres across, comfortably over the line; the same shea cooled fast gives crystals of one to five micrometres that nobody can feel.

Gloss. A surface reflects specularly only if its roughness is well below the wavelength of visible light, which is 0.4 to 0.7 micrometres. Crystals reaching five or ten micrometres at the top surface scatter incoming light in all directions and the top reads as matte or frosted long before anyone can feel grain. Dullness is therefore an early warning of coarse crystallisation, diagnosed in matte instead of glossy balm.

Oil binding. Liquid oil is held by capillary forces in the pores between crystals rather than by any chemical affinity, as set out in oleogels and wax networks. Capillary pressure rises as pore size falls, and halving the crystal size roughly doubles the surface area per gram. A slowly cooled balm has coarse pores that cannot hold oil against gravity and a warm shelf, which is the mechanism behind balm sweating.

Hardness. Two balms of identical composition, and so identical solid fat content at 20 C, can still differ in firmness, because the stiffness of a fat crystal network depends on how the crystals are connected and not only on how many there are. A dense network of small crystals with many contact points is stiffer than a sparse one of large crystals. If you are chasing hardness with a penetrometer, cool every sample identically or you are measuring your fridge.

What cooling rate cannot change

Cooling is often oversold as a fix, so be precise about the limit. Cooling rate sets microstructure: crystal number, crystal size, which polymorph forms first and how the network is connected. It does not set composition. How much of a balm is solid at 20 C or 34 C is a property of the fats you chose, which is what solid fat content curves describe. Cool a 35 percent shea formula as fast as you like and it still contains 35 percent shea, with the grain risk that carries after a month on a shelf.

Nor is the microstructure permanent. Fine crystals are a kinetically trapped state, not an equilibrium one, and a balm that spends a week cycling between 5 C at night and 40 C in a delivery van coarsens without ever fully melting, which is the whole of balm changed after shipping. Fast cooling buys you a good starting structure and time. It does not repeal thermodynamics.

The first 72 hours: why a batch is judged on day three

Setting is not finishing. A balm firm enough to handle, typically 10 to 30 minutes after pouring, is nowhere near its final solid fat content. Three processes carry on:

1. Continued crystallisation. Fat that is still liquid but thermodynamically ought to be solid keeps coming out of solution for hours, slowly, because the network restricts diffusion.
2. Polymorphic transition. The forms that crystallise first are the least stable ones, and they convert to more stable, higher-melting, denser forms over hours to weeks. The mechanism is in fat crystal polymorphism, and it is why the same balm can be both firmer and grainier at a month than at a day.
3. Sintering. Crystals in contact fuse at their contact points over time, converting a loose network into a rigid one without any change in solid fat content at all.

The result is a firmness that climbs after the pour, commonly in the region of 10 to 30 percent over the first 24 to 72 hours for a beeswax and butter balm, with most of it in the first day and a slower creep for weeks after. That band is a working expectation rather than a measured constant, since it depends on the wax and the storage temperature; measure your own with a penetrometer at day 1, day 3 and day 30 if the number matters. The long tail is covered in balm hardens over time.

So do not reformulate on day zero. A lip balm judged an hour after pouring feels soft, you add wax, and three days later you have a stick that drags. Wait 72 hours and compare against a control from the same session. Record the cooling method next to the percentages, because it moves the product as much as a point of wax does, and a batch record that omits it cannot explain an inconsistent batch later.

What chocolate tempering does, and does not, transfer

Anyone reading about fat crystallisation meets chocolate tempering quickly, so it is worth saying plainly which parts apply. Tempering dark chocolate means melting fully at 45 to 50 C to erase crystal memory, cooling to 27 to 28 C so a mixture of forms nucleates, then reheating to 31 to 32 C so the lower-melting forms melt out and only the wanted form V is left as seed. It works because cocoa butter is effectively the only fat present, its form V melts at about 34 C, and the gap between the forms is wide enough to hit.

In a balm the cocoa butter is diluted. At 15 or 20 percent of a formula whose remainder is liquid oil, that oil acts as a solvent and depresses the temperature at which cocoa butter crystallises out, so a 31 to 32 C hold that leaves form V standing in chocolate leaves nothing standing at all. There is no published tempering protocol for oil-diluted cosmetic systems, and the chocolatier's trick of seeding with pre-crystallised form V powder has no established equivalent here. Treat any recipe that tells you to temper a balm with scepticism.

What does transfer is the first step and the last. Melting fully, above the highest melting crystal present, is as necessary in a balm as in chocolate. Setting fast, rather than drifting down through the temperature range where large crystals grow, is equally necessary. Everything cocoa butter specific belongs on the cocoa butter page instead.

A cooling protocol that works

This is the sequence for tins and jars in a home or small commercial workshop. It costs nothing but attention.

1. Melt fully and hold. Any formula containing shea, cocoa, mango or another hard butter goes to 75 to 80 C (167 to 176 F) and stays there for 10 to 15 minutes after the last visible solid disappears. Clear is not the same as melted.
2. Cool the melt to the pour window before filling. Roughly 65 to 72 C for lip tubes and 60 to 68 C for tins, with the reasoning and the exceptions in pour temperatures. Pouring straight from the melt hold guarantees shrinkage faults.
3. Pour onto a conductive surface. Stand the containers on a metal tray or a stone worktop, not a towel or a wooden board. It is free cooling rate, and it keeps every container in the batch on the same schedule.
4. Get them cold within five minutes. Move the tray to a 4 to 5 C fridge, uncovered, for 20 minutes. That is enough to take a 30 g tin well below its setting range. Longer does no harm but achieves nothing.
5. Return to room temperature before lidding. A cold tin lidded in a warm room condenses water on the balm surface, and water in an anhydrous product is the one thing that lets mould grow, as mould in balm explains.
6. Leave 24 hours before handling and 72 hours before judging. Label the trial batch with the cooling method used, not just the formula.

When cooling is not the answer

The decision rule is short. If the complaint is grain, dullness, sweating or a top that will not level, change the cooling before you touch a single percentage: process is free, and it is where the worst faults live. Make one control batch with a full 75 to 80 C melt hold and 20 minutes in the fridge, wait three days and look again. If the fault has gone, it was never a formula problem. If it survives, it is compositional, and the levers are in balm texture science.

Two limits are worth carrying away. Cooling rate barely matters in a balm of wax and liquid oil with no hard butter, because nothing in it grows crystals anyone can feel, so a simple beeswax and sunflower lip balm does not need the fridge. And fast cooling is a starting condition rather than a warranty: it gives you the finest structure the formula can support on the day, after which storage temperature decides how long that structure lasts.

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