Why the same fat can set three different ways, and what each one feels like

Alpha, beta prime and beta polymorphs explained: packing, melting points and transition rates, and why cocoa butter blooms while shea turns gritty in the same tub.

A single fat can set into several different solids, each with its own melting point, density, hardness and feel, without one atom changing place inside the molecules. That is polymorphism, and it is the reason a cocoa butter balm goes chalky white on a shelf and a shea balm turns gritty in a tub that has never been opened. Both are the same process running at different speeds.

• Three families: alpha, beta prime, beta
• Cocoa butter: six forms, 17 to 36 C
• Transitions are one way
• Bloom is form VI, not mould

Same molecules, a different stack

A triglyceride is three fatty acid chains hung off a glycerol backbone, as set out in lipid chemistry for formulators. When such a molecule crystallises it folds into a rough tuning-fork shape and stacks with its neighbours in layers, chains lying alongside chains. There is more than one way to do that stacking, in the same way that a pile of identical logs can be squared off or laid loosely, and each arrangement is a different crystal with different properties.

Two things distinguish the arrangements. The first is the subcell, the geometry of how the chains pack side by side. The second is the chain length structure, whether the layers repeat every two chain lengths (2L) or every three (3L), which depends on how well the fatty acids in a given triglyceride tolerate sitting next to each other. Mixed triglycerides with one very different chain, such as the cocoa butter molecule with palmitic, oleic and stearic acids on one glycerol, often segregate the oleic chains into their own layer and adopt a 3L structure.

Nothing chemical changes between the forms. Melt the fat and the difference disappears entirely, which is why the first instruction in every fix on this page is to melt properly. Polymorphism is not degradation, it is not rancidity, and it does not make a product unsafe: it changes hardness, melting point, feel and appearance only.

Alpha, beta prime and beta: what the names mean

The three-family classification comes from Larsson's work on glyceride crystal forms, and the members are told apart by X-ray diffraction. Powder a sample, put it in the beam, and the angles at which it scatters give the repeat distances in the lattice. The short spacings, between about 3.5 and 5 angstroms, measure the distance between neighbouring chains and identify the subcell directly. The alpha form gives a single line near 4.15 angstroms, beta prime gives two near 4.2 and 3.8, and beta gives a strong line near 4.6. Those three fingerprints are the definition; everything else is consequence.

Read the melting point column across and the practical point appears: the gap between the lowest and highest form of one pure fat is around 20 C. Two batches of the same balm, identical on the ingredient list, can sit either side of a difference that large in the melting behaviour of their solid phase. That is a bigger swing than most reformulations produce, and it is invisible until you feel it.

Beta prime is the form makers want. Its crystals are small needles that pack into a dense network with a great deal of surface area, which is what holds liquid oil in place and what feels smooth. Beta crystals are large flat plates: they have less surface per gram, so they release oil, and they are over the size at which skin registers a particle.

The transitions run one way only

Fat polymorphism is monotropic. Alpha converts to beta prime, and beta prime to beta, and neither step reverses. There is no cooling regime that turns beta back into beta prime. The only way back is to melt the fat above the melting point of the highest form present and start again.

Two consequences follow, and they pull in opposite directions. The first is that the least stable form usually appears first. Nucleation barriers are lowest for the loosest packing, so a fast-cooled fat crystallises mostly as alpha even though beta is the thermodynamically preferred state, a pattern general enough in crystallisation to have a name, Ostwald's rule of stages. The second is that the fat then converts towards beta at its own pace, in the solid, without melting.

That pace depends on temperature and on composition. Warmth gives molecules the mobility to rearrange, so a balm stored at 30 C coarsens far faster than one at 15 C, and temperature cycling is worse than either steady temperature because each warm phase partly melts the smallest crystals and each cool phase deposits that material onto the survivors. Crystals grow at the expense of their neighbours, which is why a parcel that spent a week in a hot van arrives grainy without ever having melted, the subject of balm changed after shipping. The same slow conversion is part of why firmness keeps climbing in balm hardens over time.

Timescales are fat-specific and worth treating as orders of magnitude rather than figures. Alpha in a typical vegetable fat survives minutes to hours at room temperature. Beta prime in a palm-based or tallow-based fat can persist for years. Beta prime in shea converts within days to weeks at 20 C, which is exactly the interval over which a maker's smooth-on-day-three balm becomes a customer's gritty-in-a-month complaint.

Cocoa butter's six forms

Cocoa butter has the best-characterised polymorphism of any natural fat, because the chocolate industry needed it. Wille and Lutton identified six forms, numbered I to VI in order of increasing stability and melting point.

Form V is why chocolate snaps and melts on the tongue, and form VI is why old chocolate looks grey and feels dull. In a balm, cocoa butter is diluted in liquid oil rather than standing alone, so the tidy six-step ladder is blurred and the tempering routine that produces form V in chocolate does not transfer. The consequences for hardness, brittleness and usage rate belong on the cocoa butter page, and the snapping and crumbling that follow from form VI are diagnosed in cracked and crumbly balm.

Which fats stay smooth, and which coarsen

Whether a fat settles happily in beta prime or drives on to beta is decided by how uniform its triglycerides are. A fat whose molecules carry a mixture of chain lengths cannot pack into the neat parallel arrangement that beta requires, so it stalls in beta prime and stays fine-grained for years. A fat whose molecules are uniform, particularly one dominated by symmetrical triglycerides with the same saturated acid at both ends, packs into beta readily.

The kokum row is where the common summary breaks down, and it is worth being exact about. Kokum is often listed as a beta prime butter because it is famously smooth, but its composition, dominated by a single symmetrical triglyceride with stearic acid at both ends of the glycerol, is textbook beta-forming. It stays smooth for a different reason: a fat that is nearly one molecule has nothing to separate into. That distinction matters because it separates two mechanisms that get conflated. Coarsening by polymorphic transition happens within one crystal population. Coarsening by segregation happens when one fraction of a mixed fat crystallises apart from the rest. Kokum is protected from the second and slow at the first, which is why it substitutes so well for shea where grain is the problem, as mango and kokum butter sets out. Comparative melt points and hardness for the whole set are in butters compared.

Shea grain is segregation, not separated stearic acid

The folk explanation for grainy shea is that the stearic acid separates out. It does not, and free fatty acid is present at well under 1 percent in a decent grade of shea.

What actually happens is that shea is two fats in one jar. Its triglycerides span a wide range, from almost fully unsaturated ones that are liquid at room temperature to a high-melting fraction dominated by the symmetrical triglyceride with stearic, oleic and stearic acids on the glycerol, written StOSt. That fraction has the highest melting point in the mixture, so on cooling it crystallises first and by itself. If cooling is slow, those crystals have time to find each other and to grow into StOSt-rich domains rather than staying dispersed. Then the monotropic clock runs: the domains convert towards beta and grow into spherulites tens to hundreds of micrometres across, which is comfortably over the size skin detects. The result is a balm that was smooth when it was made and sandy a month later.

Two secondary factors modify the timing. Shea's 5 to 11 percent unsaturated hydrocarbons and other unsaponifiables sit between the crystals and slow the transition, which is one reason refined and unrefined shea from different origins do not grain on the same schedule. And carrier oil matters: a shea butter diluted to 25 percent in liquid oil has more room for its crystals to move and rearrange than a neat shea does. Origin, grade and usage rate are covered in shea butter, and the full timeline and rescue in grainy shea butter.

Bloom is form VI, and it is not mould

Fat bloom is the white or grey film that appears on cocoa butter, on chocolate, and on balms containing either. The mechanism is the same monotropic transition seen from the outside: liquid fat migrates to the surface, recrystallises there in the most stable form available, and grows crystals large enough to scatter light instead of reflecting it. Specular reflection needs surface roughness well below the wavelength of visible light, 0.4 to 0.7 micrometres, so surface crystals of even a few micrometres turn a gloss into a haze. The same optics produce the dull tops in matte instead of glossy balm, and it is worth saying plainly that bloom is a cosmetic defect only: the product is unchanged chemically and is not spoiled.

Because it presents as white patches on a surface, bloom is regularly mistaken for mould, and the mistake goes both ways. Anhydrous balms are hostile to microbes, and the distinction is easy to make at the bench.

The four controls that actually work

You cannot stop a monotropic transition. You can control which form appears first, how large the crystals are when they appear, and how quickly the conversion runs afterwards. Four levers do that, in descending order of return.

1. Melt completely. Hold any formula containing shea, cocoa, mango or another hard butter at 75 to 80 C (167 to 176 F) for 10 to 15 minutes after the last visible solid goes. Beta crystals of the high-melting fraction survive well above the temperature at which the butter looks liquid, and each survivor seeds its own form. This single step fixes more grain complaints than every formula change combined.
2. Cool fast through the setting range. Fast cooling produces many small crystals and gives the segregating fraction no time to gather into domains. Twenty minutes at 4 to 5 C is enough for a 30 g tin. The rates and the mechanism are in crystallisation and cooling rate.
3. Limit the beta formers. Grain risk climbs steeply with the proportion of shea or cocoa butter in the formula. Above roughly 25 percent shea, or 15 to 20 percent cocoa butter, process control stops being enough on its own. Replace part of the load with a beta prime former or with kokum.
4. Control storage. Steady and cool beats warm, and steady beats cycling at any temperature. Below 25 C with no repeated warm and cold swings is the target, which is also the advice in storing balms at home.

Where the bench runs out

Everything above is inferred at the bench from symptoms. Polymorph identification is a laboratory measurement: X-ray diffraction for the form itself, differential scanning calorimetry for the melting behaviour that follows from it, and the reasons those two methods disagree with a capillary tube are in melting point methods. Nobody can tell alpha from beta prime by touch, and any claim that a home process reliably produces a named polymorph in a mixed, oil-diluted balm should be treated as unsupported.

What you can do without a laboratory is bound the problem. If a fault appears immediately, it is not polymorphism, because transitions take time; look at undissolved solids or at cooling faults instead. If it appears over weeks and is worse in warm storage, polymorphism is the likely mechanism, and the response is process first and composition second. If a full melt hold and a fast cool produce a batch that is still smooth at eight weeks alongside an untreated control that is not, you have your answer without owning a diffractometer. And if the treated batch grains anyway, the formula carries more beta-forming butter than it can support, and no amount of process will rescue it. The trade-off you are making is honest: how much of the richness that shea and cocoa butter bring is worth the grain risk they carry, which is a formulation decision rather than a physics one, and it is settled by the levers in balm texture science.

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