Coconut milk stability depends on the emulsifier and gum system, the homogenisation step and the retort cycle. It does not depend on the fat percentage alone. A can that arrives with a hard cream layer on top and watery liquid below is not necessarily spoiled, but a brand owner will still reject it. The dispute usually costs more than the stabiliser would have.
This note explains why coconut milk separates, what each additive class does, and which numbers belong in a purchase specification. The audience is the brand owner or buyer who has to decide what a supplier’s “stabilised” claim actually guarantees.
Why does coconut milk separate at all?
Coconut milk is an oil-in-water emulsion, and its natural protein is too weak to hold the fat in suspension. Tetra Pak’s coconut handbook puts protein at 2.14 to 2.97% of the milk. Only about 30% of that dissolves in the water phase. The handbook states that separation into an oil-rich layer and a water-rich layer cannot be avoided on standing, in processing or in storage (Tetra Pak Coconut Handbook).
So the working question is not “will it separate” but “how fast, how visibly, and can the buyer re-mix it”. Additives slow the process. They do not stop it.
What do emulsifiers do, and what do gums do?
Emulsifiers sit at the oil-water boundary and lower interfacial tension. Gums and other hydrocolloids work differently. They thicken the water phase so that fat globules rise more slowly. Many gums are therefore stabilisers, not true emulsifiers.
| Additive class | Typical examples on labels | Mechanism | Main limit |
|---|---|---|---|
| Emulsifier | Sucrose esters, mono- and diglycerides, sodium caseinate | Coats the fat globule surface | Caseinate adds a milk allergen |
| Hydrocolloid | Guar gum, CMC, xanthan | Raises viscosity of the water phase | Slows creaming, does not prevent it |
| Combination | Emulsifier plus gum | Both effects | Needs trial after retort |
Retail listings show how the combinations look in practice. One UK foodservice product at 17% fat lists guar gum, sodium carboxymethylcellulose and xanthan gum together. One organic brand states it uses under 1% guar gum. Both are label claims, not test data.
The published guidance is consistent on dose. Tetra Pak notes that emulsifiers, sometimes with stabilisers, are typically used below 2% together with homogenisation. Sodium caseinate has also been shown to reduce both the creaming index and free fat in coconut milk emulsions without changing apparent viscosity (AGRIS/FAO record).
Does homogenisation fix it?
Homogenisation shrinks the fat droplets but does not end creaming. A Penn State study found it cut primary droplet size from 10.9 to 3.0 micrometres. It also increased flocculation, which raised viscosity, and all samples still creamed within 24 hours (Penn State).
Order of addition matters. Research summarised on AGRIS found that surfactant added before homogenisation gave submicron, stable droplets. Added afterwards, it displaced protein from the droplet surface and broke up the flocs.
Temperature matters too. Homogenisation works best slightly warm, while the oil is liquid. Cold milk defeats it because the fat has begun to solidify.
A short point worth stating plainly: a supplier that cannot say where in the process the emulsifier goes cannot say whether the product will hold.
What does heat do to the emulsion?
Heat is a second destabiliser. Coconut protein denatures and coagulates at about 80°C, and the Penn State work saw slight coalescence above 80°C. Heated samples also formed a larger cream layer than unheated ones.
Retort sterilisation runs well above that threshold. A formulation proven in the lab before sterilisation therefore proves little. Ask for separation and free-fat data measured after the retort or UHT step, on the finished pack.
Can lower fat avoid the problem?
Partly. A 2024 Food Chemistry study reported that coconut milk at 5% and 10% fat held up well through storage, lipid oxidation and freeze-thaw (PubMed). Most trade product sits higher. Tetra Pak cites 17 to 25% fat for packaged ASEAN coconut milk.
The Codex standard for aqueous coconut products (CODEX STAN 240-2003) sets minimums, summarised by Tetra Pak as follows:
| Product | Minimum fat | Minimum non-fat solids | Total solids |
|---|---|---|---|
| Coconut milk | 10% | 2.7% | 12.7 to 25.3% |
| Coconut cream | 20% | 5.4% | 25.4 to 37.3% |
The standard allows emulsifiers, stabilisers and thickeners where there is a technological need. Food Standards Australia notes that Codex classes these products as fruit and vegetable products, not beverages. Check the importing market’s own additive rules before accepting a formulation. Do not assume the Codex list carries over.
How does transit make it worse?
Transit adds time and temperature swings to a product that is already unstable. A container sitting at port in tropical heat accelerates creaming. A cold lane does the opposite damage by pushing the fat toward crystallisation, which is covered in the desk’s note on reefer versus dry containers for coconut milk.
For buyers weighing formats, powder removes the separation question altogether but brings its own carrier and allergen trade-offs. See the liquid versus powder landed-cost comparison.
What belongs in the purchase specification?
Four lines carry most of the risk. First, fat content as a minimum, with the test method named. Second, the full additive list with INS numbers and maximum levels. Third, a separation or creaming-index limit measured on the finished pack after sterilisation, with the storage period stated. Fourth, an allergen statement covering caseinate and any carrier used.
Add a retained-sample clause as well. Both parties then hold a pack from the same lot to settle any dispute about separation on arrival.
What this means for buyers
- Procurement managers (A): specify the additive list by INS number and the maximum level, not “stabilised”. Require separation data after retort or UHT.
- Importers (B): confirm the destination market accepts every additive on the label before the first container is booked.
- Brand owners (C): decide whether a visible cream layer is acceptable for the end customer, and put a shake-and-re-mix instruction on the pack if it is.
- Allergen-sensitive programmes: check whether the emulsifier is sodium caseinate, which carries a milk declaration.
FAQ
Is a cream layer in coconut milk a defect? Not by itself. Coconut milk separates naturally on standing. It becomes a defect when the cream will not re-mix with shaking, or when free oil is visible, which indicates coalescence.
Which gum is best for coconut milk? No single answer is supported by the sources reviewed. Guar, CMC and xanthan all appear on retail labels. Run a trial on the finished pack after retort and measure separation, not label claims.
Does homogenisation remove the need for stabilisers? Rarely at standard fat levels. Homogenisation cut droplet size in published work, but samples still creamed within 24 hours.
What fat content does Codex require? At least 10% fat for coconut milk and 20% for coconut cream, per CODEX STAN 240-2003 as summarised by Tetra Pak.
Need a spec locked across origins? Send the desk an RFQ.