Sodium Citrate in Gelato: What It Does and How to Dose
Table of contents
Sodium citrate is not a stabilizer, not a sweetener and not an emulsifier in the everyday sense. It is a calcium sequestrant and a buffer, and in a gelato lab it solves exactly two problems: milk protein that curdles when acid arrives, and dairy that turns grainy when it should turn silky.
Trisodium citrate dihydrate: a mildly alkaline salt that does its work on calcium, not on water.
What Sodium Citrate Actually Is
The powder sold as sodium citrate in food work is almost always trisodium citrate dihydrate, Na₃C₆H₅O₇·2H₂O, molar mass 294.10 g/mol. In the United States it is affirmed GRAS under 21 CFR 184.1751, which describes it as the sodium salt of citric acid, anhydrous or with two moles of water, and permits it in food with no limitation other than current good manufacturing practice. In the European Union it carries the number E 331(iii) and is authorised as a food additive under Regulation (EC) No 1333/2008.
It is not an exotic import into a dairy mix. Bovine milk already contains citrate naturally, on the order of 1.5 to 2 g per litre, in equilibrium with the calcium in the casein system. Adding sodium citrate pushes an equilibrium that is already there. The salt is highly soluble and mildly alkaline — a 5% solution lands near pH 8 — but the alkalinity matters less than the anion, which is where the real work happens.
The Calcium Trick That Makes It Work
Casein does not float in milk as loose protein. It is packed into micelles held together in large part by colloidal calcium phosphate, which acts as the mortar between casein sub-units. Remove the mortar and the micelle loosens.
Citrate is a tridentate chelator. It grabs Ca²⁺ out of solution; the solution then pulls calcium back out of the colloidal phosphate to restore equilibrium; the micelle partially dissociates into smaller, more soluble protein fragments. Those fragments are better emulsifiers than intact micelles: more surface-active protein, more fat globules coated, a smoother mix.
This is not a gelato invention. It is the principle behind processed cheese, where 21 CFR 133.169 requires an emulsifying agent and lists sodium citrate among the permitted ones, capped so that the solids of that agent are no more than 3% of the weight of the finished cheese. A gelato lab is borrowing cheese technology at a tenth of the dose.
Quick reference. Sodium citrate does not thicken, does not sweeten and does not control ice. It binds calcium and it resists pH drops. Every legitimate use in gelato traces back to one of those two actions.
Figure 1 — Milk sits near pH 6.7; casein flocculates as the mix approaches its isoelectric point at pH 4.6. Citrate's buffering strength is concentrated in the band between them.
Buffering: Where Acid Meets Milk
Casein has an isoelectric point at pH 4.6. Fresh milk sits at roughly pH 6.6 to 6.8. Everything a gelato maker does with fruit, yogurt or natural cocoa moves the mix from the first number toward the second, and somewhere on the way the protein stops being dissolved and starts being a precipitate. That is the grain you feel on the tongue.
The citric acid system has three dissociation constants — pKa 3.13, 4.76 and 6.40 — and a buffer resists pH change most strongly within about one unit of a pKa. Two of those three sit inside the danger band between milk pH and the casein isoelectric point. That is why citrate is the buffer of choice here.
| Situation | What goes wrong | What citrate does |
|---|---|---|
| Yogurt or fruit gelato on a milk base | Acid drives local pH toward 4.6, casein flocculates, texture goes sandy | Holds the mix above the flocculation threshold while the acid is dispersed |
| Natural (non-alkalised) cocoa in a dairy base | Natural cocoa runs acidic, roughly pH 5.0 to 6.0, and drags the mix down | Absorbs the drop; the alternative is switching to a dutched cocoa |
| Cheese gelato (gorgonzola, ricotta, pecorino) | Aged cheese proteins refuse to disperse and break into strings | Dissociates the casein network so the cheese melts into the base |
Working acid-forward flavours regularly is worth pairing with a pH meter rather than guessing, and with knowing how titratable acidity differs from pH — the two numbers fail in different ways.
Milk at pH 6.7 meeting fruit acid: the moment citrate is bought for, and the moment it is easiest to overshoot.
Dose, Sodium Load, and Where the Line Is
Working practice in gelato labs puts sodium citrate at 0.1% to 0.3% of the mix — 1 to 3 g per kilogram, a fraction of what processed cheese is allowed. Below 0.1% the buffering capacity is too thin to hold against a real fruit addition. Above roughly 0.4% a soapy, faintly saline note appears, with a slick mouthfeel that reads as cheap rather than creamy.
The sodium arithmetic is worth doing once. Trisodium citrate dihydrate is 23.4% sodium by weight (three sodium atoms at 22.99 g/mol against a molar mass of 294.10):
| Dose in mix | Sodium added per kg of mix | Sodium added per 100 g serving |
|---|---|---|
| 1 g/kg (0.1%) | 234 mg | 23 mg |
| 2 g/kg (0.2%) | 469 mg | 47 mg |
| 3 g/kg (0.3%) | 703 mg | 70 mg |
For scale: 100 g of whole milk carries about 43 mg of sodium (USDA FoodData Central). At 0.2% you are roughly doubling what the milk fraction brought in. That is not a nutritional problem at these levels, but it is a flavour one, and it is why the ceiling exists. If the base already carries added salt, count both — the interaction with salt in gelato is additive on the palate long before it is additive on a label.
What It Does to Your Freezing Curve
Any dissolved salt depresses the freezing point, and a salt that dissociates into four ions does more of it per gram than sucrose does. Expressed as an anti-freezing power factor relative to sucrose:
| Form | Molar mass | Ideal particles | PAC factor vs sucrose |
|---|---|---|---|
| Sucrose (reference) | 342.30 | 1 | 1.00 |
| Trisodium citrate dihydrate | 294.10 | 4 | ≈ 4.7 |
| Trisodium citrate anhydrous | 258.07 | 4 | ≈ 5.3 |
At 2 g/kg of the dihydrate that is roughly 9 g of sucrose-equivalent freezing work per kilogram, with no sweetness attached. Treat it as an upper bound: the figure assumes complete dissociation and ideal behaviour, and a trivalent anion in a concentrated dairy mix does not behave ideally. The practical instruction is not to ignore it — add 3 g/kg to a mix already at the soft end of its range and expect it to draw softer still. Mechanism in freezing point depression, bookkeeping in PAC.
At 0.1% to 0.3%, a kilo of mix takes 1 to 3 grams. This is a 0.01 g scale job, not a teaspoon job.
Four Jobs It Cannot Do
Most disappointment with sodium citrate comes from asking it to be something else.
- It does not build viscosity. No gum, no body, no chew. If the mix is thin, it stays thin — that is what stabilizers are for.
- It does not control ice crystal size. It lowers the freezing point slightly, changing how much ice forms at a given temperature, but it does nothing to the crystal growth mechanism.
- It does not rescue a badly balanced recipe. A mix short on total solids will still be icy, buffered or not.
- It is not an emulsifier in the mono- and diglyceride sense. It frees up protein; it does not adsorb at the fat interface. For fat destabilisation control, see the emulsifier comparison.
One interaction catches people out. Citrate binds calcium, and some hydrocolloids need calcium to set. Sodium alginate gels through calcium bridging, and iota-carrageenan is calcium-dependent too — dose citrate into either and you weaken the gel you were building. Kappa-carrageenan is potassium-dependent and largely indifferent, one more reason the carrageenan types are not interchangeable.
How to Add It Without Ruining a Batch
The salt is fully soluble and needs no hydration schedule, which makes it one of the easier additions in the lab.
- Weigh on a 0.01 g scale. At 1 to 3 g per kilo, a 1 g scale rounds you into error.
- Disperse into the cold liquid phase, before heating and before the sugars. It dissolves almost instantly.
- Never add it dry to a hot, acidified mix. Localised alkalinity where the crystals land leaves an off-note that does not fully disappear.
- Add acid last and slowly, with the mix moving. Buffering capacity is finite; a slug of lemon juice into a still mix wins locally no matter how much citrate is present.
- Taste at 0.1% before going higher. The buffering effect is often complete well below the point where flavour suffers.
On a cocoa-based recipe, decide first whether the choice between natural and dutched cocoa already solves the pH problem for free. On a yogurt base, the ingredient order in a yogurt gelato recipe matters more than the additive does.
The target: dense and silky, with no trace of the grain that protein flocculation leaves behind.
If the texture problem turns out to be crystalline rather than proteinaceous, sodium citrate is the wrong tool entirely — start with the diagnostics in sandy gelato texture and with how casein and whey behave differently under heat and acid.
