Why Sauces Split: Emulsions and Pan Sauces

What an emulsion is

Oil and water do not mix. An emulsion is an unstable state in which one of them has been broken into tiny droplets and dispersed through the other. Mayonnaise, hollandaise, a cream pasta sauce, the opaque broth of ramen — all emulsions.

Unstable is the operative word. An emulsion is always trying to separate, and what holds it together is an emulsifier.

Emulsifiers worth knowing

• Egg yolk (lecithin): the most powerful. Mayonnaise, hollandaise, Carbonara.
• Starch: physically traps the droplets between water and fat. Pasta water, potato cooking water, a flour roux.
• Mustard, honey, grated garlic: weak, but genuinely effective — the reason a spoon of mustard goes into a vinaigrette.
• Myosin: muscle protein from meat or fish. Salt dissolves it out, and dissolved it holds fat — which is the whole of what makes Weisswurst and Kamaboko work. How it differs decisively from the ones above is below.
• Gelatin: the odd one out, because you do not add it — the cooking makes it. Collagen out of bones and skin dissolves into the pot and holds fat in the water, and that is the white of Seolleongtang. It is why the last section of this guide inverts every temperature rule above.

And every example on that list is a pourable sauce. Emulsions also set. The ganache in Chocolate Truffles is the chocolate's cocoa butter dispersed in the cream's water, held by the chocolate's own lecithin and cocoa particles. It sets firm enough to roll rather than pour, and it is still an emulsion — which is why the rescues below work on it.

Pasta water is not water

Water that has boiled pasta carries dissolved starch, and that starch acts as the glue between olive oil and water. It is why the sauce in Aglio e Olio clings to the strands instead of splitting into a puddle of oil.

Hence the advice to save a cup before draining. Plain water thins a sauce; pasta water thickens it, and it is salted, so it seasons at the same time.

Temperature is usually the culprit

The most common reason an emulsion breaks is heat.

• Egg-based sauces coagulate above about 70°C. Make Carbonara over a live flame and you get scrambled eggs, not sauce. Take the pan off the heat and use only its residual warmth.
• Cream splits if you boil it hard — the fat clumps and the whey separates. Bring it just to the edge of a simmer.
• Cold butter dumped into a hot pan melts and separates at once. Lower the heat and stir in cold butter one piece at a time (monter au beurre). The fat emulsifies into the liquid and the sauce turns glossy.

When heat completes an emulsion instead of breaking it

Every temperature rule above is about getting too hot while you make it. A myosin emulsion is exactly the reverse.

• The danger is at the beginning, not the end. The paste for Weisswurst splits on the spot if it goes past about 12°C while it is being mixed, because the fat melts out. That is why the meat, the machine and the ice are all kept cold. Hollandaise breaks because it got hot at the end; this breaks because it got warm at the start.
• And heat sets it permanently. Once cooked it never splits again. Every other emulsion in this guide rests on the premise that it is always trying to separate; here cooking is the completion.
• In exchange there is no way back. The rescues below are for liquid emulsions. A split sausage paste cannot be whisked back together.

An emulsion you buy is not an emulsifier

Every emulsifier above is something you add in order to make an emulsion. When a recipe buys mayonnaise instead, that is not what it is doing. It is buying the finished thing whole: the oil already dispersed and held, thick without heat, stable at room temperature, and setting or softening with the temperature of its fat rather than with cooking. Which is why it does jobs a sauce cannot.

• An adhesive. Elote says it outright — "the mayonnaise is glue, not sauce." Everything that follows it is either a dry powder (cotija, chilli) or a thin liquid (lime), and neither would stay on a smooth, wet, curved cob. Thick, tacky and stable at room temperature is the job description.
• A structural binder, and the fridge is where it sets. Crab cakes cannot draw out the myosin listed above: the crab arrives boiled and its protein is already denatured, so there is nothing to dissolve. The fat in the mayonnaise and the butter stands in, and the page draws the conclusion — "if the glue is the fat, then the temperature at which that glue sets is in the refrigerator, not over the heat." Skip the thirty minutes and the first cake collapses as you turn it.
• A frying medium. Grilled cheese calls mayonnaise in place of butter "a swap of ingredients, not a trick": butter's milk solids start to burn at around 150°C, which is the exact temperature the bread has to reach, and mayonnaise has no milk solids in it to scorch.
• A substitute that cannot split, at a price. The aioli in Patatas bravas is properly garlic and olive oil alone, and the list above files grated garlic among the weak emulsifiers. Mayonnaise "brings egg yolk, the strongest emulsifier there is, so it cannot split — at the cost of tasting of egg rather than of garlic."
• What thickens it further is more of its own emulsifier. Tamago sando crushes cooked yolk through it. That is not catching fresh oil; it is stiffening what is already caught until it holds its shape between two slices of bread.

It is the same property as the ganache above, which is an emulsion and still firm enough to roll, and the section immediately below is that property's failing face: stable at room temperature also means not stable on lukewarm ingredients. Coleslaw salts and drains its vegetables first, which is that failure stopped at the source.

When a finished emulsion breaks because of what you put it on

There is one more case, and it is not during the making but after. The mayonnaise in Olivier Salad breaks when it is folded into lukewarm ingredients: the fat softens so the droplets coalesce, and steam rising off the food condenses into the sauce and thins it. That is why the potato and egg are cooled completely first, and why a lukewarm one leaves a puddle in the bowl by the next day.

German Potato Salad dresses the same potatoes while they are hot — but its dressing is vinegar, mustard and bacon fat, so there is no emulsion to break, and hot starch absorbs it well.

Acid is the other culprit

Pour lemon juice or wine into cream, or cream into them, and you get different results. Dairy added to a strongly acidic pan curdles immediately. Reduce the acid first to soften it, then add the cream.

And a culprit with no villain in it: the water left

Everything above breaks the emulsion by damaging the emulsifier or by adding something. There is a quieter cause with no villain in it at all — the continuous phase evaporates. An oil-in-water sauce needs the water to hold the oil dispersed in; keep the pan on while you chase a thicker result and there is eventually too little left to disperse into, and the oil comes back out. Aglio e olio states the endpoint: boil it off chasing thickness and "there is nothing left to emulsify into, so it slides apart into oil at the bottom of the bowl."

Note this is the exact reverse of the lukewarm-mayonnaise case above, where steam condensing into a sauce thinned it. Same sauce, same ingredient, opposite direction — which is why "it looks too thick" and "it looks too thin" are not opposite problems with one answer between them.

Naming it is what makes the rescues below legible. Two of the four are adding water back, and read without a cause they look like tricks. They are not: they restore the phase that left.

What you add back with is decided by what is holding the sauce. Where the emulsifier is egg, plain water is fine — eggs benedict loosens a stiff hollandaise "with a teaspoon of water at a time", because the yolk is unaffected and the sauce only needed volume. Where the emulsifier is starch, plain water dilutes the very thing doing the work (cooking noodles puts it as "too much water dilutes the starch"), so what goes back is the starchy cooking water — which is why cacio e pepe adds it "bit by bit" to the end rather than topping the pan up once.

And that water is not a pasta thing. Bryndzové halušky keeps a cup of the water its potato dumplings boiled in and says so outright: "it does the same job as pasta water — loosened with it, the sauce clings to the dumplings instead of sliding off." The bryndza is mashed smooth with five tablespoons of it off the heat, and that page gives the other half of this guide's own temperature rule in one line: put the pan back on the flame and the cheese "turns grainy and splits into oil". Dao xiao mian adds the seasoning consequence — if the broth has gone salty, loosen it with the noodle water rather than plain water, because plain water costs you the body as well as the salt.

Bringing a split sauce back

Things to try before starting over:

1. Off the heat, and let it cool. Most emergencies are simply too much heat. 2. Add a splash of cold water or an ice cube and whisk hard. Works well for the hollandaise family. 3. Whisk a fresh yolk in a clean bowl and dribble the broken sauce into it. You are re-emulsifying from scratch; this has the highest success rate. 4. Add a little starchy water. For a pasta sauce, two tablespoons of pasta water usually fixes it.

Thick, but not emulsions

Some things are easy to confuse. The creaminess of Risotto is starch released from the rice, not an emulsion; a butter sauce on Gnocchi is one, but the softness of Chocolate Mousse is fat crystals and aerated foam. Keeping the mechanisms apart is what tells you which thing to fix when it fails.

Not everything made of chocolate belongs here, though. The ganache in Chocolate Truffles above is a real emulsion — a mousse is a foam and cannot be brought back once it breaks, but a split ganache can. Filing it here because it sets means throwing away something fixable.

And a milky broth is an emulsion. The white of Seolleongtang is bone fat and gelatin emulsified by a hard boil, which is why boiling long and hard is the method, and Bouillabaisse boils hard deliberately to drive its olive oil into the broth. In most cooking "a hard boil splits it" is the rule; when the emulsifier is gelatin, it inverts.

White is not a diagnosis

A broth going opaque does not tell you an emulsion happened. Galbitang spends its whole recipe keeping one clear, and it separates four things that cloud a broth: fat emulsifying at a rolling boil — "which is seolleongtang, the same beef bones taken to the opposite result" — starch off the tteok, an onion left in until its pectin breaks down, and fat left on the meat. Samgyetang states it as a rule: its broth clouds in about forty minutes, and what clouds it is starch escaping from the glutinous rice inside the bird plus a young chicken's collagen. "One colour, two causes."

So the test is not the colour, it is whether violent agitation was required. An emulsion has to be beaten into existence — seolleongtang needs a rolling boil, and the same bones at a bare simmer give a clear yellow broth. Starch and pectin need nothing of the kind; they arrive on their own, whether or not you wanted them. Which puts a white soup in the same box as Risotto above: knowing which mechanism it is, is knowing what to do about it.