Partial Coalescence — How Fat Holds the Air Bubbles Up

The structure of gelato carries a demand that looks like a contradiction. Advanced Chapter 1 took the emulsion of fat globules as something whose dispersion must be kept. Advanced Chapters 3 and 4 said that "breaking it in part" during freezing is what builds the structure. Is it to be stabilized, or broken? The key that unties this is partial coalescence (coalescenza parziale). The paradox of Advanced Chapter 4 — that an emulsifier weakens the emulsion — clears up in this one term.

Coalescence, and partial coalescence

The fat in a mix is dispersed as an oil-in-water emulsion. Each globule is covered by a film of protein or emulsifier, and they repel one another and stay dispersed. In an ordinary emulsion, two droplets breaking their films and merging into one larger droplet — coalescence — is a fault to be avoided. Let coalescence run and the fat separates; the emulsion collapses.

In gelato, though, something close to it but not the same is put to work on purpose. It happens only when the fat globules are partly crystalline, and it is partial coalescence. Liquid droplets merge into one round globule. But a globule holding a skeleton of solid fat crystals inside cannot go back to a sphere. A crystal jutting out of one globule enters the liquid oil of another. The two link, and the crystal skeleton stands in the way of merging completely. Keeping something of their original shapes, they become a lump out of true. So the fat globules join not as round lumps but as irregular clusters. That is why it is called "partial" coalescence.

Why does the crystal go in, and why does it not come out?

There is something to ask here. Why does the crystal enter the neighboring globule, and why does it not come out? This question has to be split in two.

Going in takes force. The stage at which a crystal pierces the film of the neighboring globule is in fact the rate-limiting step of partial coalescence. By the literature, fat globules that meet reach partial coalescence only once in a million times. So what is hard is not the meeting but breaking the film. That is why shear tells. A high shear stress presses the globules together and helps the crystal through the film. The thinner the film an emulsifier makes, the easier the passage; the thicker and more viscoelastic, the more it resists. The deepest part of the paradox of Advanced Chapter 4 — that an emulsifier weakens the emulsion — lies right here.

Staying in is thermodynamics. A crystal jutting out from the surface of a globule is surrounded by water. For a fat crystal that is an uncomfortable place. Enter the oil of the neighboring globule, though, and it is surrounded by oil. A fat crystal is better off surrounded by oil than by water — the interfacial energy falls, which is to say the hydrophobic effect. So a crystal that has once entered stays there. The two globules remain joined rather than parting.

Force to enter, thermodynamics to stay. Neither alone accounts for it.

The same reasoning tells with time. As two joined globules merge a little further, the area of oil exposed to water falls further still. The system wants to go that way. So the joint made by partial coalescence, as time passes, grows stronger and harder to break. A cluster is at its most fragile the moment it forms.

How far the crystal juts out tells as well. The further a crystal juts from the globule surface into the water, the more readily it enters the next globule. A crystal buried inside the globule cannot reach anything to pierce. Where in the globule the crystals form — scattered inside, or lined up at the interface — is settled by how it is cooled. Here is another reason why aging is not merely time spent waiting.

Neither liquid nor solid will do — what aging means

Partial coalescence needs the fat globules to hold liquid and solid at once. Both extremes fail.

If all of it is liquid, what happens is ordinary coalescence. The globules merge into one large round globule, and the emulsion breaks. If all of it is solid, they cannot merge at all. Hard particles meeting merely gather together — aggregation — without linking. There is no liquid oil anywhere for a crystal to pierce.

Even so, "all solid is safe" cannot be said. A fully crystalline fat globule can change shape and gain surface, until the emulsifier no longer covers it. Exposed non-polar faces then draw on one another, and wide aggregation can follow — a different road, but the emulsion breaks all the same. Both extremes only mean "no partial coalescence"; they do not mean "nothing happens".

Partial coalescence holds only where liquid fat and crystals live together. The crystal keeps a skeleton and refuses to return to a sphere. The neighbor has liquid oil for it to enter. Both are needed.

Nor is living together enough on its own. Raise the solid fat content and the readiness for partial coalescence first rises, reaches a peak, and from there falls away. Too few crystals and there is no skeleton; too many and there is no oil left to pierce. Partial coalescence runs best in between. The fat must neither melt away entirely nor set entirely.

What makes this state is the cold of aging, taken up in Advanced Chapter 6. Held at about 4 °C for several hours, the fat inside the globules crystallizes in part. Not all of it sets; much of the inside stays liquid. When Advanced Chapter 3 said that a partly set state raises the whipping quality, this partial crystallization was the preparation for partial coalescence. Aging is not merely time for flavors to settle. It is a physical step that builds into the fat the ability to coalesce partially.

And milk fat is a fortunate material in this respect. It holds a partly crystalline state across a wide band of temperature. That is why partial coalescence matters so particularly in dairy products. Ice cream, whipped cream, butter and margarine all stand on this one principle. Gelato belongs to that line as well.

Why should things fall out so conveniently? The reason is that milk fat is not a single substance.

The melting point of a fat is settled by the chain length of its fatty acids and their degree of unsaturation. Where on the glycerol they sit tells as well. If there were only one pure triacylglycerol, the melting point would be one point. It would melt all at once at a temperature, as ice melts at 0 °C. But a natural fat is a complex mixture of many triacylglycerols with different melting points. They melt little by little, the highest first. So melting is not a point but a band.

Nor is it a simple sum. The higher-melting triacylglycerols dissolve into the lower-melting ones. How a mixture melts is not the weighted sum of how its parts melt.

And it is this band that lets the solid fat content be chosen by choosing a temperature. Temperature settles the solid fat content, and that settles partial coalescence. To fix the temperature of aging is to follow this chain of two and choose how strongly partial coalescence will tell. Had milk fat been a single triacylglycerol, there would have been no choice at all. Above one temperature all liquid, below it all solid — no question of aiming at the top of a hill.

That said, the temperature is not the only dial. Where the temperature cannot be moved, there is the option of choosing a fat that melts differently. Change the melting band itself and the solid fat content at the same temperature changes with it. And, as the next section shows, how it is cooled changes the number and the form of the crystals as well. At the same solid fat content, crystals of a different character give a very different readiness. Same solid fat content, same behavior — that does not hold.

How it is cooled settles the crystals

What aging gives, though, is not only the quantity called solid fat content. What kind of crystals form is settled by the cooling as well.

First, number and size. The same tug of war seen with ice is at work here (Approfondimento 2). Cool fast, well below the melting point, and many small crystals form. Cool slowly, to just under the melting point, and there are few large ones. The rate of nucleation rises more steeply with falling temperature than the rate at which crystals grow. The logic seen with ice carries over to fat unchanged.

Quality changes too. Cool slowly and the molecules have time to settle properly into the lattice. Cool fast and there is no such time — the molecules are packed carelessly, and the crystals come out full of defects and loosely packed.

And one level deeper, something else changes. How they mix.

Cool fast and the high-melting and the low-melting crystallize at about the same time. What forms is a homogeneous crystal in which the various triacylglycerols are intimately mixed — a solid solution. Cool slowly and the high-melting set first, the low-melting later. What forms is an uneven crystal, with regions rich in the high-melting and regions poor in them.

Here is where it tells. Whether a solid solution or an uneven crystal forms changes how that fat itself melts. Density changes, and so does how it compresses.

The band written of above lets the solid fat content be chosen by temperature. That band is itself rewritten by how the fat is cooled. Temperature settles the solid fat content, true; but the very table matching temperature to solid fat content is remade by the speed of cooling.

And one thing more. Once crystals have formed inside a globule, the crystals gather into a three-dimensional net and hold liquid oil by capillary force. Gathered crystals go on to merge in part and strengthen the net. Here too, freshly made is most fragile — the same story as the joint of the cluster, now inside the globule.

Further, the same fat can take several crystal forms. A triacylglycerol can pack its molecules in more than one way. This is polymorphism. The order of stability is fixed: β is the most stable, then β′, and α the least. Yet a fat often crystallizes not in the most stable form but in a metastable one. The barrier to forming a nucleus is lower there. Being easy to form and being stable are two different things. And a metastable crystal shifts, over time, toward the stable form.

One last point to press. All of this happens inside the small vessel that is a fat globule. The crystallization of emulsified fat is not the same as the crystallization of fat in bulk. Heat travels differently, and the globule surface is a physical constraint. Where crystals form, what form they take, how fast they move to a stable form. On all of it, an intuition drawn from bulk fat does not necessarily carry over.

Shear pulls the trigger

What pulls the trigger of partial coalescence, on fat globules thus prepared, is the shear of the mantecazione.

Shear first lays on two places to meet. One is to make the globules collide hard in the water phase. The other is to draw the globules onto the surfaces of the bubbles raised by stirring, and set them side by side there.

On top of that, shear speeds partial coalescence along three paths.

• It raises the frequency of collision. Stir, and the globules meet more often
• It raises the efficiency of collision. In a shear field the globules roll against one another. A jutting crystal more readily takes an angle that lets it enter its neighbor
• It helps the piercing itself. A high shear stress presses the globules together and eases the crystal through the film at the interface

Frequency, efficiency, piercing — answering in turn to "do they meet", "do they meet well", and "can it get in". So at the same speed of rotation, a disturbed flow gives faster partial coalescence than an orderly one. The film can break in part and expose a fat face — the partial de-emulsification of Advanced Chapters 3 and 4. The crystal then enters more easily still, and partial coalescence runs on in a chain.

That is to say, the very stirring that folds in the air is the driving force of partial coalescence. As freezing forms ice crystals, the unfrozen phase concentrates and the fat globules crowd together, and that too raises the frequency of collision. Behind the claim of Advanced Chapter 7, that rapid freezing with agitation settles the structure, this linking of the fat is under way.

The net that holds the bubbles

Clusters of fat made in this way settle onto the interfaces of the bubbles taken in by stirring. Partly coalesced fat covers the bubble surface, and the clusters join one another to string a continuous net of fat between bubble and bubble.

This net holds the structure of gelato up from within. The bubbles are cradled in the net of fat and steadied, and are less apt to collapse. The same net gives a smooth, "dry" quality, good shape retention, and resistance to melting. Whipped cream stands in peaks by exactly the same principle of partial coalescence. Advanced Chapter 1 counted the air as one of the four components. It can work as a structural material because of this quiet labor of the fat. Of all the roles fat plays, steadying the bubbles is held to be the most important.

Not all of the fat links, though. Typically about three tenths of the fat is in a partly coalesced state, and the rest is scattered as separate globules. Linked fat and lone fat: both together is the real state of things inside gelato.

The paradox of the emulsifier

Here the paradox of Advanced Chapter 4 is untied. An emulsifier such as a monoglyceride is generally used to steady an emulsion. In gelato, though, an emulsifier is used to displace the protein of the fat globule membrane and weaken the film. It is there to promote partial de-emulsification under shear. The aim is not a stable emulsion but a measured destabilizing. When Advanced Chapter 4 put the point of emulsifiers at the control of partial de-emulsification, it was pointing at this partial coalescence.

The paradox shows plainly in the size distribution. Made without emulsifier, the fat globules sit in a single small peak at about 1 µm, made by homogenization. Add an emulsifier and part of that is destabilized, and a second peak rises at about 10 µm — the clusters formed by linking. The emulsifier is put in precisely to raise that second peak.

The dial is not only the amount. The kind of emulsifier tells too. The fatty acid tails of a monoglyceride come saturated and unsaturated. An unsaturated tail is kinked at its double bond, so the molecules do not pack neatly and take up room. As a result they drive more protein off the membrane, weaken the interface further, and push partial coalescence along. Which emulsifier, and how much — both settle how far the fat will link.

Go too far and it turns to butter

Partial coalescence must be neither too little nor too much. Where it does not run, the bubbles are not held, the structure is coarse, and it melts readily.

And where it runs too far? The emulsion turns inside out.

Let partial coalescence spread and the clusters of fat go on joining. At last the state passes from oil-in-water to water-in-oil — from oil scattered through water to water scattered through oil. This is phase inversion. It is what churning to butter is, and it is why cream beaten too long becomes butter.

What is striking is that this very inversion is the goal in another industry. Making butter, margarine and spreads is nothing but a process of causing the inversion on purpose. The net of crystals formed in the oil then holds the water droplets. Too many crystals in that net and it is too hard to spread; too few and it slumps under its own weight. So choosing a fat with the right melting behavior becomes the first concern of that industry.

Gelato is the work of stopping short of that inversion. The same physics is used, and the trades divide into those who drive it on and those who hold it back. The fineness of the globules from homogenization, the fat, the emulsifier, and the degree of crystallization in aging all settle this balance. The "dry and smooth" of gelato stands on a narrow optimum between collapse and excess.