Overrun and Air Bubbles — Air as the Fourth Ingredient

Advanced Chapter 1 counted air bubbles among the four phases that make up gelato. Alongside ice, fat and the unfrozen phase, they are a structural material in their own right. But air is invisible and weighs almost nothing, so it is easily overlooked as something to design. This piece takes that fourth ingredient head on. The number at the center of it is overrun.

Overrun is the volume gained

Freeze the mix while stirring it, in the mantecazione, and air is taken up inside; the volume grows. Overrun is that gain in volume, given as a percentage.

Measuring volume on the floor is a nuisance, though. So in practice the same vessel is filled and weighed. Air carries almost no mass, so a mix and a gelato packed into the same vessel share a volume and differ only in mass. Take the ratio of the densities, then, and overrun comes out without measuring any volume.

Fill a cup of known capacity, level it off, and weigh — that alone yields the number.

At an overrun of 100 percent the volume is twice what it was — half of the finished gelato is air. Industrial products force air in with a pump, so that figure passes 100 percent easily. Artisan gelato, by contrast, stays within what stirring takes up on its own. As a guide, milk-based gelato does not go much past 35 to 40 percent. Fruit gelato is water-based and short of the protein and fat that build a protective film, so it settles around 20 to 30 percent. And that 35 to 40 percent is an average, not a rule. The literature points one step further.

A mix at 32 percent total solids takes an overrun around 32 percent; at 38 percent, around 38. The solids settle how much air can be carried — that is how the rule reads. The band of 35 to 40 percent is nothing but the other side of a fact. Crema-based total solids sit at roughly 32 to 42 percent (Advanced Chapter 5).

The literature is plain, though, that the rule does not carry over to water-based gelato. A fruit sorbetto is high in sugar and has neither egg nor fat, so air enters it less readily. It is set apart, with 25 to 30 percent taken as a satisfactory level. How far a rule reaches is part of the rule.

There is also a way to cut the air on purpose. If the aim is a dense, tight serving, put more mix into the cylinder. The volume of air left inside falls by that much, and less of it is taken up. The adjustment is made by how much is loaded, without touching the formulation.

One more thing follows when the air goes up. The more air there is, the smaller both the ice crystals and the bubbles become. The gelato then reads as less cold, and melts more slowly. Air is not filler; it is a design variable that moves coldness and melt.

The character seen in Foundations Chapter 1 — dense and rich — follows directly from this low overrun. Of that chapter's close, "Rich, yet not heavy", the rich half is explained by this one point: there is little air. The not heavy half is a matter of fat, and goes to Approfondimento 7 and the aftertaste.

What the bubbles are doing

Why put air in at all? With none at all, a frozen mix is a hard block of ice — impossible to scoop, and no melt to enjoy. The bubbles do several jobs at once.

First, they lighten the structure and make it easy to scoop. Second, they push in between the ice crystals, thinning the ice and softening the impression of cold and of grit. Third, fine bubbles themselves lend smoothness to the mouthfeel. Air is not filler: it is a material that actively builds texture and melt. Put in too much, though, and the flavor thins and the corpo is lost. Gelato chooses a low overrun because it sets the balance of rich against light at a point that does not lean too far toward light.

Size, not amount

Here is the heart of it. The number called overrun stands for the amount of air. But what settles the mouthfeel is less the amount than the size of the bubbles and how even they are. The same shape of thing was seen with ice crystals (Approfondimento 2).

Fine, even bubbles give a smooth and creamy structure. Coarse, uneven, large bubbles give the opposite — a hollow, unreliable texture, the "spongy" defect listed in Advanced Chapter 13. At the same overrun of 30 percent, whether fine bubbles are spread evenly or large ones sit sparsely makes all the difference in quality. Air, like ice and fat, is a material to be spoken of in terms of "how fine the grain is".

The bubbles are a product of the whole system

So how are fine, even bubbles to be had? The answer has already been given by the deep dives so far. What folds air in finely is the mechanical shear of the mantecazione. The same stirring that mass-produces ice nuclei (Approfondimento 2) drags air in and breaks it into fine bubbles.

And air goes in only at the very start of freezing.

The literature draws the line by temperature. Between 0 °C and −4 °C — that is the window for taking air up. Below −4 °C the gelato takes up no more air. So the freezing must not be stopped partway. Stop it and the air already taken up escapes, and the product collapses on the spot.

That one line settles the order of the work. "Adding air later" cannot be done. Overrun is decided outright in the first few minutes in the mantecatore. The ice nuclei of Approfondimento 2 are decided in those same few minutes — so most of this step is packed into them.

Aging carries a condition of its own. Put the mix in warmer than +4 °C and it loses, during aging, the air it had already taken up. Being cold is also for the sake of the air.

Which materials help the air, and which get in its way, the literature lists apart as well.

• Helping — egg yolk, egg white, stabilizers in measure, skimmed milk powder, caseinates, some minerals
• Getting in the way — too much fat, MSNF (milk solids-not-fat), sugar, cacao or nut paste; an excess of calcium salts

Skimmed milk powder is worth noting: it appears on both lists. In measure it helps; in excess it gets in the way. The account in Advanced Chapter 3 — that MSNF is pressed from above and below — turns up here in the same shape.

Raising a fruit sorbetto with raw egg white is not to be recommended. One reason is hygiene: putting raw egg white into a mix that will take no further heating is risky. If it is used, choose a pasteurized one. The other is structural: an ice confection raised with egg white will show syneresis and sink easily if the stabilizer is short.

Breaking the air up, though, is not enough — the bubbles soon grow. Two mechanisms hold that back.

One is the adsorption of protein. Milk proteins gather at the bubble surface and line up, hydrophilic parts toward the matrix, hydrophobic parts toward the air. This lowers the surface tension at the boundary and weakens the very driving force of the coarsening seen below. The layer of adsorbed protein also stands in the way, bulk against bulk, and keeps bubbles from sticking to one another.

The other is the adsorption of fat globules. Clusters of partly coalesced fat cling to the bubble surface and clothe it like an armor of particles (Approfondimento 5). This way of steadying a dispersion — solid particles adsorbing at a boundary — is called Pickering stabilization (stabilizzazione di Pickering). And the thick unfrozen phase (Approfondimento 4) slows the very movement of bubbles past one another.

Here the difference between gelato and whipped cream tells. Whipped cream carries fat past three tenths — enough to clothe the whole bubble surface in fat. So it can hold its foam on fat alone, with no emulsifier added. Gelato has less fat and cannot cover the surface. The shortfall is made up by using an emulsifier to push partial coalescence along — and that is why Advanced Chapter 4 called for emulsifiers.

A good overrun, then, is a product of the whole system — it comes only when fat, protein, stabilizer and stirring mesh. Chasing the overrun figure alone will not give good bubbles. The weight put on machine performance in Advanced Chapter 7, and the discussion of fat and stabilizers in Chapters 3 and 4, tie together here.

The bubbles grow too

And here is a fact not to be missed. A dispersion of fine bubbles, left alone, coarsens too. Just as with ice crystals (Approfondimento 3), fresh is best for air as well.

A mean diameter of a little over 20 µm on leaving the freezer grows to nearly double through hardening alone. Put it through swings of temperature and bubbles past 100 µm appear. Bubbles are at their finest the moment they are made.

There are two roads to coarsening. One is coalescence — neighboring bubbles touch, the film between them breaks, and the two become one (the same logic as with fat globules). The other is disproportionation (sproporzionamento), and that one is a little more involved.

Inside a bubble, surface tension squeezes, and the pressure runs higher than outside. That difference Δp grows larger the smaller the bubble's radius r.

And how readily a gas dissolves runs in proportion to pressure — Henry's law. So the air inside a small bubble dissolves the more readily into the matrix around it. The dissolved air spreads through the matrix and turns back to gas at a large bubble, where the pressure is lower. Air moves on balance from small to large, and the small bubbles vanish while the large ones grow.

The story should look familiar. The smaller a thing is the less stable it is, and the small are eaten by the large. This is exactly parallel to the Ostwald ripening of ice seen in Approfondimento 3. There the driving force was that small crystals have a higher solubility. Here it is that small bubbles carry a higher pressure and so dissolve more readily. Ice and air coarsen by the same logic.

This is just why the two stabilizations named above tell — the lowering of surface tension by protein, and Pickering stabilization by fat. Lower the surface tension γ and Δp falls with it, and the driving force of disproportionation weakens at the root. Making the bubbles fine and keeping them fine are not the same work.

With that, the design of all four phases — ice, fat, unfrozen phase, bubbles — is on the table. How they come together in the end into the single sensation of melt is the subject of rheology (Approfondimento 10).