Ice Cream


My history with ice cream probably mirrors that of most other cooks: from not really knowing what I was doing to becoming fully obsessed with its inner workings. As one begins to grasp a better understanding of ice cream, it becomes at once easier and oddly far more difficult. The more you know, the more you realize you don’t know.

Let me be honest – this is not going to be the epic treatise on ice cream or the last word on a subject that I’m wholly unqualified to author. I will instead propose a handful of obvious (or not so obvious) concepts that might simply serve as a launching point deeper into the science of the subject.


1. While there may be no one ‘ideal’ ice cream formula, one can assemble that formula much like an algebraic equation based on the desired end result. The key to success is knowing which components are proportionally fairly static and which are variable. And then it’s about knowing how your ingredients supply these basic components.

One can very generally place ice cream formulas and their constituent components within the following ranges:

Milk Fat 10% – 16%

Egg Yolk Solids 0% – 2%

Nonfat Milk Solids 9% – 12%

Sweeteners 12% – 16%

Stabilizers and Emulsifiers 0% – 1.0%

Water 55% – 64%

There are, of course, exceptions. Gelato-style products often have a fat content in the 7-8% range; soft serve products may contain 5% fat or less. My own go-to formulas tend to fall within a range of 7-9%, though technically I could never commercially call it ‘ice cream’, which is defined by U.S. law as containing a minimum 10% milk fat.

Crucial to understanding how to build an ice cream formula is knowing the composition of your ingredients. Of course, I think this basic information is important no matter the preparation at hand. With knowledge of an ingredient’s composition, structure, and function comes true power to the cook. Rather than thinking of milk as simply ‘milk’, one must look at it as a system of water, fat, protein, and sugar; it’s structure is at once an emulsion, a suspension, and a solution.

Below, a useful chart for comparing the composition of milk and its commonly used derivatives:

Dairy Composition

And if we employ egg yolks in the formula, it is helpful to know the following:

One “Large” Egg Yolk = 20g

Comprised of, approximately:

50% water

10% proteins

30% fat

10% lecithin


Milk fat and sweeteners will be covered further below, but these charts alone can immediately set us on the path of formulating new ice creams, or reverse-engineering existing recipes to see where its components may fall along the formulation spectrum.

2. Water containing dissolved solids such as salt and sweeteners are affected by something we refer to as colligative properties. These solutes will raise the boiling point of water on the high end of the temperature range, and at the low end, they lower the freezing point of water. It is this very property of freeze point depression that makes ice cream possible at all – that at serving temperatures below water’s freezing point it is soft enough to scoop and chew.

Different solutes – for the sake of our discussion, sweeteners – will lower the freezing point of water to different degrees. The measurement that we use to correlate freeze point depression is a sweetener’s molecular weight – the lower the molecular weight, the greater the effect of freeze point depression. Sucrose, for example has a molecular weight of about ‘342’, with fructose coming in at about ‘180’ and an average glucose at ‘428’. With this we can say that a solution of fructose will lower the freezing point water nearly twice that of sucrose, while a glucose solution will raise the freezing point. What does this mean for an ice cream maker? Simply put, we can use multiple sweeteners to modify the freezing point – the relative firmness or softness – of an ice cream.

Different sweeteners also have a different ‘sweetening power’, which allow us the ability to fine tune the perceptible sweetness in addition to adjusting freeze point depression, all while maintaining a fairly constant percentage of total sweeteners. Sucrose is given a sweetening power of ‘100’, with fructose at about ‘125’, and glucose somewhere in the range of ‘50’ (glucose can offer a confusing range of properties based on how it is processed – that may well be a different discussion at a different time). Thus, for the sake of comparison, replacing some of the sucrose in a formula with fructose will simultaneously give us more sweetness while also lowering its freezing point. Far more useful to us is the fact that added glucose will offer less sweetness while also raising the freezing point, giving us firmer textures at higher temperatures. It is also helpful to know that lactose, with a relatively low perceptible sweetness will still lower the freezing point at the same rate as sucrose.

The chart below offers a rough comparison of these properties among a range of sweeteners. RFDP refers to relative freeze point depression in relation to sucrose, which is given the arbitrary factor of ‘1’. SP refers to sweetening power. The column under Max. % refers to a generally accepted use in ice cream formulas. These figures were culled from various sources over a long period of time – one may see slight differences between sources – especially with regard to sweetening power – but I think this gives us a good rough guideline to work with.

Sweetener Properties

It may also be worth noting the molecular weight and RFDP of sodium chloride – salt – at ‘54’ it will lower the freezing point of water over six times that of sucrose (which why we put salt on icy roads and not sugar!). And ethanol – a component in alcohol – will lower the freezing point by a factor of seven times that of sucrose, with its molecular weight of ‘46’. Formulating ice creams with alcohol can often be frustrating; two general rules of thumb to consider are the need for a reduction in dissolved solids (down to 23-25%) and the addition of a maximum of about 7% pure alcohol. Formulations must also be adjusted when adding ingredients like chocolate and fruit – which might bring sweeteners, water, or fats of their own. I know, the math just got a lot harder.

3. In addition to supplying creamy mouth feel, the milk fat content of ice cream will determine its basic physical structure. The best way to understand the structure of ice cream is to step back and consider first the structure of whipped cream. As we whip heavy cream, we can begin to visualize individual fat particles swirling around the continuous phase of water, slamming into each other almost as if in a mosh pit (my favorite way to describe it). We know that cream whips up best when it’s cold; this is because at low temperatures most of the milk fat is crystallized – solid – which allows the individual particles to stick to one another while maintaining to some extent their own identity (as opposed to simply fusing into larger and larger fat particles). With help from some of the milk proteins, these partially coalesced fat particles begin to form a kind of ‘scaffolding’ – a solid structure – that also traps the air bubbles that are incorporated into the cream as it is being whipped. Ok, that’s whipped cream.

Understanding the structure of whipped cream helps us understand the structure of ice cream because, on a microscopic level, they are really quite similar – the only differences being that there is usually a lot more ‘stuff’ dissolved in the water phase of ice cream (sweeteners) and that some of the water exists as ice. Below, a set of graphics that helps illustrate this idea, courtesy of my friend Cesar Vega, and one of his mentors in ice cream, Douglas Goff (2007):

Whipped Cream  Ice Cream

Also important in the formation of the structure of our finished ice cream is the relative size and dispersion of the milk fat particles. We almost always heat our ice cream bases in order to dissolve sweeteners, cook proteins, and pasteurize the final mix. This heat will liquefy the previously crystalline milk fat; upon cooling the milk fat will have a tendency to form large fat particles. In recent years I have become fanatical about homogenizing the mix after cooking, to aid in breaking up those fat particles, which ultimately give way to better structure. A thorough buzzing with an immersion blender, while not the perfect tool, will certainly yield better results than skipping the step outright. An aging period is also important, among other reasons, as it allows those milk fat particles to properly crystallize.

In short, understanding how the milk fat in our ice cream behaves on this underlying structural level can lead us to make certain determinations on how we process ice cream, its overrun, and its melting qualities. More food for thought: spinning ice cream in a batch freezer allows this partial coalescence of fat to occur (the ‘scaffolding’), while processing the same ice cream in a PacoJet merely ‘slices’ the already frozen base into finer particles. An interesting experiment to try is comparing the melt down of identical ice cream formulas processed in each machine – which do you think might melt faster? Why?

More food for thought: ‘re-spinning’ ice cream should be discouraged solely for hygienic reasons, but we can also begin to imagine how it can be problematic from a structural point of view…


4. Ice cream is made up of a lot of ice. Obviously, right? Ice defines its nature, yet improper formulation or handling can result in the ice emerging as negative attributes – too much of it, or in too large a crystal size. Two important concepts to remember:

The amount of solutes in the unfrozen water phase determines volume of ice crystals that form.


The rate and speed of freezing the base mix determines crystal size – the lower the temperatures, the faster the base freezes to produce the smallest possible ice crystal. These ice crystals will always be at their greatest number and smallest size the moment they are extracted from the machine – they can never get smaller.

In other words, where the type of sweeteners we choose will determine the freeze point depression and overall sweetness of the ice cream, the sum total of those sweeteners will determine how much of the water will turn to ice. Easy. Also interesting to consider is the idea of freeze concentration: as a solution freezes, only pure water crystallizes in to ice, which means the concentration of solutes in the remaining unfrozen water increases, which also means that the freezing point of that water continues to drop as more water turns into ice. Thus, even at a temperature of about 3˚F/-16° C – below the typical serving temperature of ice cream – only about 72% of the total water in a base mix is frozen as ice. The rest remains unfrozen as a very concentrated sugar solution.

And then we turn to keeping those ice crystals as small as possible. It’s all about speed and temperature. A high end batch freezer that can process ice cream in a few minutes will make better ice cream than lower end methods that may take much longer to freeze. It’s a classic example of getting what you pay for. Rather than using visual clues to determine when the ice cream is ‘done’, I typically spin my ice creams to a temperature of about -5˚C/23˚F, (at this point only half the water in the mix has frozen) and transfer to a blast freezer to fully ‘harden’ the ice cream. From here it makes sense that as the ice cream is exposed to increasingly higher temperatures some of that frozen water will melt, forming increasingly larger crystals if and when the temperature drops again. This is usually referred to as thermal shock. Related but slightly different, is accretion, the fusing of large ice crystals stored at higher temperatures over time. For example, the acceptable ‘shelf-life’ – texturally speaking – of ice cream stored at -4˚F/-20˚C may be up to two weeks, but increase the storage temperature to 5˚F/-15˚C and that shelf-life dramatically drops to one or two days.

Spin your ice cream as quickly as possible and store it as cold as possible.


5. Stabilizers and emulsifiers do different things. Stabilizers collectively refer to a category of additives  – most often polysaccharide hydrocolloids – that act upon the water phase alone. Technically speaking, stabilizers do not interact with or directly influence emulsions of fat and water.


Stabilizers are responsible for adding viscosity to the unfrozen portion of the water contributing to overall mouth feel, and enhancing  the ability of the base mix to hold air during the freezing process. Binding water stabilizes it, so that it cannot migrate within the frozen product. Without the stabilizers, the ice cream would become coarse and icy very quickly due to the migration of this free water and the growth of existing ice crystals. Stabilizers improve (slow) melt down and help to prevent thermal shock.

 Emulsifiers are a group of compounds in ice cream which aid in developing the appropriate fat structure necessary for the smooth eating and good meltdown characteristics desired in ice cream. Milk proteins present act as initial emulsifiers and give the fat its needed stability. Supplemental emulsifiers are added to ice cream to actually reduce the stability of this fat emulsion by replacing proteins on the fat surface, leading to a thinner membrane more prone to coalescence during whipping.

Emulsifiers are characterized by having a molecular structure which allows part of the molecule to be readily ‘anchored’ in water, and another part of the molecule to be more readily ‘attached’ to fats: hydrophilic and lipophilic. When we use egg yolks in an ice cream base, the 10% lecithin that they contain performs this function to some degree. Common emulsifier additives used in commercial stabilizer blends include mono- and di-glycerides or polysorbate 80.

When the mix is subjected to the whipping action of the batch freezer, the fat emulsion begins to partially break down and the fat particles begin to destabilize. As previously mentioned, the air bubbles which are being beaten into the mix are stabilized by this partially coalesced ‘scaffolding’ of fat. If emulsifiers were not added, the fat globules would have a ‘chaotic’ structure, resistant to coalescing, resulting in a weaker structure and less desirable texture. Interestingly, as the milk fat content of the base increases, the need for a stabilizer blend decreases.

Attached here is my standard formula that, after years of tweaking, I often use to build from as I plug in other ingredients. But starting from zero on your own is well worth the exercise.

Vanilla Ice Cream – Opusculum

There are numerous resources from which I’ve gathered this information over time, and from which the subject can be further explored. A few of my favorites include:

Ice Cream, by H Douglas Goff , Richard W Hartel

Frozen Desserts, Francisco Migoya


i Segreti del Gelato, Angelo Corvitto

University of Guelph (Ontario), Dairy Education Series

Also of interest, from my ice cream guru, Cesar Vega:

The Kitchen as Laboratory: Reflections on the Science of Food and Cooking, by César Vega, Job Ubbink and Erik van der Linden




To learn more detail on the underlying science of ice cream, I’ll be teaching a short form class on the subject on July 30th 2014 at the Institute of Culinary Education in NYC!


11 thoughts on “Ice Cream

  1. Bra f’cken VO. Absolutely Outstanding. This is the Licence to push the boundaries with confidence and science on my side.

    1. Jeremy,

      I would anticipate an eventual posting solely on the subject of stabilizer blends, as the subject can get quite deep. It’s true that not all blends are created equal, but as a general rule anything handy would work here. I am currently using ‘Cremodan 30′ – which is comprised of Mono and Diglycerides, Sodium Alginate, Locust Bean Gum, Carrageenan, and Guar Gum.

  2. Hey chef,
    I’m loving the blog! I got two questions about your recipe. The first one is why do you add the cream after your anglaise is cooked and the other is if the ice cream would be ready to serve after being pacojetized!
    Thank you so much! Looking forward for the next posts.

    1. Ppezo,

      Great question, and one that there isn’t necessarily a scientific reason for – I usually add the cream at the end to begin the cooling process. It’s a habit I picked up long ago and just stuck with it! On a small scale it’s fine, but on a commercial scale it would ideally be added prior to pasteurizing the mix. As for the PacoJet, virtually all of my recipes work well, but they do generally require some degree of hardening time after processing. In most kitchens that use a PacoJet – in my observation – tend to process an hour or so before service, allowing the ice cream or sorbet to firm back up in the freezer. It depends a lot, of course, on the formulation and freezer temperatures as well…

  3. Thanks so much for this – I made the vanilla already and it was fantastic. My question, if I may… would you suggest the same 6g amount of Cremodan for any similar quantity of ice cream? Or would the amount vary according to the kind of ice cream made? I’m doing a parsley ice cream and doesn’t call for it but as a pastry/ice cream neophyte I figure the stabilizer might “cover” for other small mistakes that could affect consistency. Many thanks – Marc Leandro

    1. Marc,

      If you look at the total yield of the vanilla ice cream (just shy of 2000g), and then the amount of stabilizer blend (8g), one could say that it represents 0.4% of the total mix – in general that’s a good place to start with such things. Not all stabilizer blends are created equal, however. If I had a stabilizer that contained just hydrocolloids (such as guar and locust bean gums) and no emulsifier (such as mono and di-glycerides), I might add 0.2% of the stabilizer and 0.2% of the emulsifier separately, for a total of 0.4%. If the formula changes and the fat content goes up, one would begin to decrease the total stabilizer.

      Hope that helps!

  4. I was doing some calculations on my own mix recently, when I realized that the whole milk I get in the US is typically 3.25% fat, not 3.6% fat. Good to keep in mind if you’re targeting a specific fat percentage in your final mix.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s