Introduction to Flavor

Please see updates on the new Underbelly Blog

Easier...

Earlier in this blog series I suggested that good texture was easier for industrial ice cream manufacturers, but that good flavor was easier for the home or pastry cook. This is because some of the tools that promote the finest texture—high-pressure homogenizers, high-end ice cream and gelato machines that can freeze the mix in mere minutes, blast freezers / hardening cabinets—are out of reach for many small shops. 

Meanwhile, good flavor depends, in part, on generous amounts of good flavor ingredients, which are almost always expensive. Premium vanilla beans (around $400 a pound in 2018) are a minor splurge for a pastry chef in a michelin-starred restaurant, or a home cook who doesn’t need to turn a profit. Might as well use the good stuff, and throw in an extra bean if it suits your whims. But for a manufacturer making ice cream by the boxcar, such price premiums are a killer—especially in the context of retail markups and price fluctuations, either seasonal or unpredictable (caused by draught, blight, politics, war, etc.).

But Not Easy

While it's not so challenging to get better flavor than Haagen Dazs or Ben & Jerry, that's not a particularly high standard. Good flavor is easy. Great flavor is hard.

This blog has been silent for many months, in part, because I’ve been researching and experimenting with flavor, and discovering along the way that every answer spawns several new questions. Surprisingly, there’s very little high-quality published science on getting the best flavors into ice cream. The breadth of the topic doesn’t help. A spice like vanilla behaves very differently from an herb like thyme, which behaves differently still from an herb like peppermint. None of these are at all like coffee, and none create the problems you’ll find with ingredients that add water, like fruits, or ones that add fat, like chocolate or nuts. Every flavor is a new adventure, demanding different science, presenting different complications.

When I worked at an ice cream shop way back in the 20th century, our approach to flavors was mostly formulaic. We'd usually use the most concentrated flavor ingredient possible, in the form of commercial extracts, dry spice powders, and cocoa powder. With problem ingredients (fruit, or our house-made, water-based coffee extract) we’d compensate for the extra water by throwing in extra cream. Mix it together, spin it, call it done. This is how most ice cream shops do it. The approach makes decent commercial ice cream, and it lets you use the same base for all your flavors, and gives you a process you can teach someone in a couple of days. But it’s not the path to the best ice cream, and it’s not what we're going to be discussing in the next few posts. 

Taxonomy

There are a few basic ways to get flavors into ice cream. I've divided them into the following categories:

Simple recipes (Blend the flavor ingredient into the base)

-Extracts or essences

-Matcha

-Low-bulk spices (cardamom, star anise, peppercorn

Modified recipes (Blend in the flavor ingredient. Compensate by adjusting total fats, sugars, milk solids, etc.)

-chocolate

-nuts

-fruits

-spirits 

-alternative sweeteners (honey, molasses, etc.)

Infusions Brew flavor into liquid ingredients at specific temperature and time, strain out)

-herbs

-high-bulk spices (vanilla)

-aromatic vegetables (ginger)

-coffee

Recipies with additional processes

-caramelizing sugar

-reducing wine or spirits

-toasting nuts

-browning butter

Compound recipes (Combining two or more of the above)

-dark sugars ice cream (caramelizing sugar, modifying base recipe)

-brown butter cognac ice cream (browning milk solids, reducing spirits, modifying base recipe)

-bourbon smash ice cream (infusion, modifying base recipe)

-coffee (infusion, modifying base recipe)

What is Flavor?

Flavor describes a complex set of physiological-psychological perceptions, based on taste sensations from the tongue, olafactory sensations from the nose, and a whole range of expectations and biases created by what we see, hear, believe, and remember. The science of flavor is young and rapidly expanding. The lore goes back about as far as recorded history.

In this series we're mostly going to be talking about the tongue and nose. The psychological / perceptual sciences are a bit above my pay grade (I happen to get paid nothing for this ... thank you for asking). 

You may have been tought that the tongue perceives just four tastes: sweet, salty, sour, bitter. This is pretty close, but we now believe that in addtion to these four, there is also at least one type of savoriness (umami), which describes our response to glutamic or aspartic acids in meat, seaweed, and aged cheeses; there is a taste for alkalinity (think of soap); there are possibly many distinct types of bitterness; and there may be other distinct tastes associated with fat, metals, and calcium.

There are also sensations that are similar to tastes but that are actually physical sensations unrelated to the tongue's taste receptors: pungency (the chemical burn from hot chilis); astringency (the grainy and puckery sensation from tannins in red wine, black tea, and rhubarb); coolness (triggered by mints, ethanol, camphor, and some sugar alcohols); numbness (triggered by some Sichuan chilis and by cloves). 

This describes around a dozen sensations in the mouth. The nose, however, knows no such humble limits. We can detect thousands—possibly tens of thousands—distinct aromatic compounds. And we're discovering new ones all the time. Here's a simplified chart breaking down organic aromatic compounds by family and molecule size. This includes a range of them from herbs to jet fuel, but is far, far from exhaustive:

Table of Aromatic Compounds

Some of this is academic; some may be useful when we have to solve problems. For example, you may find that an herb contains aromatic compounds that you like and ones that you don't. If they are soluble at different temperatures, this will help you decide the temperature of your infusion. Some compounds may be much more soluble in fat than in water. If you want those compounds, then infuse into the cream. If you don't, then infuse into the milk. And so on.

But let's start with something simple. Here's a recipe from the first group, aptly titled Simple Recipes, because you don't have to make any adjustments to the base recipe, and you don't have to use any specieal technique (besides turning the blender on):

Simple Recipe Example: Green Tea Ice Cream  

-Ingredients to make 1 liter of Standard Base

-12g matcha powder 

-Combine matcha with the other dry ingredients and make as usual. Easy peasy.

Notes

Matcha is powder ground from whole green tea leaves. In Japan it's whisked into hot water and enjoyed as a full-bodied tea. Since the powder is expensive, matcha is considered a more special drink than ordinary brewed green tea. Matcha comes in several grades, from "ceremonial" to "premium" to "café" to "culinary" to "classic." The highest grades are distinguished by their delicacy and subtlety, and so are almost always consumed straight without any milk or sweetener. The delicacy does not hold up to the dairy and sugar in ice cream, so it makes the most sense to use a good quality, fresh, culinary grade in this or other green tea ice creams.

For a very intense matcha flavor, increase the quantity to 18g powder. If using 18g and you find you need to decrease bitterness, add  0.4g citric acid and 1.3g additional salt (increase salt in recipe to 2g total)

This is a standard flavor balancing tactic, not just in ice cream but in all cooking, sweet and savory. Some universal guidelines:

-To counteract bitterness, increase salt or acid
-To counteract acid, increase fat or bitterness
-To counteract sweetness, increase acid or bitterness
-Salt will bring forward mid-range, warm, savory flavors
-Metalic flavors are often the product of imbalanced acids

Next post: Coffee Ice Cream (aka Simple Cooking Made Complicated)

For further reading:

Compoundchem.com's food and flavor chemistry pages

James Kennedy Monash's aromatic chemistry site

Harold McGee: On Food and Cooking

Modernist Cuisine

The Art of Flavor

How Does Our Sense of Taste Work? [Informed Health Online, via PubMed]

Part 1 of this series: Introduction
Part 2 of this series: Components
Part 3 of this series: How to Build a Recipe
Part 4 of this series: Basic Recipe Examples
Part 5 of this series: Techniques
Part 6 of this series: Sugars

Part 7 of this series: Stabilizers
Part 8 of this series: Emulsifiers
Part 9 of this series: Booze

Part 10 of this series: Solids, Water, Ice

Part 11 of this series: Introduction to Flavor
Part 12 of this series: Ice Cream Flavor: Coffee

Part 13 of this series: Coffee Ice Cream Addendum: Origin Notes and Minutiae

Part 14 of this series: Chocolate Ice Cream
Part 15 of this series: Chocolate Ice Cream Addendum

Ice Cream: Solids, Water, Ice

Please see updates on the new Underbelly Blog

This post is addendum to the post on How To Build a Recipe, and the post on Sugars. I want to clarify the importance of solids—which is really a reflection of the importance of water. These are key ideas—if you master them, you will be well on your way to texture Ninjahood.

Scanning electron micrograph of ice crystals

We’ve discussed how ice cream is made up of three physical systems: an emulsion (fats suspended in water); a foam (air dispersed in a solid fat network); and a sol (solid water dispersed in liquid water).

Here we’re going to look at that sol—the interactions of water and ice. And we’re going to look at the effects of all the nonfat solids that are dissolved or suspended in the water.

It’s important—and instructive—that the ice is mostly pure water. This is the case because freezing acts as a purifying process; when ice crystals form, they expel dissolved and suspended solids. The result, besides much lower concentrations of stuff in the ice, is an increased concentration of stuff in the liquid portion of the water. 

This phenomenon is called Fractional Freezing. It creates an interesting system in which there’s no single freezing temperature for ice cream, or for any water-based solution. Instead there’s a temperature range between the extreme points where none of the water is frozen and where all of it is frozen. Between these points, you get a mix. The colder the temperature, the greater the proportion of frozen water, and the higher the concentration of solutes in the liquid water.

Fractional freezing works because of the colligative properties of water regarding freezing point depression.* The stronger a solution, the lower the freezing point. So when a bit of a solution freezes, strengthening the concentration of the remaining liquid portion, that liquid’s freezing point drops. And so on. This process is continuous and self-regulating. And pretty cool. 

So What? 

We need ice to make ice cream, but we don’t want too much. With too much ice, you get a popsicle. In order to have a lower proportion of ice, we need a higher proportion of other stuff. 

Some of that other stuff is fat. But we likewise don’t want too much, or the ice cream will be too rich and the flavors too muted. Some of that other stuff is air, but we really don’t want too much of that, or the ice cream will be too fluffy and insubstantial. 

The remainder is liquid water, and dissolved or dispersed nonfat solids. These two are intimately related, because 

1) the more nonfat solids, the lower the percentage of total water, and

2) the more nonfat solids, the greater the portion of that water that stays liquid

So, in general, solids are good. Solids with low molecular weights** depress the freezing point the most, while all solids displace some of the water. 

We especially like milk solids—specifically the nonfat portion—because they effectively concentrates the milk. We get more of the functional qualities of milk, like emulsification, freezing point depression, and improved body, but without added water. And milk tastes good, the way ice cream should.

Sometimes we get solids from the flavor ingredients: chocolate and cocoa, fruit pulp, nut butters, matcha powder, coffee solids, etc.

Remember from the How to Build a Recipe post, we typically aim for the following levels of solids in a well-balanced recipe:

Nonfat Milk Solids: 10-12%. Or higher for low-fat ice cream.

(everything in milk besides the water and fat)

Total Nonfat Solids: 22–25% 

(everything in the ice cream besides water, alcohol, and fats)

Total Solids: 37–42%

(total nonfat solids plus total fat. everything besides water and alcohol)

If solids levels are too low, ice cream can lack body, freeze too hard, and may have textural problems like iciness. If solids levels are too high, ice cream can be excessively chewy or cakey. If the milk solids specifically are too high, you can get grainy textures. 

Just remember that by managing the solids, you’re managing the water.

Periodic Table of the Elements. The number int the top left of each box is that element's atomic mass (when a molecule is made of many atoms, we add the individual atomic masses together to find the Molecular Mass). Lower number = smaller molecule = more molecules at a given weight = greater freezing point depression.

*Colligative properties are based on concentration of dissolved solutes in the water, and on the size of their molecules, but not on any special chemical qualities of those solutes. Two equally concentrated solutions of two chemicals that have the same molecular weight will depress the freezing point of water equally. This is handy to know; it gives us just two values to consider when adjusting ice cream’s hardness.

**Molecular Wuh?? The molecular weight (also called molecular mass) of substance is a number that tells us the size and mass of its individual molecules. Molecules are made of atoms, and atoms are made of protons, neutrons, and electrons. The protons and neutrons have equal mass, while the electrons are so tiny and insubstantial that we can ignore them. So the molecular weight is the total number of protons and neutrons. 

A high molecular weight means the individual molecules are relatively massive. This means that a gram of something with a high molecular weight contains fewer molecules than something with a low molecular weight. And this, in turn, is why substances with a low molecular weight are more powerful at depressing the freezing point: for a given mass, there are more individual molecules in dispersion in the water, exerting their colligative influence. 

Next post: infusing flavor (yes, we’re eventually getting to the good stuff)

Part 1 of this series: Introduction

Part 2 of this series: Components

Part 3 of this series: How to Build a Recipe

Part 4 of this series: Basic Recipe Examples

Part 5 of this series: Techniques

Part 6 of this series: Sugars

Part 7 of this series: Stabilizers
Part 8 of this series: Emulsifiers
Part 9 of this series: Booze
Part 10 of this series: Solids, Water, Ice
Part 11 of this series: Introduction to Flavor
Part 12 of this series: Ice Cream Flavor: Coffee

Part 13 of this series: Coffee Ice Cream Addendum: Origin Notes and Minutiae
Part 14 of this series: Chocolate Ice Cream

Part 15 of this series: Chocolate Ice Cream Addendum

Booze Flavored Ice Cream

Please see updates on the new Underbelly blog.

The holidays are upon us—good cheer, warm cockles, and dining while captive audience to your white supremacist uncle. I’m not promising boozy ice cream will take the edge off. But it can’t hurt.

Spirit flavors are well trodden territory. If you can find it in a bottle, or make it in a still out back, someone’s probably made ice cream with it. I’ve made—or tried or seen or heard rumors of—ice creams made from every cordial and liqueur, malted and distilled grain, wine from every fruit (fortified in every kind of barrel), stout and ale and lambic and mead, brandy, sake, kefir, and all forms of home-grown something steeped in home-brewed something else.

The new frontiers seem to be about reimagining classic (or transgressive) cocktails as ice cream flavors. More on this later.

Just about the only booze I’ll recommend against is vodka, or other “neutral” spirits like Everclear. These are neutral in the sense that most of the character of the original fermented ingredient has been distilled and filtered out, and no new flavors have been added. These liquors all taste different straight up, but the subtleties are masked by any cocktail, and certainly by anything full of sugar and cream. All they add to ice cream is problems.

And alcohol indeed presents some problems. As a tiny molecule with a low molecular weight, ethanol (the alcohol that we choose to drink, because it’s non-toxic, more or less) has high powers of freezing point depression. 

(Please see the post on Sugars for a review freezing point depression in ice cream). 

By weight, alcohol depresses the freezing point more than any sugar (over seven times more than table sugar), or any other ingredient we'll likely use. This is why liquor-flavored ice creams are often soft, and often melt almost instantly on the plate.

We handle this by reducing other ingredients that have high freezing point depression. Additionally, we may have to limit the amount of alcohol. 

You can limit the alcohol either by reducing the quantity of spirits, or by simmering a portion of it to remove much of the alcohol, while concentrating the less volatile flavors. I find the latter method helpful with milder flavored liquors, like Cognac, or with alcoholically weaker ones, like wine and beer (which need water, rather than alcohol, boiled off). In many other cases, the boozy ingredient has more than enough flavor even in modest quantities, so the amount that tastes good presents a manageable amount of alcohol.

In all cases, it's helpful to reduce other ingredients with high freezing point depression. The first step is to simply eliminate the dextrose. It has double the depression of table sugar (sucrose), and is about a third less sweet. You may wish to increase the table sugar a bit to compensate. If the spirit you're using is sweet, this may be unnecessary. Keep an open mind to making your ice creams much less sweet than what’s typical; you may find that they just taste better all around—less cloying, with more dimensions and subtleties to the flavors.

You may wish to reduce the invert syrup a bit. And it’s often helpful to slightly increase the stabilizers. I find that just increasing the locust bean gum 20-25% is all that’s necessary, and that it helps without causing any textural weirdness.

The wee ethanol molecule. Molecular weight 46g/mol.
Looks like a Jeff Koons sculpture, but bites.

How Much Liquor?

A reasonable upper limit is 30g pure alcohol per 1000g mix. (3%)

This works out to: 

     -60g 100 proof booze

     -75g 80 proof booze

     -150g 40 proof booze / 20% fortified wine

     -200g 30 proof booze

     -230g 13% wine

These are suggested maximum quantities. In many cases, you’ll choose to use less.

With most hard liquors: 

75g is very strong.

45–60g gives nice flavor and a kick

With most liqueur and fortified wine:

45–60g is often plenty

If you’re using liquors that have lighter flavors relative to the alcohol content (cognac, sake, grappa) you may wish to to use more, but to simmer off the alcohol from a portion of it. For example, with cognac ice cream, I simmer 115g of cognac until it’s reduced by about 2/3. At this point most of the alcohol is gone. I add this along with 30g unreduced cognac. 

For weaker spirits, like wine, the problem isn't the added alcohol as much as the added water. You can reduce a portion of the wine with heat, but not most of it; heat will kill the aromatic flavors. I find it best to replace a portion of the milk with the wine—then add a commensurate amount of nonfat dry milk, to make up for the lost solids.

Ultra boozy ice cream is becoming a trend, with companies like Tipsy Scoop churning out 5% alcohol concoctions. When you get into this range, it's more about a sweet frozen cocktail than about a liquor-flavored ice cream. Getting you buzzed takes a higher priority than texture or flavor. And you have to start carding people. If you want to head in this direction, just follow the instruction here, but take everything farther.

Liquor Quality

My general rule is to use liquor that you’d happily use in a sweet, tropical cocktail. Not necessarily the finest (you’re not going to taste all the subtleties through the dairy and sugar). But go for full flavors, and nothing that tastes bad.

For example, with cognac, I use an affordable VSOP grade that I’d be happy to drink, but that I wouldn’t be tempted to ruminate over with my pompous friends. I like Gosling's rum—bold, full-flavored, if not the most complex. Save the really good / delicate stuff for drinking straight.

Of course, some varieties of liquor are proprietary, and if you want their flavor profile you have to buy their brand. Like Campari, or Grand Marnier (see below). Some of these are going to cost you.

Summary of basic compensations:

-eliminate dextrose.

-Increase sweetness only as much as necessary by increasing sucrose

-If booze is a liqueur with added sugar, take into consideration. You may actually be able to reduce sucrose or keep it the same. 

-keep around 10% trimoline for its textural properties

-keep nonfat milk solids around 12%

-increase Locust Bean Gum up to 25%, depending on total quantity of added alcohol and water.

A Guide for Wine Ice Creams (this is a work in progress)

For 1000g base:

-720g wine (2 cups) divided in half

-Reduce one portion of the wine  by 2/3, concentrating and removing most alcohol

 (total volume will now be 1-1/3 cups. mass slightly higher than 360g)

-160g milk (2/3 cup)

-240 g cream (1 cup)—standard proportions for lower fat ice cream

-add additional 30g dry milk beyond your usal amount, to compensate for reduced fresh milk

eliminate dextrose, increase sucrose 20g

-increase locust bean gum 25%

Beer flavors (consider stouts and lambics!) would probably benefit from similar treatment (but beware IPAs and other hoppy brews—the bitterness does not boil off and will concentrate if you make a reduction).

If you experiment with this template, please let me know how it goes.

Thoughts on Cocktail Flavors

In some cases you can stick pretty close to the cocktail recipe. In others, you might find that some ingredients add too much alcohol relative to their flavor.

Gin, for example. High alcohol content, subtle flavors. And many of the flavors are aromatic, so you risk losing them if you reduce over heat. Assuming you don’t have a rotary evaporator, one solution is to add your own flavors through infusion. Think about what’s in there: juniper berries, angelica root, corriander, cardamum, citrus. Try infusing a lot of these into a small quantity of gin, to make an intense supergin that will hold up to dilution. Or infuse these flavors into the cream.

Generally, cocktails that are already sweet (anything from variations on the Old Fashioned to Tiki drinks) will work better than dry cocktails. I don’t see much potential for Martini ice cream (but maybe as a sorbet …)

Obviously, leave out the simple syrup. And avoid vodka cocktails (see above. And they're not real cocktails). 

If you’re making your own creations, remember that alcohol is a great solvent. Many flavors will infuse into the booze itself more readily than they will into the dairy. And alcohol is great for macerating dried fruits. The best rum raisin ice creams are alwways made with raisins soaked in a portion of the rum. Many of us got our first childhood buzz off of this flavor, and still remember it fondly (or don’t).

Go to good cocktail bars. Figure out what the mixologists are doing. The best of these guys are essentially chefs. Notice how they layer and balance the flavors. Notice how in some cases all the individual flavors remain identifiable; in other cases, an entirely new flavor is created by the combinations. Try to figure out what's going on. Look at the structure of classic cocktails (and the often similar modern ones). Try to accomplish the same effects in the flavors you create.

Sample Recipe (single spirit)

Grand Marnier Ice Cream

If you haven't had the pleasure, Grand Marnier is a delicious liqueur made by infusing orange and spices (especially cloves) into cognac. Most other orange liqueurs taste like Kool Aid in comparison. This recipe is based on the standard ice cream base described in a previous post, with the modifications outlined above. I’ve replaced a portion of the trimoline with chestnut honey; this provides some dark, bitter, and savory notes that help support the liqueur's flavors.

Makes 1 liter. Written for an immersion circulator and blender, but should be easily adaptable to other techniques.

360g whole milk (3.3% fat)*

7g chestnut honey (replace w/ invert syrup if you don’t have it)

5g invert syrup

70g granulated sugar 

55g nonfat dry milk*

0.7g salt

1.0g locust bean gum (tested with TIC Gums POR/A, soluble at 74°C)

0.4g guar gum

0.2g lambda carrageenan

2 large egg yolks (36g)

360g  heavy cream (36% fat)* 

60g Grand Marnier (40% alcohol = 24g. 14g sugar)

Optional: 1/4 cup stuff to mix in: candied ginger, candied orange peel, shaved chocolate, white raisisns macerated in a portion of the grand marnier, etc.

*Use the best quality milk and cream you can get. Nothing ultrapasteurized. Low-temperature pasteurized is ideal. Homogenized products will give best texture. Avoid cream with added stabilizer (unknown variables). Dry milk must be 100% skim milk solids, processed without high heat. There should be no off odors either when it's dry or when it's mixed. Store sealed in freezer.

-set immersion circulator to 75°C / 167°F

-thoroughly mix powdered ingredients.

-separate eggs and set yolks aside. freeze whites for other use.

-measure milk, honey, and invert syrup into blender.

-set blender speed to create a vortex; add powdered ingredients. cover and blend on high for 30 seconds to disperse the stabilizers. 

-add cream, yolks, Grand Marnier.

-briefly blend again

-pour mixture into 1gal ziplock bag.

-add weight (recommended, to keep bag from floating) and evacuate the air.

-cook in water bath for 45 minutes to set custard, hydrate stabilizers, denature milk proteins.

-gently agitate bag after 5 and 15 minutes. if you see air accumulated in the bag after 15 minutes, release it, and carefully reseal bag.

-mix will be pasteurized (pasteurization time after reaching this temperature is under 2 minutes).

-remove bag from water bath. open and pour hot mix into clean blender container (or a square container if using a homogenizer or stick blender). remove weight (with tongs). use bag to squeegee off any mix. temporarily seal bag and keep handy. 

-blend on highest speed for 30 seconds to homogenize.

-pour mix back into ziplock bag.

-chill bag in ice water bath (use ice bath to evacuate the air when sealing bag). carefully agitate to cool. Try to cool to refrigerator temperature. 

-refrigerate at least 8 hours, below 38°F / 3°C to age mix / pre-crystalize fat.

******

-pour into ice cream machine: snip off bottom corner of bag, and squeeze out mix as if using a pastry bag. or squeeze out into an intermediate container that’s easy to pour from.

-spin in the ice cream maker. With a mulitispeed machine, use a slow setting (this recipe works best with a low overrun). Ideal drawing temperature is 23°F / -5°C.

-evaluate when surface texture of ice cream first looks dry. if it needs more overrun, continue on higher speed. if it needs to cool more, continue on lower speed. 

-optional: mix in the mix-ins.

-harden for several hours (preferably overnight) in a cold freezer. freezer should be set to -5°F / -20°C or lower. Ice cream will have to warm up several degrees before serving. 20 to 30 minutes in the fridge works well. Ideal serving temperature is 6 to 10° F / -14 to -12°C.

Sample Recipe 2 (cocktail)

Bourbon Smash Ice Cream

The Bourbon Smash is one of my favorite summer cocktails. It would be one of my favorite winter cocktails, too, but it can be hard to find fresh mint outside the summer months. I just made a batch of this from the last surviving mint from my garden, which wasn't in the best shape (there's an inch of snow on the ground, and the leaves are mostly turning black). Can you come up with a better version for winter? How about less citrus, and replacing the mint with cardamom or cloves?

This recipe illustrates a few techniques beyond the basics: zesting a lemon and incorporating zest with the dry ingredients (hint: a fine Microplane is probably the most efficient tool for this); incorporating fresh citrus juice (strain, and add after the cream and eggs have been incorporated, to keep the milk proteins from curdling); and herbal infusion (done at low temperature—important with mint, to protect its delicate flavor) and the addition of essential oil, to compensate for the difficulty of infusing mint's full flavor.

Makes 1 liter. Written for an immersion circulator and blender, but should be easily adaptable to other techniques.
[edited 6-2018. Reduced mint essential oil from 0.5g to 0.1g. For reasons that will be obvious to anyone who tries it either way]

340g whole milk (3.3% fat)*
12g invert syrup

85g granulated sugar
55g nonfat dry milk*
4g (aprox.) lemon zest (from 1 small lemon)

1g salt
0.8g locust bean gum. increase to 1.0g if you get icy textures (tested with TIC Gums POR/A, soluble at 74°C)
0.4g guar gum
0.2g lambda carrageenan

2 large egg yolks (36g)
360g  heavy cream (36% fat)*

24g (aprox) lemon juice (from 1 small lemon)
65g bourbon (43% alcohol = 28g)
2g Angostura bitters

12g very fresh mint leaves

0.1g mint essential oil

*Use the best quality milk and cream you can get. Avoid anything ultrapasteurized. Low-temperature pasteurized is ideal. Homogenized products will give best texture. Avoid cream with added stabilizer (unknown variables). Dry milk should be 100% skim milk solids, processed without high heat. There should be no off odors either when it's dry or when it's mixed. Store sealed in freezer.

-set immersion circulator to 75°C / 167°F

-thoroughly mix powdered ingredients.

-separate eggs and set yolks aside. freeze whites for other use.

-measure and combine lemon juice, bourbon, and bitters. set aside.

-measure milk and trimoline into blender.

-set blender speed to create a vortex; add powdered ingredients. cover and blend on high for 30 seconds to disperse the stabilizers. 

-add cream and yolks. blend briefly.

-add bourbon-lemon-bitters mixture. briefly blend again.

-pour mixture into 1gal ziplock bag.

-add weight (recommended, to keep bag from floating) and evacuate the air.

-cook in water bath for 45 minutes to set custard, hydrate stabilizers, denature milk proteins.

-gently agitate bag after 5 and 15 minutes. if you see air accumulated in the bag after 15 minutes, release it, and carefully reseal bag.

-mix will be pasteurized (pasteurization time after reaching this temperature is under 2 minutes).

- while mix is cooking, rinse mint leaves and trim off browned parts and most of the stems.

-remove bag from water bath. pour mix into clean blender container (or a square container if using a homogenizer or stick blender).

-add essential oil

-blend on highest speed for 30 seconds to homogenize.

-put mint leaves in ziploc bag.

-pour mix back into bag.

-chill bag in ice water bath (use ice bath to evacuate the air when sealing bag). carefully agitate to cool. Try to cool to refrigerator temperature. 

-refrigerate at least 8 hours, below 38°F / 3°C to age mix / pre-crystalize fat.

******

-strain out mint: snip off bottom corner of bag, and squeeze out mix as if using a pastry bag, into a strainer placed over a pourable container

-pour into ice cream maker.

-spin in the ice cream maker. With a mulitispeed machine, use a slow setting (this recipe works best with a low overrun). Ideal drawing temperature is 23°F / -5°C.

-evaluate when surface texture of ice cream first looks dry. if it needs more overrun, continue on higher speed. if it needs to cool more, continue on lower speed. 

-harden for several hours (preferably overnight) in a cold freezer. freezer should be set to -5°F / -20°C or lower. Ice cream will have to warm up several degrees before serving. 20 to 30 minutes in the fridge works well. Ideal serving temperature is 6 to 10° F / -14 to -12°C.

Next post: introduction to Flavor

Part 1 of this series: Introduction

Part 2 of this series: Components

Part 3 of this series: How to Build a Recipe

Part 4 of this series: Basic Recipe Examples

Part 5 of this series: Techniques

Part 6 of this series: Sugars

Part 7 of this series: Stabilizers
Part 8 of this series: Emulsifiers
Part 9 of this series: Booze
Part 11 of this series: Introduction to Flavor
Part 12 of this series: Ice Cream Flavor: Coffee

Part 13 of this series: Coffee Ice Cream Addendum: Origin Notes and Minutiae

Part 14 of this series: Chocolate Ice Cream
Part 15 of this series: Chocolate Ice Cream Addendum

Ice Cream Emulsifiers

Please see updates on the Underbelly Blog

This has so far been the most challenging post to research and write; there’s a lot of arcane and incomplete knowledge on the topic, much of it of questionable relevance to the mission of artisanal ice cream making. If you favor recipes with one or more egg yolks per liter of base, and don’t want extra homework, you can skip this one.

If you’re interested in making egg-free ice creams with the best possible texture, or are just a glutton for minutia, read on.

Electron micrograph of a milk fat globule. Here's where the action is.

Basics

An emulsifier is a two-faced molecule that binds to water on one side, oil on the other. You’ll see one side labelled hydrophilic, the other labelled either hydrophobic or oleophilic—different jargon, same idea.

The adage that oil and water don’t mix still holds true after all these years. This would be a problem for much of the world, especially the plant and animal parts, if it weren’t for emulsifiers helping to hold things together.

Like, for instance, milk. Milk is a colloid made from a polar chemical (water) and non-polar chemical (milk fat), which cannot mix. Technically speaking, they are not miscible. But they can form an emulsion—a more complex arrangement in which immiscible ingredients coexist, with some degree of stability.

Because it’s an emulsion, Milk is creamy, opaque, and homogenous. The fat molecules are divided into many microscopically small globules, which are surrounded by emulsifying chemicals that keep them from glomming together and rising to the top. These emulsifiers include fatty chemicals called phospholipids, and proteins called wheys and caseins.

We’re mostly interested in the caseins, which make up about 80% of the milk proteins. They gather together in clusters called micelles, which expose their hydrophilic and hydrophobic surfaces on different sides, allowing them to bind both water and fat—thereby keeping the milk from separating.1 If you look at milk under an electron microscope, you see big round fat globules, covered in a thin membrane of casein micelles. This casein, with its hydrophilic end facing outwards, keeps the fat globules connected to the water and discourages them from joining each other.

Emulsifiers in Ice Cream

Milk Fat and Casein. Click to enlarge.

Left: fat globule with membrane made of casein micelles

Center: crystalized fat. Crystals protrude, but casein prevents them from bumping into other globules.

Right: crystalized fat partially destabilized by emulsifiers. Most of the protective casein is gone,
so the fat globules are better able to bump into each other and form a larger structure.

Thanks mostly to the casein, milk and cream are already pretty stable emulsions—so why are emulsifiers so important in an ice cream recipe?

The reason is that the milk and cream emulsions are a bit too stable. A big part of making ice cream involves the same process as making whipped cream, which requires destabilizing the emulsion. As long as the emulsion is stable, those fat globules can't join to create a foam network. With traditional methods, you can only whip cream containing over 30-something percent fat. With lower fat percentages, there’s enough water between the fat globules to keep the emulsion stable. Ice cream usually has far too little fat.

So to whip ice cream, you need extra help to partially destabilize the milk fat emulsion. Then the fat globules can partially coalesce—science talk for the fat glomming together just enough to form a stable foam.2

This is what added emulsifiers are for. They weaken the emulsion, by binding to some of the existing emulsifiers (casein micelles) and carrying them away. This allows the protruding fat crystals from from different fat globules to make contact and form a network. The result is ice cream to whip into a foam more easily, and with better texture and stability.

Ice Cream Emulsifying Ingredients

Egg Yolk

Egg yolks may be the most familiar emulsifier in the kitchen; they hold together mayonnaise, hollandaise and béarnaise family sauces, Caesar’s salad dressing, and chocolate marquises.

Yolks are usually cooked to a custard consistency, which also thickens the water portion of the ice cream (as a stabilizer—see the post on Ice Cream Stabilizers). Yolks are occasionally used without taking advantage of their thickening ability, but they must at least be cooked to the point of pasteurization.

Yolks contain a number of proteins and lipids that work as emulsifiers. By far the most important among these is lecithin. For adequate effect on ice cream emulsification, the mix requires 0.5% to 1% egg yolk. This equals 1/3 to 2/3 yolk per liter of mix, so a single yolk is more than adequate. 2 yolks (3 to 4%) are necessary to get significant thickening / stabilization. For the richest French custard-style ice creams, 4 to 6 yolks per liter is more common, with 8 to 10 being close to the outer fringe.

I would consider anything with 10 yolks per liter a frozen custard rather than an ice cream. Custard texture and flavor will dominate, or at least will compete heavily with everything else. 

Non-Egg Emulsifiers

The following are useful primarily if you’re making ice cream without eggs. Egg-free ice creams can be lighter, can have more vibrant flavors and a cleaner finish, and of course will be completely without egg flavors (although ice creams made with as few as two yolks per liter have essentially no egg flavor). Eliminating eggs can be especially useful in ice creams that have a lot of extra fat from the flavor ingredients: chocolate, nut butters, avocado, etc.  With these flavors, eliminating eggs is a way to to avoid the flavor damping and greasiness that can come with too much total fat content.

Lecithin

Lecithins are complex mixtures of fatty components derived either from eggs or soybean oil. They allow egg-free ice creams to be made with the same, all-natural emulsifier used in custard ice creams. Soy lecithin is as effective as egg lecithin and costs less. A single “large” egg yolk contains roughly 1.5g lecithin. An egg-free recipe can be made with 1.5 to 4.5g lecithin per liter (0.15% to 0.45%) Higher quantities may be helpful in recipes with lots of added fat.

It’s possible that in very delicately flavored ice cream, lecithin will have a detectible flavor of its own. I haven’t found this to be the case (I’ve tried straight soy lecithin out of the jar and found its flavor much milder than milk powder). Flavor may vary with the quality grade of the ingredient. Seek a high quality, very refined lecithin that has a mild flavor, and always disperse it thoroughly in a powerful blender.

Buttermilk Powder 

is similar to skim milk powder, but has a greater number of phospholipids that have emulsifying power. There’s some research into using this ingredient in place of traditional emulsifiers. I haven’t seen research on ice cream specifically, so I can’t recommend quantities or techniques. In larger quantities, it will of course taste like buttermilk. Buttermilk flavored ice cream might be good for first experiments.

Polysorbate-80

Is an emulsifier derived from oleic acid and from a form of sorbitan. It works in minute quantities, so it's both inexpensive and completely flavorless. It belongs to a family of emulsifiers called sorbitan esters, which are believed to be the most active at the fat interface of the ice cream, making them the most efficient at destabilizing the fat emulsion.3 

While the health scares you’ll read the internet are imaginary, I can’t find much compelling reason to use polysorbate in artisanal ice cream. Its advantages over lecithin and egg yolks are in cost and and improved whipability. Cost is a minor issue in small volume production, and most high quality ice creams are intentionally dense (low overrun) and so don’t need improved whipping. If you'd like to experiment, standard concentration is between 0.02% and 0.04%.

Mono- and Diglycerides

These are emulsifiers which occur naturally at very small concentrations in seed oils. Commercial products are manufactured, using a glycerolysis reaction between triglycerides and glycerol—components of fats from either plants or animals. This ambiguity concerning their source gets them flagged by some vegans, although there is no chemical difference in the end product. Unlike sorbitan esters, glycerides are believed to be most active at the fat-air interface, helping the ice cream form a finer-textured foam. I have not seen this confirmed by studies.

Because they naturally form together, the mono- and di- versions are usually used together and packaged together. The monoglycerides, however, are thought to be the functional component. These typically make up about 40% of the blend. It's also possibly to use monoglycerides by themselves which are usually isolated to glycerol monostearate (GMS) or occasionally glycerol monooleate. These work in much smaller quantities but are more expensive than the usual mon- / di- mixture.

Versions of mono/diglycerides are sometimes packaged for chefs as “glycerin flakes,” although this is a bit of marketing nonsense, since there’s no glycerin in there. The vendors are fighting the bad reputation of ingredients that are manufactured or hard-to-pronounce. There is plenty of internet hysteria about emulsifiers—most of it dreamed up by charlatans like the Food Babe who deserve your active contempt. 

If you'd like to experiment, standard concentration is comparable to lecithin: between 0.1% and 0.2%.

3-D diagram of a ß-Lactoglobulin whey molecule.
When it denatures (cooks), it unwinds.

Denatured Whey Proteins

Whey proteins make up  about 20% of milk’s protein content, vs. Casein’s 80%. Whey is much less stable than casein when exposed to heat. At temperatures that are typical for pasteurizing ice cream and thickening custard, whey proteins begin to denature—they unfold from their original shape, exposing previously concealed surfaces to the molecules around them.

Another culinary term for using heat to denature proteins: cooking.

When the whey proteins denature, they expose rows of active, hydrophobic (water-hating / oil-loving) molecules to other proteins and fats they encounter. There are several different types of whey proteins in milk, all of which react differently to different temperatures and environmental conditions when they unfold.

In other words, this is complicated. Not only is my own understanding of the science imperfect, but the science itself, with regards to ice cream, is immature.

Here’s what we know: different types of whey protein denature at different temperatures. They partially denature over time above 60°C / 140°F. They don’t fully denature until they reach 90°C / 194°F or higher. When they fully denature, some may form insoluble aggregates that are harmful to the ice cream's texture. 

When they partially denature, they develop different kinds of molecular attractions. Some of them are attracted to the casein micelles. When they bind to the casein micelles, they form stronger bonds than the casein’s original bonds to the fat globules. When enough of this happens, the emulsion partially destabilizes—which is exactly what we use all those other emulsifying ingredients for.

Other denatured whey proteins are attracted to each other. At some temperatures they begin to form aggregates that make a gelatin-like network in the water phase of the ice cream. Which is to say, they behave like a hydrocolloid stabilizer (see the post on Stabilizers). 

What this tells us: we can, in theory, use partially denatured whey proteins as both an emulsifier and a stabilizer. Some industrial manufacturers, like Haagen Dazs, appear to get all of their stabilization from the whey proteins (although they use eggs as emulsifiers). Others, like Jeni’s Splendid, get all of their emulsification from the whey proteins (although they use starch as a stabilizer). In both cases, partially denatured whey proteins probably play a role in the ice cream’s body and texutre.

How much whey do we need? That's unclear, but a recipe that's high on nonfat milk solids is essential. You're not going to get enough from straight milk and cream. I'd suggest adding nonfat dry milk powder to raise the milk solids into the 10% range (which will improve the ice cream for many unrelated reasons as well).

What’s the optimum time and temperature to get these effects? It’s somewhere between 60°C and 90°C, half a minute and 90 minutes. There is no published definitive research on this, and what exists probably expresses ranges with varying characteristics and tradeoffs.

Back when Jeni’s made their egg-free bases in-house, in a big batch processor (as opposed to a continuous processor, which is more common in dairies and manufacturing plants), they cooked it at 75°C for an hour. That's a relatively low temperature and a very long time.

I cook my ice cream at 75°C for 30 minutes, plus 15 minutes for it come to temperature—for a total of 45 minutes in a heated water bath. This works well for my methods—but I’m not counting on the milk proteins for everything. I use gums to stabilize, and a couple of yolks (or some lecithin) to emulsify. I’m only counting on the proteins for some added body and smoothness. 

If you want to rely on the whey completely, you have some experimenting to do. You’ll probably need:

• a formula with a lot of milk solids

• fresh milk and cream that have been low-temperature pasteurized

• or raw milk, and the knowledge required to handle it without killing anyone

• nonfat dry milk that’s been spray dried at low temperatures

And you may need to cook different portions of the whey at different times and temperatures, to optimize them as both stabilizers and emulsifiers.

Ice Cream With No Emulsifiers

It’s possible to make ice cream with just cream, milk, and sugar, skipping the eggs and everything else. The result is called “Philadelphia Style,” although it doesn’t appear to have any roots or any particular following in Philly. This name may be the coinage of a 19th Century Journalist who needed a catch-all for East Coast ice creams that aren’t French. 

This is good news for the citizens of Philadelphia, since the name is rather slanderous. Philly-style ice cream can have vibrant, clean flavors, but it pays for them with a litany of texture defects: it’s icy, it rapidly gets icier, it whips poorly, the foam structure lacks smoothness (it can be grainy) and lacks stability (it deflates). Some people champion this style, and they’re lucky, because it’s easy to make. But I can’t think of other reasons to recommend it.

Closing Thoughts

Egg yolks work really well. If you don't want the flavor of eggs, or the flavor-dulling of heavy custard, you can use just a couple of yolks, or as little as half a yolk. If you really want to banish egg entirely, lecithin works nicely. I make eggless ice cream bases for flavors like chocolate and nut butters, where the flavor ingredients add tons of their own fat. In these cases a bit of lecithin (equal to what's in two or three eggs) and a bit of extra stabilizer takes care of the emulsion perfectly.

If you want to delve into the more efficient manufactured ingredients, there's a world of glycerides and polysorbates to play with. See the ingredient sources listed in the Stabilizers post. If you find any advantages to these ingredients in Artisanal ice cream, I'd like to hear from you.

You can also take advantage of the natural emulsifying power of cooked whey proteins. The whey's already there, and you're already cooking the mix, so why not take advantage? Generally, the emulsifying and stabilizing power of these proteins will be enough to supplement the more conventional ingredients, rather than replace them.

It may be possible to replace the conventional ingredients entirely, although there's only one reason to do so: “label friendliness.” For retail ice cream, the ingredient list has become a marketing document, and therefore a place to play to the whims of an often-uneducated public.

As long as people believe locust bean gum is somehow less natural or less desirable than tapioca starch—even though it’s a superior ingredient—manufacturers will feel pressure to use it. Or to use nothing at all. Denaturing the whey proteins provides a way—even if somewhate ironic and cynical—to use science in the service of pretending your ice cream is as science-free as Grandpa’s. 

Never mind that Grandpa’s ice cream was never all that good. The idea of Grandpa’s ice cream, in all its nostalgic splendor and purity, has created a standard everyone with a retail label aspires to. 

If you’re making ice cream at home or in a pastry kitchen, you’re lucky. You can use the ingredients that make the best product, without concern for labels and mythology. For you, getting some emulsion and texture control from the whey proteins isn’t mandatory—it’s just the cherry on top.

Next Post: Booze!

1Eventually, milk will separate. Fresh from the cow it will separate over time just from gravity (fat is lighter than water) or faster if put in a centrifuge (the way dairies separate cream from skim milk today). Emulsifiers make an emulsion relatively stable, but not perfectly so. We can add stability by thickening the water portion of the emulsion (we call anything that does this a stabilizer) or by breaking the fat globules up into a greater number of smaller ones. This physical process is called homogenization. Most of the milk we buy has been run through a high-pressure homogenizer, which blasts the fat globules into a bajillion nano-sized ones—they’re so minute and plentiful that the emulsion stays stable almost indefinitely.

2Full coalescence would mean whipping the cream into butter, which we’re not trying to do.

3Ice Cream By Robert T. Marshall, H. Douglas Goff, Richard W Hartel, p.85


For further reading:

Oil-in-Water Emulsions Stabilized by Whey Protein—Effects of Heat Treatment and High Pressure Homogenization
Instability and Partial Coalescence in Whippable Dairy Emulsions (abstract)
Studies on heat-induced interactions and gelation of whey proteins
The Effects of Polysorbate 80 on the Fat Emulsion in Ice Cream
Hydrocolloids as Emulsifiers and Emulsion Stabilizers
Ice Cream. Robert T. Marshall, H. Douglas Goff, Richard W Hartel
Emulsifiers in Food Technology Robert J. Whitehurst.
The Science of Ice Cream. Chris Clark

Part 1 of this series: Introduction
Part 2 of this series: Components
Part 3 of this series: How to Build a Recipe
Part 4 of this series: Basic Recipe Examples
Part 5 of this series: Techniques
Part 6 of this series: Sugars
Part 7 of this series: Stabilizers
Part 8 of this series: Emulsifiers
Part 9 of this series: Booze
Part 10 of this series: Solids, Water, Ice
Part 11 of this series: Introduction to Flavor
Part 12 of this series: Ice Cream Flavor: Coffee

Part 13 of this series: Coffee Ice Cream Addendum: Origin Notes and Minutiae
Part 14 of this series: Chocolate Ice Cream

Part 15 of this series: Chocolate Ice Cream Addendum