Introduction
Vegan ice cream is everywhere these days, and it’s not just herbivores who demand them. Some pleasure-seekers are trying to reduce their impact on the planet. Others are bothered by the treatment of dairy cows at industrial farms. Others are lactose-intolerant. Others are just jumping on a bandwagon. A marketing executive at a national brand told me that interview subjects often believed non-dairy ice-creams were better for them—even though they couldn’t articulate how or why.
Apologies to everyone for the long dark spell here at the blog. We’ve been busier than usual, with multiple consulting clients and some other projects. The questions never stopped pouring in, and we have lots to talk about. Stay tuned.
Backstory
Over the last year several people have approached me for help with dairy free ice cream. They ranged from small shop owners to product managers at major manufacturers. And they were serious. My stance quickly changed from “I don’t know anything about that stuff” to “I’d better learn all about it!” The journey has been an exciting one, fueled as much by good luck and the enthusiasm of my clients as by science. Which makes sense, because compared with traditional ice cream, there’s very little science to draw from. With dairy ice creams, I’m able to build my software modeling tools on top of well-tested axioms and equations. With dairy-free, I had to rely on quite a few educated guesses. Happily, in short time, I’ve been able to move from guesswork to some pretty great ice cream with surprisingly little trial and error.
What’s the Big Deal?
Great dairy-based ice cream relies on the special qualities of milk fat, milk proteins, and milk sugars. They all contribute, in numerous interconnected ways, to the textural qualities we usually strive for: smoothness, creaminess, the right density, the right elasticity, vibrant flavor release, and a creamy sense of lingering flavors. Not to say dairy ingredients by themselves make it easy; if they did there’d be no need for this blog. But vegan ingredients add a whole new layer of puzzles to solve. And we don’t get to benefit from centuries of other people’s experiments when solving them.
The Greasy Truth
I’ve sampled versions from several small shops, and from several commercial manufacturers. These include a much-lauded NYC and LA-based maker that sells vegan ice creams for $12 a pint. I’ve found a range of texture and mouthfeel problems with all of them. Topping the list, especially with the artisanal makers, is greasiness. Every bite leaves my mouth feeling coated with oil or wax, as if I’ve taken a bite out of a beef tallow candle. The coating sometimes takes minutes to fully dissipate, and it contributes nothing to a long, lingering flavor release (which is often a benefit with high-fat ice creams). I find they embody the worst aspects of very high-fat ice creams (slow, muted flavor release, and a greasy sensation) without the benefit of lingering flavor.
Some of these ice creams contained over 20% fat! I find this unappealing, but there’s little surprise here that so much fat would lead to greasiness. This can likewise be a problem with ultra high-fat dairy ice cream.
Not all the commercial non-dairy ice creams suffered from greasiness. Some just felt and tasted insubstantial—like a diet ice cream. This is the result of a low-fat formula, when nothing is done to compensate for the lack of richness.
These textural problems result from the lipid profile, which describes the mix of fatty acids that make up a fat. It can be summarized by the ratio of saturated fats, monounsaturated fats, and polyunsaturated fats (although there’s more to it than this). No plant-based fats match the unique signature of milk fat. The one that comes closest is palm oil, which presents other problems that take it out of consideration1
Some plant-based fats are pretty good—as long as you use very little of them. These include cashew oil (my favorite; more on this later) refined coconut oil (in very small quantities, unless you want coconut flavor), and cocoa butter. The total of these fats needs to be kept below 8%, to keep prevent greasiness. But then you’re making a low-fat ice cream, which may lack textural qualities you’re looking for.
The Underbelly Vegan Ice Cream Solution
Don’t try to adapt a dairy ice cream recipe just by substituting plant-base fats for milk fat. Instead, think in terms of making a sorbet with a bit of added plant-based fat for richness. If you find yourself objecting that sorbets are too icy, too sweet, or don’t have a rich enough texture, allow me to suggest that these are problems with bad sorbets, not good ones. And we’ve already cracked the code on how to make a creamy, smooth, not-too-sweet, rich-textured sorbet.
Here’s how to apply this knowledge to plant-based ice cream:
- Use cashew milk as the base. It has a pretty good lipid profile, is almost completely bland (you probably won’t notice the taste of cashews creeping in on your flavors); the cashew solids contribute favorably to the texture; and sustainably grown cashews are available and not terribly expensive. As a bonus, cashew milk is almost absurdly easy to make if you have a high-powered blender. No straining required. You can essentially just blend the cashews and water with your other ingredients and be done with it.
- Use vegetable fiber—particularly inulin—to enrich the texture. Inulin has long been used for this purpose by gelatistas in Italy. I’ve found it to be a perfect enhancement for sorbets. And it works wonders for non-dairy ice cream. It may seem strange, but this fat-free ingredient creates the creamy textures of dairy fat much better than any plant-based fat I’ve tried.
- Boost the solids levels with big sugar molecules, like those in atomized glucose2, glucose syrup, or tapioca syrup. They improve the density and texture of the ice cream by increasing the total solids. They reduce the total amount of water in the formula, and they bind to a portion of the water that remains, helping to stabilize it and prevent ice crystal growth. Look for a low DE number (dextrose equivalence). The lower the DE, the less these sugars will contribute to sweetness. DE 42 is usually the easiest to find, and works well.
- To increase the fat a bit beyond what you can reasonably get from cashews, use refined coconut oil. For ice creams with milder flavors, if the coconut flavor becomes noticeable, you can use a blend of coconut oil and cocoa butter.
- As with dairy ice cream, use a blend of sugars to balance sweetness and freezing point depression. Sucrose (table sugar), atomized glucose, and trehalose are all useful. The inulin has a small amount of sweetness and can be considered part of the sugar blend.
- Use an emulsifier. You won’t get any of the natural emulsifying powers of the milk proteins. Or of egg proteins. A bit of emulsifier will help create a foam structure and keep the ice cream from being too dense. I’ve been using gum arabic, at about 1% of the weight of the total fats.
- Use stabilizers that hydrate cold. I’ve been using a 50/50 blend of sodium carboxymethyl cellulose and guar gum. Use about 0.7% the weight of the total water (recipe weight minus total solids). This is a higher level than I use in dairy ice creams.
- Use salt. Milk solids are full of salts. None of the non-dairy base substitutes offer this benefit. 0.2% of the recipe weight in salt will help bring the other flavors into focus.
Sample Recipe: Vegan Ice Cream Neutral Base
628g Water
145g Cashews (raw, unsalted)
105g Sucrose
60g Atomized Glucose DE40
10g Trehalose
25g Inulin
1.5g Gum Arabic
2.25g Carboxymethyl Cellulose
2.25g Guar Gum
2g Salt
12g Coconut Oil (refined)
8g Vanilla Extract (optional … useful while testing, to check flavor release, and see if mild flavors can compete with the cashews)
Instructions
- Soak the cashews in the water overnight, refrigerated
- The next day, blend the cashews and water on high speed for 1 minute in a high-powered blender. Pause in the middle to scrape down the sides of the blender, so you don’t leave any chunks.
- Stir together the powdered ingredients. Turn blender on low to create a vortex; gradually pour in powdered ingredients until incorporated. Blend on high for 30 seconds.
- Add coconut oil and extract (if using). Blend on high for 30 seconds.
- Chill until mix is below 4.5°C / 40°F
- Spin in ice cream machine; harden in freezer.
Analysis
Total Fat: 7.6%
Milk Fat: 0%
Total Solids: 35.4%
Solids Nonfat: 27.8%
Acidity: 0.12%
Alcohol: 0.3%
Stabilizer/Water: 0.7%
Inulin: 2.5%
Gum Arabic: 0.08%
POD: 138 / 1000g
PAC: 215 / 1000g
Absolute PAC: 371 / 1000g
Rel. Hardness @ -14°C: 82
Note on relative hardness: This is a metric I created to predict the final hardness of ice cream with greater accuracy than industry standard metrics (PAC, etc.) RH takes into account the proportions and types of fats that influence the hardness of an ice cream, including cocoa butter, coconut oil, and other nut oils. Target value for standard ice cream is 70–75, which corresponds to an ice cream without added hardening fats that has a 70–75% ice fraction. This is considered ideal for scooping and serving.
The equation I devised does NOT take into account the hardening properties of milk fat; it just presumes a typical amount. The result is that dairy ice cream made to a relative hardness of 75 has a good consistency, but non-dairy ice cream with the same value is too soft. My plan is to make the formula more sophisticated, so we can work with dairy and non-dairy using the same values. In the meantime, we’re using a target of 80–85 for the relative hardness of non-dairy ice cream.
Tasting notes
I think it’s pretty good. It’s better by miles than any other dairy-free ice cream I’ve tried, including super-premium artisanal brands. It’s better than any published recipe that I’ve tried. It’s not as good as the best dairy ice creams. I feel it falls short in two key areas: flavor release and mouth coating. I find that flavors emerge slowly and in a slightly muted fashion from this base. And it still leaves a slight sense of a neutral coating on my mouth. These are the complaints I mentioned above, and the precise problems I set out to solve. I’ve managed to minimize them, but not eliminate them.
For another data point, my partner thinks I’m crazy. She thinks this ranks with the best ice cream she’s had, and wants me to make nothing but. I mention flavor release and mouth coating, and she doesn’t know what I’m talking about. Your mileage may vary.
Notes for Commercial Production
This recipe is unpasteurized. I tried a low temperature pasteurization (~75°C @ 30 minutes) followed by homogenization, and the results were bad. The cashew milk went from bland to a fairly strong cooked nut flavor that overwhelmed everything.
For commercial production, the best course may be to source prepasteurized nut milk, and to use certified aseptic dry ingredients and flavorings. This is a work in progress.
Footnotes
- Palm oil is highly susceptible to oxidation and rancidity, which makes it challenging to use in quality ice cream. Worse, much of the world’s palm oil comes from unsustainable farming practices, which contribute to deforestation and reduction of of biodiversity, including the destruction of Orangutan habitat. This includes much of the “Certified Sustainable” palm oil, which is still dubious.
- Don’t mistake atomized glucose for dextrose powder. The latter is pure, refined glucose. Atomized glucose is a corn syrup that’s been spray-dried. In addition to glucose, it’s full of dextrins and other saccharides, and is both less sweet and has less freezing point depression that pure glucose. As such, it can help add bulk without unbalancing the flavor or the structure of the ice cream.
Sounds like you pretty much nailed it. I have been working alongside very similar lines. I was also majorly disappointed by some NY brands I was highly recommended. Better stick to one’s own.
OMG! The Master returns and with a quiver full of brand new arrows! Welcome back, you have been missed.
Thank you for posting this recipe. Can’t wait to try it, and I’m glad to see another blog post!
I’m curious how you are calculating an Absolute PAC of 371. If I do (PAC / Water) I get 215 / 646 = ~333.
I get the same as you
I’m using an “absolute PAC” value that includes the effect of salts, including dairy salts. In this recipe it’s just plain old added salt, which at 2g has a PAC(sa) of 23.
Thanks for the recipe, can’t wait to give it a shot. Just curious what type of inulin you are using. I’ve read the kind that’s readily available is a lot less effective than long chain HP inulin for ice cream.
I’m thinking the same 😀 In Vietnam there is a distributor who can get me long chain inulin but I need to order at least 5 bags x 25kg which will last eternity in my small shop.
It’s great to see another blog post! This was a great read and I have so many thoughts/questions. Why have you switched to using gum arabic instead of lecithin? It seems like gum arabic is an inferior emulsifier, at least in terms of amount needed to achieve similar particle size. It seems like mixing gum arabic and lecithin is often done to achieve better emulsion particle size that either alone too.
Sources:
https://blog.modernistpantry.com/wtf/the-secret-of-emulsifiers-in-dressings/
https://www.sciencedirect.com/science/article/abs/pii/S0260877420303022#:~:text=According%20to%20the%20enthalpy%E2%80%93entropy%20compensation%2C%20whey%20protein%20concentrated%2C,delivery%20systems%20to%20encapsulate%20and%20stabilize%20lipophilic%20molecules.
Lecithin should work also. I chose gum arabic because of its slight thickening ability, and also because the soy lecithin I had recommended heating to dissolve. But most soy lecithins should be cold-soluble. I think it would be worthwhile to experiment with it. The gum arabic has yielded good results; it’s possible that lecithin will give different ones.
Any thoughts on using a mixture of coconut oil and high oleic sunflower oil? There’s some studies showing that it matches well with butterfat and in my experience, it helped prevent greasiness with cashew ice creams.
http://icecreamscience.com/vegan-ice-cream-selecting-a-non-dairy-fat/
Also it seems like this recipe would especially benefit from using a pacoject or creami — might get you that higher flavor release and handle the lower fat levels well. I’m super excited to give this a shot.
Thanks so much for sharing the fruits of your hard work! I’ve found everything you’ve said to be true. I’ve had a lot of trouble with greasiness but inulin has proven to be quite the magic ingredient. I’m also having trouble with flavors being a bit muted, so I’ll increase the flavoring. I plan to try a batch with tapioca powder in place of inulin because it’s far cheaper where I live.
Update: I’ve found the tapioca starch thankfully works well enough for me. Maybe it’s my imagination, but I think it makes the ice cream more slimy than inulin, so I’m using a bit less. I’ve also think I’ve found the solution to the slow flavor release: inclusions. Add in some stuff with flavor that hits you right away. I made some peanut butter ice cream with the recipe (swapped cashews with 200g peanut butter, removed salt, reduced sucrose a little). The peanut butter flavor was AMAZING but came on too slowly, so we mixed a bunch of Oreo crumbs in (like a mcflurry) and the sweetness of the Oreos hit right away and then a pure and powerful peanut butter flavor took over. 10/10 best ice cream I’ve ever had, better than any dairy ice cream.
Love the article! Very nice recipe, and vegan ice cream indeed seems to be on the rise. Like a lot of vegan products, actually.
I’ve looked around a bit to see what ingredient replacements well-known commercial brands use. Most of them replace the fats simply with coconut oil, sometimes combined with soybean oil. The MSNF replacement is a bit more interesting. Here are three solutions I found:
Ben&Jerry’s: almond paste 3% + pea protein (some small %)
Hertog (well-known Dutch brand): soybean flour
Local supermarket brand: pea protein + tapioca starch
Out of curiosity, have you tried any of these ingredients to compare them with the cashew solution?
Question, would you use this recipe to make ice cream for milk intolerant (fine with cheese so we can go with lactose intolerant?
Absolutely. There’s no lactose or other dairy ingredient.
Echoing many others, this blog is such an invaluable resource.
I may be missing something obvious here, but what was your full calculation to reach a PAC/1000g of 215?
After adding all the salt, sugars (even in the cashews themselves) and alcohol from the vanilla extract I still don’t get up to nearly 215, so wondering what I’m missing…thanks in advance
Yes, why isn’t this recipe rock hard?
Hi. wish to understand the concept of using gum arabic, Guar gum and CMC, all together. What is the seasone of using all of them. TNX
Gum Arabic is an emulsifier. It should help the ice cream’s whipability. The other two are stabilizers, for thickening and suppressing the formation of large ice crystals. CMC is most effective at ice crystal suppression; guar is most effective at adding body. But they both do a bit of each.
My partner (who recently went vegan) was gifted a Vitamix 5200, and this recipe is outstanding! The ease with which it comes together is frankly mind-boggling, and an utter delight. I will admit that I’ve ended up tweaking it very slightly via icecreamcalc to accommodate the ingredients I have on hand (dextrose and maltodextrin instead of glucose syrup solids) and our tastes (we sometimes go as high as 160 POD but usually shoot for more like 145-150), but this method and the basic amounts are superb, and a great springboard for other flavors. I was able to actually make a very decent coffee version by just substituting strong coffee for the water in the recipe, and a bananas foster version where I used about 360g very ripe banana to stand in for some of the water, inulin, and sucrose, then added a splash of rum and used brown sugar instead of granulated. I have some ‘banana butter’ to use as a ripple that is bananas reduced with a bit of molasses, lemon juice, rum, and cinnamon. I also blended up a Pecan Pie flavor that used some maple syrup as part of the sugar mix, and swapped out a bit of roasted pecans for part of the cashews; I’m really stoked to churn that one up later this evening. I’m so excited to explore this avenue of ice cream making further. Thank you so much!!
could I use egg yolks to replace gum arabic if I am not worried about it being vegan? I eat vegan ice cream to avoid dairy only.
Sure, try it.
Unfortunately for me this recipe resulted to spongy/gelly texture and mouth feel.
Same happened when I tried to recreate the sorbet recipe.
I’m not sure where the sponginess comes from, but as these are the only recipes that I use CMC in, I have started blaming it.
I will try to reduce the stabilizer amounts in my next batch to see if it helps.
I haven’t had similar problems with the recipes using LBG and cooking.
Can you let us know what brand / type of CMC you used? Unfortunately with many gums, and with CMC in particular, there are different varieties that can yield very different results. It’s my fault for not specifying what I used and being clear about the considerations.
I used the Modernist Pantry version, which has the advantage of being easy to find in small quantities in the U.S.. But it has the disadvantage of lacking more specific data about its properties. Like many other M.P. products, it’s probably a repackaged gum from Ingredeon. They never reveal exactly what they’re giving you.
If you’re getting nasty textures, I’d first try getting your hands on the M.P. version, if possible. If not, I’d cut way back on the CMC quantity.
This can be an issue with any recipe that includes gums. They are powerful ingredients, and they’re not all the same just because they have the same name. Anytime you get an exaggerated texture, try scaling back.