The Trouble with Soy

Posted by Stella Cochrane on

With the rise of so many different dietary intolerances any good coffee menu needs to include at least one diary milk alternative for their customers. The trouble is that many of these are extremely volatile to heat and don't make for a very smooth latte. Not ones to shy away from a challenge, our coffee partners over at Five Senses Coffee, set about investigating the chemistry behind this phenomenon and coming up with a solution which doesn't turn the stomach.

Following article by Jeremy Hartley, Five Senses Coffee

'“I'll have a decaf, no foam, skinny, tall soy latte.' It's a modern American cinematic cliche, a caricature of the fashion conscious urban coffee consumer with the conspicuously complicated demands! However, for the barista on the cutting edge of the speciality coffee industry, this order might present more challenges than just remembering the details. The trouble is that soy has a frustrating tendency to turn into tofu right before your eyes - and that's not a good look!

The problem is exacerbated for those of us working in the specialty coffee industry. As we push the envelope of lighter roasts to preserve the flavour of the beans, we also preserve more of the natural acids in the coffee. The same chemicals which make our brew brighter and clearer don’t play nicely with colloidal proteins. Stay with me here – with a little bit of protein chemistry it all makes sense.

Curds, whey and colloids
There is a lot to be said for proteins. If you have the right ones handy, you can make spider webs, enzymes, even antifreeze and, if you can suspend them in water, a refreshing milky beverage. The trick is to keep them suspended in water.

Proteins are large, bulky, sticky molecules and by rights, they should clump together and drop out of solution. The reason they don’t is that they are covered in charged (positive or negative) chemical groups. The electrostatic charges on the molecules repel each other and they can never get close enough to entangle and clump. When this is the case, protein strands will stay suspended in liquid indefinitely. It’s called a colloid. The protein isn’t dissolved in the water, rather it’s a solid which floats around in it and never settles.

It’s a fine situation for a refreshing milky beverage until you begin to add acid. The proteins in your drink are endowed with an overall negative charge. Acids are strongly positive. As you add more acid to the drink, the acid begins to cancel out the negative charges which are keeping the protein strands apart from each other. The protein strands link up and clump together. And voila, curds and whey. If you were the kind of kid who always wondered what Miss Muffet was eating, let me help you out; the curds are the chunks of coagulated protein and the whey is the watery stuff in between. I wondered for years.

It’s a very useful process if you want to get proteins out of water. Bacteria eat the lactose sugars in milk and excrete lactic acid, and you end up with yoghurt or, eventually, cheese. If you start with soy bean extract, it’s a step on the way to tofu. It’s not so handy if you’re pouring latte art.

In coffee, acid is only half the story. Not only are you adding acid in the form of your espresso brew, but the milk is heated as well. The heat speeds up the process of acid-induced coagulation, but also has its own effects. If you go far enough with your milk steaming, the heat can damage the protein structure (denature it) in a way which makes the clumping problem worse. It’s what you see when you cook an egg and it turns from clear and runny to white and chunky. So if you’re making cheese or tofu, a little acid and heats is OK, I’d even recommend it, but it’s a difficult balance for a smooth soy latte.

Unscrambling the egg

five senses coffee soy milk testing for common man coffee roasters
It’s not every day you get a chemistry problem to solve in your coffee, and reports of baristas fighting back tofu in their cups around the nation was too much chemistry for us to resist the challenge. We roasted coffee. We collected soy milk. We bunkered down in the Rockingham cupping lab and we went to work.

First up, we made few basic measurements. We tested the pH of soy milks straight out of the bottle. Then we added measured amounts of acetic acid to samples of each milk and streaked them onto a black sheet to visualise clumping. We certainly got very different results from each of the milks. Some went to yoghurt almost straight away (I’m looking at you So Good), some thickened a little but wouldn’t go fully chunky, even with the addition of lots of acid. I’ve never spent so long peering into coagulated soy milk. It’s not something I want to do again.

five senses coffee testing soy for common man coffee roasters

Next in line was the coffee. We brewed up a range of our roasts and measured the pH. The results were surprising. The short version is that you can’t reliably predict the acidity of a brew by the taste. Some of our least acid-tasting coffees measured as the most acidic with objective equipment. For these roasts, it seems that the acid in the bean is too well balanced on the palate to perceive it.

Then we put it all together. Shots were poured of both our most and least acidic coffees, and soy milk was steamed and poured into each. Importantly, a milk thermometer was used to keep the micro-foam temperature at a standard 65°C.

five senses coffee testing soy common man coffee roasters

The most surprising result was how few of the cups obliged with curds and whey. Only one of the milk brands tested separated to any real extent and the rest held together, even in our most acidic coffee. The milk which did fully separate, did so in all three of the coffees tested. But the separation was worse in the more acid one. It was a surprising result given how often one hears reports of curdling soy milk. I hypothesise that in many of these reports, overheating the milk is a factor. Alternatively, it could be that you’re using Pureharvest Nature’s Soy — whatever that turned to would probably have been delicious with oyster sauce.

The overall lack of curds isn’t to say that all the milks performed equally. Some definitely produced a more even and aesthetically pleasing micro-foam than others. The stand out performers were Bonsoy and Macro Organic Soy Drink. Some might be surprised that the ‘barista quality’ milks we tested were among the worst — they didn’t curdle, but it looked as if it was a real effort for them.

Finally we held the taste test. Have you seen those YouTube clips of babies getting their first taste of lemon? That moment when they realise just how horrible the world can be is comedy gold. You should stop reading right now and google it, or set up a soy milk taste comparison for two people with exquisitely sensitive pallets; it’s basically the same show.

I think the taste testers thought I was trying to poison them with burnt plasticine and acetone if you believe my cupping notes. Not cool “barista quality” soy. Not cool.

In an amazing stroke of fortune (or design), it turns out that the two most presentable drinks were also the ones which didn’t make anyone dry retch. This is not a product endorsement, but based on this experiment, next time we make a soy coffee, we’ll reach for the Bonsoy first, followed by Macro Organic Soy Drink.

So if you’re having trouble with curds in your soy milk, it might not be the milk which is the problem (unless you are using Pureharvest Nature’s soy, in which case it almost certainly is). Try keeping a better eye on your milk temperature and see if that solves the problem, you may be denaturing your proteins. If your milk still isn’t presentable, try reaching for the two milks which worked best for us, Bonsoy and Macro Organic. If all else fails, we can recommend a lower acid coffee from our range — it’s all in the chemistry!'

First appeared on Five Senses Coffee's blog.

Don't forget you can get a perfectly made soy drink at the CMCR cafes, just take a look at the microfoam on this beauty!

common man coffee roasters perfect soy cappuccino

← Older Post Newer Post →