I don’t even remember why I started, but I mix my own cola from first principles. It’s a fun hobby and I get to experiment with the flavour components – therefore, in theory, ending up with a cola that’s particularly close to what I like. It’s been taking a fair number of iterations, though. I’m currently on my third batch (two of 45 litres, this one is 22.5 litres). Here’s my journey, including the fine details that some other sources don’t mention. I’ll close with the recipe I used and what I’ll change in the next batch.

Sources and prior art

Here’s what I used for inspiration:

• The Pemberton recipe
• OpenCola
• OpenSoda cola – 2009 version
• Cube-Cola

My ingredients are sourced from various German vendors:

• Dragonspice for the essential oils and the gum arabic
• buXtrade for sucralose, vanillin and caffeine
• My local supermarket for the vodka, sugar for the caramel colouring and carbonated water (I might do my own carbonation later on, but for now I want to focus on getting the mixture right)
• A friend of mine for the food-grade phosphoric acid, though there are a few vendors on the internet where I can get this from, too

All right, let’s get into the fun details. I’m going to skip over some of the details covered in depth in the other sources, and add my own spin and the things I’ve learned that I didn’t see mentioned anywhere else. Still, if you want to make your own cola, too, this should hopefully be detailed enough.

Towards my own recipe

First attempt

I started out trying a slightly tweaked version of Cube-Cola – the main deviations is that I used phosphoric acid instead of citric acid to get closer to the “proper” taste profile of cola. I also made my own caramel colouring because I couldn’t find a good source of the industrial double strength colouring they used. I’ll explain how in a bit.

The other huge deviation is one of the reasons I got into this whole thing: I was drinking a lot of cola and consuming a lot of sugar in the process – probably way too much sugar. However, I never really could get into the sugar-free colas being sold around here, which are typically sweetened using acesulfame potassium, aspartame and sodium cyclamate.

The two options I looked at were rebaudioside A (one of the components of stevia) and sucralose.

• Rebaudioside A, if refined enough, is said to have a much reduced bitter aftertaste compared to most stevia sweeteners, but it’s also a fair bit more expensive. The upside is that it’s more “natural” (whatever that means).
• Sucralose has a fairly neutral taste and is stable in acidic conditions (such as a cola concentrate that contains phosphoric acid).

Going purely from price – the initial cost for the whole project is quite high, mostly due to the many different essential oils – I ended up choosing sucralose, and I’m quite happy with the result. Obviously sucralose doesn’t taste exactly like sugar, but at least it doesn’t have the typical metallic or bitter aftertaste.

So I mixed what’s mostly Cube-Cola, with those changes, guesstimating the amounts:

Since sucralose is said to be 600x sweeter than sugar, I just divided the amount of sugar for a 45 L batch by 600 and ended up with 8.5 grams.

Phosphoric acid woes

Calculating the amount of phosphoric acid took a lot more time, and I got it wrong in my first batch. I had a 85% solution of phosphoric acid, meaning that it’s 85% phosphoric acid and 15% water (100% phosphoric acid is kind of difficult to get and handle).

The Cube-Cola recipe uses 32.5 mL of citric acid for a 45 L batch, and the OpenCola recipe equates 17.5 mL citric acid with the same volume of 75% phosphoric acid, so I calculated how many grams of phosphoric acid there were in 32.5 mL:

• The density of hypothetical pure phosphoric acid at room temperature is 1.87 grams per mL – the 75% solution is 1.579 g/mL and the 85% solution is 1.689 g/mL.
• 32.5 mL = 51.32 grams of a 75% solution comes out to 38.49 grams phosphoric acid (20.58 mL) and 12.83 grams water (12.83 mL). Scaling that to fit an 85% solution means we end up with 26.81 mL total, or 6.79 grams of water and 38.49 grams of phosphoric acid (totalling 45.28 grams).

Somehow I got some of this completely wrong and ended up with 38 mL of phosphoric acid. While it probably didn’t destroy my teeth, it did taste a fair bit too tart in the end. Whoops.

Mixing oils

Oils don’t naturally mix with water, so an important part of the process is making an emulsion from the oils. This needs an emulsifier - in the OpenCola/Cube-Cola recipes, gum arabic is dissolved in water and the oils blended into that.

The Cube-Cola folks recommend using a whisk attached to a power drill to really mix up the emulsion, but that seemed unnecessarily complicated. I tried getting a dedicated stirrer but the best I could get was a wimpy little battery-driven hand-held milk frother - which kind of worked okay, but it took quite some time. I added a very small amount of vodka to help stabilize the emulsion (many oils dissolve in alcohol), which I think helped the process along a lot.

Making caramel colouring

Caramel colouring is basically a solution of very strongly caramelized sugar in water. The extreme caramelization turns it a very dark brown and it stops tasting sweet - instead you get a bitter flavour and a rather different smell.

Warning: caramel gets very, very hot – way hotter than boiling water – and can cause terrible burns. You don’t want it on your hands, and you want it in your face even less. Be very, very careful.

The process is not difficult – aside from the risk of serious injury:

• Grab a pot (with lid) you won’t mind ruining a little. Also have a spoon handy that can stand very high temperatures - metal will work, but then you can’t leave it in the pot throughout the process, otherwise it will become too hot to handle. Also, don’t use a tiny pot because the caramel will boil and it inflates quite a bit in the process.
• Take two parts sugar and one part water (but don’t add the water just yet). I used something like 350 g sugar and 175 mL water to start out with, but using less will make this easier and let you do more attempts over time.
• Put the sugar in the pot. On high heat, let it dissolve and then caramelize.
• Make sure you have sufficient ventilation - the smell can get quite intense.
• Stir occasionally so all of the sugar is caramelized equally.
• At some point the caramel will boil and sizable bubbles of air will burst on the surface. You’re not done yet - this stuff needs to turn a very, very dark brown. I didn’t let it caramlize enough the first time which made it quite weak.
• Once it has reached a sufficiently dark shade of brown (this may take some experience to recognize), we get to the fun part. Brace yourself!
• Add the water and immediately close the lid – a lot of steam will form. You may need to have your kitchen hood active on its highest setting.
• Turn the heat to low. Give the devil pot a few minutes to settle down. The steam in the pot will help melt and dissolve any caramel that might have gotten stuck to the sides of the pot during the violent uprising that probably happened when you added the water.
• You should end up with a somewhat viscous, dark brown liquid. Transfer it into a sealable glass jar (needs to be heat-proof), close it and let it cool to room temperature. Add a bit more water and stir if it got too syrupy.

Adding the other ingredients

To the oil mixtures, I added (in this order – you don’t want to add the acid first and the water later, it can splash and ruin things, including your eyes):

• caramel colouring (I’ll get to the exact amounts later)
• phosphoric acid (careful, it’s corrosive)
• caffeine powder
• sucralose – all the other recipes work by creating sub-batches of cola syrup from the concentrate, combining the concentrate with sugar syrup or sweetened water, but I thought that sounded tedious and I wanted to try a ready-to-serve concentrate.

I ended up with a jar of concentrate and all I needed to do to get a 1.5 L bottle of cola is add one teaspoon (plus a little bit) to a bottle of carbonated water. This first 45 L batch lasted me a few weeks.

The result

Aside from being way too tart, there were a few more issues:

• Some of the oils (lavender and nutmeg, maybe also coriander) left a somewhat unpleasant aftertaste after a while.
• I didn’t realize, for the first few bottles, that the sucralose doesn’t fully dissolve in the concentrate – it’s too saturated and sucralose doesn’t dissolve fully in concentrated doses in the first place. So, I ended up having some fairly non-sweet bottles.
• The flavour profile was different in each bottle, suggesting that the emulsion wasn’t that good after all and some of the flavour components were used up mostly in the first few bottles. Later bottles tasted a fair bit blander.

After toiling to consume all of that concentrate, I got ready to make the second batch…

Second attempt

I made these changes:

• I tweaked the flavour mixture – using less of the oils that I found to cause an unpleasant aftertaste, and generally tinkering with the balance. Notably, I used a proportion of cassia oil that was closer to the OpenCola recipe than the Cube-Cola one.
• I used way less phosphoric acid – this time I undercompensated, I went with something like 23 mL.
• I used more sucralose, because the first batch didn’t seem sweet enough.

I also added a bit of vanillin, inspired by the Pemberton recipe. Vanillin is the primary flavour component of vanilla – all the other flavours are missing, so the taste is a lot less complex, but it’s also really cheap and ridiculously strong.

To estimate how much I needed, I consulted a number of sources regarding how much vanillin was contained in typical vanilla extracts, and settled on adding 0.33 grams of vanillin per 45 L batch. This works fairly well.

Finally, I added a few new flavour oils that I thought might complement the taste, notably grapefruit oil and clove oil.

This time, I gave up on the milk frother and got myself a cheapish magnetic stirrer - which is basically a plate you put a container on, and a small magnetic stir bar (coated in PTFE) in the container, and the plate rotates the bar using the magic of magnetism.

I got mixed results (har har) out of this. The bulk of the emulsion settled faster, but it seemed like the stirring process tended to push the oil to the sides of the container, so in the end I had to do a lot of fiddling to get everything caught up in the mix.

The result was quite promising, though this time it wasn’t tart enough due to the low amount of phosphoric acid, and that made it taste rather too sweet and fruity. Specifically:

• The aftertaste was improved, but still a little too strong in my estimation. I resolved to cut out the neroli oil next time (it wasn’t really noticeable to me and it’s really expensive), and I cut down on the problematic oils again. I liked what the new oils did to the taste, though, and decided to add a little more of those in the next batch.
• Flavour performance still varied from bottle to bottle, so the emulsion still wasn’t good enough.
• By remembering to stir the concentrate with a spoon each time I used it (paying attention to the bottom where all the undissolved sucralose was partying hard), I got way more consistent sweetness out of the batch.

Third attempt and notes on future changes

I’m getting there! To make this easier to follow, this is the full recipe and guide.

I made an important change to the mixing process: this time I used a double-whisk hand mixer with just one of the whisks installed. This made the emulsion happen a lot faster. Also, I believe it makes for a superior emulsion, as explained by “Blasted Bill Putt” in his Open Soda videos: due to all the shearing forces with the whisk hitting the sides and bottom of the container, the individual oil drops get “cut” a lot more often, resulting in smaller drops (we’re talking about microscopic drops here), which is what a successful emulsion is all about.

This batch also introduces more flavour components, inspired by the list of ingredients on “Red Bull Cola” (available in parts of Europe) which is a rather opinionated take on cola. Here’s what showed up for the first time in this batch:

• Cardamom oil
• Galangal extract
• Macis (nutmeg blossom) oil instead of nutmeg oil – the difference is quite subtle, though

The recipe (without the changes, see next section!)

Warning: when I say glass, I mean glass (a ceramic mug will do, too, but don’t use plastic – gum arabic sticks to it like mad and oils can dissolve it).

• In a tallish glass that just barely fits a hand mixer’s whisk, dissolve 10 grams of gum arabic (powdered, food grade!) in 20 mL of water. I suggest adding the gum arabic first, whisking gets less dusty that way.
• In a small separate glass, collect the oils. My bottles of oil have a dropper that outputs roughly 0.025 mL per drop, I’ll add drop counts based on that:
  • 2 mL (80 drops) of (sweet) orange oil
  • 3.25 mL (130 drops) of lime oil
  • 0.5 mL (20 drops) of cassia oil
  • 2 mL (80 drops) of lemon oil
  • 0.75 mL (30 drops) of macis oil
  • 0.2 mL (8 drops) of coriander oil
  • 0.1 mL (2 drops) of lavender oil
  • 0.25 mL (10 drops) of grapefruit oil
  • 0.2 mL (8 drops) of clove oil
  • 0.2 mL (8 drops) of cardamom oil
  • 0.2 mL of galangal extract – this is hard to measure out, and harder to mix, since the extract is very sticky… so I may not bother with this in the future.
• Add 0.33 grams of vanillin to the gum arabic solution. Vanillin doesn’t dissolve that easily, so it’s important to do this as early as possible.
• Optionally, add 5 mL of vodka (~1 tsp) to blend the oils better. The Open Soda folks say that this improves the blending of flavour notes, too. YMMV.
• Start stirring the gum arabic solution to get an even consistency. Use the “power drill with kitchen whisk attachment” approach as championed by the Cube-Cola folks, or a double-whisk hand mixer with a single whisk installed. Either way, make sure you do this long enough to break up all the clumps that formed in the first step.
• Keep stirring and slowly add the oil mixture. Put the mixer on as high power as you dare without spilling anything. Let it touch the sides and bottom of the glass a little to increase the shearing force.
• Keep stirring for several minutes – I mean it. You can test it if you’re not certain whether you’re done: add a couple of drops of emulsion to a glass of tap water, mix and let it sit for a few minutes – no oil should collect on the surface.
• Add 190 mL of caramel colouring (“normal” strength) and whisk to get a smooth blend.
• Add 26.8 mL of 85% phosphoric acid or 32.5 mL of 75% phosphoric acid. That’s 45.27 grams or 51.31 grams respectively – weighing may be easier depending on your equipment.
  You need food-grade (“U.S.P."; or analysis grade, often called “p.a.” or “A.C.S.") phosphoric acid – lower grades may contain contaminants that are very much not good for your health.
  Warning: phosphoric acid is corrosive, handle with care. It can’t hurt to use hand and eye/face protection. Have a few water-soaked paper towels or some such handy, and use them to dilute/wipe off any spills as quickly as possible. Let it air-dry before you bin them, to be on the safe side.
• Add 4.5 grams of caffeine strained through a fine sieve, to break up any lumps. You may need to whisk again because it will tend to clump together.
  This will yield a drink with caffeine content comparable to what’s in the big brands. You can go a little higher, but don’t overdo it. Caffeine can kill you. The highest concentration of caffeine that can be legally sold in cola drinks in Germany would equal 11.25 grams of caffeine in this recipe. That’s not a recommendation to go that high, just a frame of reference.
• Add 8.5 grams of sucralose. I’m not sure whether straining helps here - some of the sucralose crystals are quite large, but if you do strain it through a sieve, it still doesn’t dissolve fully in the concentrate. Either way, whisk again.

Done! A 1.5 L bottle of carbonated water needs about two teaspoons of concentrate, or roughly 6 mL of concentrate per litre.

Warning: the concentrate is still quite acidic, avoid spills and wash it off generously if comes into contact with anything. Keep face away from any potential splashes.

First tests indicate that this could still use some changes:

• The lime/lemon taste is a little too dominant. The overall balance of citrus flavours doesn’t feel quite right.
• The clove oil kind of doesn’t really do anything, but I did like the taste of it in the second batch when it was overrepresented due to insufficient emulsion.

Cutting corners

This is a lot of components. If you want to try it yourself but don’t want to get quite as many different ingredients, here are some suggestions:

• You can use citric acid instead of phosphoric acid – it’s a lot cheaper and easier to get. For the above recipe, go with 32.5 mL.
• The most important flavour components of cola are citrus, cinnamon and vanilla. You could probably get by with just these oils: orange, lime, lemon and cassia, and the vanillin (or vanilla extract if you like, but you’ll have to look up the right amount yourself).
• Caffeine does not affect the taste. Personally I don’t really care about it that much but I wanted something “authentic”. You can easily leave it out.
• Caramel colouring can be bought ready-made, the standard stuff you get should probably be used just as described above.
• Allegedly you can get an emulsion with just alcohol instead of the gum arabic sludge. I haven’t tried it myself and I don’t know how much you need. The recipes I found used alien units like “quarts” and “oz”, which as far as I know is the domain of some kind of wizard.

Making a bottle of cola

The basic idea, of course, is adding the right amount of concentrate to a bottle of carbonated water, e.g. 9 mL for an 1.5 litre bottle. Here are a few things to pay attention to:

• Before you use your teaspoon to shovel concentrate into the bottle, give the concentrate a good stir – after all, some of the sucralose has settled at the bottom and you want it a little more distributed than that.
• It’s a good idea to skim off any drops at the bottom of the spoon (I use the top of my jar) before transfering it to the bottle – dripping is a great way to create a somewhat corrosive mess.
• Angling the spoon just right takes a bit of practice. You can cheat using a small funnel if you prefer.
• Add the concentrate very slowly (I spend like ten seconds on each teaspoon), this reduces foaming. You can get a lot of foam while pouring in the concentrate.
• Keep the bottle cap handy and screw it on fairly quickly after you’ve finished adding the concentrate. The foam is still working and this will head off a small fountain.
• With the cap in place, turn over the bottle to let any concentrate trapped in the foam mix with the bulk of the water (and possibly re-mix any concentrate that sunk to the bottom without fully dissolving). Of course, this will increase foaming once you re-open the bottle…
• When re-opening the bottle, go very slowly, to avoid the cola christening you Foamy McFoamface.

Planned changes

• 2.5 mL (100 drops) of orange oil
• 1.5 mL (60 drops) of lemon oil
• 0.25 mL (10 drops) of clove oil – maybe 0.3 mL if this still doesn’t come through in the overall flavour
• Leave out the galangal extract – I do think I like the taste, but it’s really tedious to mix in with everything else due to its very high viscosity.

Well, there you have it. Experiments will continue until all volunteers have died from cola poisoning.