I didn’t drink coffee for 26 years. Back in 1986, a small version of me had an exciting but regrettable encounter with a machine which dispensed instant coffee powder and sugar. It was exciting, way too exciting. I did my best to forget and move on, and I didn’t give it another thought until the advent of Breaking Bad. I was half way through a PhD in drinking water chemistry when I saw it — Gail’s magnificent, beautiful, bubbling apparatus for brewing the perfect cup. My mind locked on and the penny dropped — coffee was chemistry, and I was good at chemistry.
Two years later and I’m measuring out water samples in the Five Senses cupping lab alongside barista champions and Q graders. We’ve been galvanised into action by Maxwell Colonna-Dashwood’s water-centric World Barista Championship routine. It’s about time the topic got some real attention, and we didn’t want to let the chance go by.
Water isn’t an ‘ingredient’ in coffee
You often hear that water is the most important ingredient in coffee because it makes up 98% of your drink; but it’s much more than that. Water isn’t just an ‘ingredient’ that you add to roasted coffee seeds — water is the solvent which extracts the flavour compounds from the seed. It has a very active role. To dissolve flavour compounds in water, the water has to form chemical bonds with solids in the seed and carry them away.
To a chemist, brewing coffee is a kind of selective solvent extraction.
There are delicious chemicals in the seed, and there are some truly nasty ones. The whole trick of brewing is for your solvent to collect the delicious ones and leave the others in the puck. We use temperature to change our solvent characteristics all the time in coffee. Take cold brew for example; cold water is quite a different kind of solvent to hot water and it extracts an entirely different profile from the seed.
Shout-outs to Maxwell and Chris
Knowing what difference water can make, Maxwell’s WBC presentation caught my attention — and everyone else’s too. In case you missed it, here’s my best shot at fitting the water-related bits in a nutshell.
Maxwell starts his presentation by introducing three important impurities which are found in our water — magnesium, calcium and ‘buffer’. He presents the judges with a filter brew made from ‘soft, empty’ water which he describes as ‘OK’ and having ‘slightly sour acidity’. Then they drink a filter brew made from water with a high calcium content, but having the same Total Dissolved Solids (TDS) as the first. For this brew, Maxwell suggests that the high buffering capacity destroys other flavours. Finally they drink from a brew where ‘all the ingredients are working together’, which we are to understand makes a much better cup.
For the espresso brews which he is judged on, Maxwell comes prepared with a number of roasts from light to dark, and matches the roast to his measurements of the local water quality. The local water has a high alkalinity, so he uses a lighter roast and faster flow rate during the extraction to compensate for the dulling of acids. During both the following discussions and the opening presentation, Maxwell describes the effect of each of these components on coffee flavour. He suggests that magnesium ions in water aid the extraction of sharp, fruitier flavours, calcium emphasises heavier, creamy notes — and that ‘buffer’ is antagonistic towards sharper, acidic notes.
But there’s more! This presentation stems from a collaboration with Chris Hendon from the University of Bath and there is a published scientific journal article to prove it (Hendon et el, 2014). The journal article describes a computer modelling experiment where the binding energy between magnesium, calcium or sodium ions and seven different flavour components of the coffee seed is calculated. This provides a prediction about how the presence of these ions will affect the flavour of the brew.
The results predict that these ions will increase extraction of each of the seven flavour components by binding to them. Magnesium will have the greatest effect on extraction, calcium slightly less and sodium still less again. While the magnitude of the effect of each ion is different, the balance of effect is roughly the same. So the model doesn’t predict big differences in the flavour of the brew, so much as in the overall extraction — which isn’t what Maxwell said on the practical side of things, so it left me wondering.
It’s fantastic to see some solid science on the topic of coffee brewing, so a very genuine shout-out to these guys for putting in the work.
I know that a single graph can represent a lot of hours of work, but I was curious for some TDS or sensory data to back up the predictions of the computer modelling. That’s the nice thing about science; someone publishes something new and the rest of the world tries it and comments.
Respecting the chemistry
If you’ve been appreciating the buzz about water, but are feeling lost on the detail, you’re about to be brought up to speed. The three components (impurities, really) we are talking about are calcium, magnesium and carbonate.
Calcium and magnesium are metals dissolved as charged particles in water (Ca2+ and Mg2+). The positive charge is important, because most of the flavour compounds in coffee are negatively charged when dissolved in water, which means they are attracted to the positively charged metal ions. If you’ve heard talk of ‘hard water’ by the way, it doesn’t mean whisky. It means water with a lot of calcium and magnesium in it. There, I made a dad joke. I’m not even sorry.
Carbonate isn’t a metal. It moonlights as CO32-, which is a negatively charged cluster of carbon and oxygen atoms. If you come across the terms ?alkalinity? or ‘buffer’ when talking about water for coffee, it is almost certain to refer to carbonate.
The important quality of carbonate for us is its ability to soak up acid (H+). Carbonate can bind up acid to become HCO3- or H2CO3 and then release it again if the acid concentration falls. So carbonate helps to keep the acid level steady — it locks it up when there is too much of it around and it releases it again when there is a shortage. This is why carbonate is called a ‘buffer’, because it provides a buffer against changing acid levels.
Acids are associated with sharp, lively but sour flavours in coffee, so you can imagine a bit of buffer might promote balance, but too much could really flatten things out.
Aside from taste, carbonate also has a role in the health of your equipment. It’s a two edged sword. A little bit of carbonate is useful because of that balancing trick where it binds up acid, but if there is enough calcium around, then heat encourages the formation of calcium carbonate. Congratulations, you just made limestone with your coffee machine.
Brewing for graphs
Chris Hendon’s journal publication suggests that magnesium ions in the brew water will increase the extraction of a number of flavour compounds and that calcium will have a similar effect, but less so. I really felt the lack of published TDS data to back this up, so an experiment was designed to allow me to collect my own data.
The basic idea was to prepare Clever Coffee Dripper (CCD) brews using water with varying magnesium chloride concentrations and to measure how this affected the TDS with a refractometer. Using the CCD is one of the more repeatable brewing techniques to start with, but I needed to make sure it was at its most reliable. This was done by weighing the correct amount of low TDS reverse osmosis permeate water (TDS of 4 ppm) into a Bonavita kettle set to 96°C, and then dosing it with analytical grade magnesium chloride, so that the entire volume of the kettle could be poured steadily into the CCD over 25 seconds.
Five replicates at six different concentrations makes 30 brews, each from their own precisely-dosed kettle. I thought it would be more fun. Lots more fun. It was such a pretty graph though, that I forgot about the grinding, the washing and the syringes when I first laid eyes on it. I’m going to call it William, unless it’s in trouble, and then I’ll call it ‘Figure 1: the effect of magnesium chloride concentration on the TDS of brewed coffee’.
What William tells us is that on average, the TDS of coffee is increased by increasing the magnesium chloride concentration, up to a point. If you go high enough, adding more magnesium has a lowering effect on TDS. The most likely explanation for this in layman’s terms is that small amounts of dissolved magnesium increase the solubility of solids in the coffee by binding to them and pulling them into the water, but if there is enough magnesium in the water, it begins to compete with the coffee solids for chemical space. The concentration required to see a lowering effect is quite high, at around half a gram per litre of magnesium chloride, so it’s unlikely to come to that in practice.
Much more interesting, though, is that even with all the care taken to make each CCD brew the same, the variation in the TDS caused by the brewing variability is generally much larger than the variation caused by changes in the magnesium concentration. In other words, as far as total TDS goes, it’s likely that the way you pour liquid into your CCD is affecting the brew much more than the amount of magnesium in the water.
But can you taste it?
I’m a big fan of the VST refractometer, but the nuances of coffee flavour are much more subtle than total dissolved solids. When I’ve got one handy, I prefer to use a Q grader. As an added bonus, all of the Q graders I know are quite a lot funnier than a refractometer, so when I got five of them together, it was the most fun experiment I’d done in ages.
This experiment involved making up a series of water samples with varying levels of magnesium chloride, calcium chloride or sodium bicarbonate. The water samples were tasted on their own, and so was the coffee brewed from them, for which we used Yirgacheffe seeds roasted for filter. There is a lot of sensory data there, so I’m going to cut to the chase with some conclusions for each.
At 100 or 200 ppm, coffee brewed with MgCl2 was sweeter, brighter, stronger and had more body than coffee brewed in straight RO water. At higher concentrations though (300, 400, 500 ppm), grassy, chalky flavours began to emerge. There was also a general perception that the quantity of acid increased, but the quality decreased.
Calcium chloride was an interesting one. In plain water samples, it was quite unpleasant in concentrations from 200 ppm and above. It made water with a gritty, powdery mouthfeel which tasted chalky (Guess what chalk is made of?). In coffee though, there was none of this chalkiness or mouthfeel. At 100 ppm, the brew was rounder and sweeter than the plain water brew, and had a creaminess which the magnesium-enriched samples lacked. An unpleasant dryness began to creep in from 200 ppm though, as well as some earthy notes and muted acidity.
Sodium bicarbonate is quite delightful in plain water (It even makes it taste wetter!). At 100 ppm, it cleaned and balanced a brew nicely, taking the harshness out of the acidity. Add a little more though, and the cup starts to get flat and boring. It also becomes dry, earthy and lacking sweetness.
Drawing some conclusions
What does all this experimenting mean for our brewing, and how does it compare to what is already known about the effect of water quality on coffee? The most comprehensive work on the subject is probably the SCAA-published ‘The Water Quality Handbook’ (Beeman, Songer and Lingle, 2011). It includes guidelines for calcium hardness, TDS and total alkalinity. There is not a lot of information in there about the effect of specific ions, although it refers to research suggesting that the presence of calcium in the water is detrimental to the solubility of some flavour compounds, which could disagree with the conclusions of Hendon, who suggests that the effect is largely positive.
At the end of the day, though, the most important information is sensory and, to my mind, the most novel aspect of Maxwell’s data is the suggested sensory effect of magnesium and calcium. Our sensory data is certainly in agreement with this, at least in filter coffee, and at least in concentrations of around 100 ppm of calcium chloride or magnesium chloride.
If I were to bring all this together, I would say that at concentrations of around 100 ppm, magnesium chloride accentuates acidity, sweetness and body, calcium chloride adds a round sweetness and creaminess and bicarbonate buffer leads to a more balanced acidity. At concentrations much higher than this, though, all three of them were detrimental to the brew quality.
Put yo lab coat on!
Science is a little about knowing things, and a lot about trying things. Chris and Maxwell have made some interesting suggestions about the specific ions which make up TDS. We gave them a good test, and hopefully we’ve been able to fill the ideas out a little, but the real win will be to hear back from baristas who have tried and critiqued our own contribution. I’m even going to give you instructions along with some figures comparing this water with what might be coming out of your tap.
I call it 70/30 water. It’s 70 ppm sodium bicarbonate and 30 ppm magnesium sulphate, and here is a link with instructions to prepare it with ingredients you can buy in a supermarket. Try it. Improve it. Brew some Colombian, Indian or Sumatran. Tell us we’re wrong, or tell us we’re right — in the end, it’s all good for the coffee!
- Beeman, David, Paul Songer, and Ted Lingle. Water Quality (Specialty Coffee Association Of America Handbook Series, Second Edition). SCAA, 2011. For more info: SCAA Standard Water for Specialty Brewing
- Hendon, Christopher Holman, Lesley Colonna-Dashwood, and Maxwell Colonna-Dashwood. The Role of Dissolved Cations in Coffee Extraction. Journal of Agricultural and Food Chemistry (2014).
- Hartley, Jeremy. 70/30 Water. Five Senses Coffee (2014).