Research from Coffee Next and the University of New South Wales finds out the secrets of the best water for coffee

About this collaboration

The Seven Miles Coffee Next team joined forces with the University of New South Wales (UNSW) UNESCO Centre for Membrane Science and Technology to gain clarity on how water quality affects the flavour of espresso. Using my background in water and analytical chemistry, and UNSW’s expertise in treating and engineering water quality, we now understand what needs to be controlled, and what doesn’t, in order to extract key flavours in espresso for any café anywhere in the world.

Who is this research for?

  1. Anyone who wants to get the best cup of coffee out of their espresso machine
  2. Anyone who wants to place a filter on the water supply to protect their machine
  3. Anyone who thinks they need to dose their water with specific salts to get the best taste out of the coffee

What did we find? Top 3 Summary

water for coffee infographic

The Background. Why water science is harder than people think

It’s been an interesting experience coming into coffee and seeing what ‘those in the know’ believe happens to their coffee when they change water properties. And I don’t just mean temperature or pressure changes.

There has been a bit of work recently investigating how taste changes by adding or removing salts, hardness and other such things. Indeed, we at Seven Miles install over 500 filters (of various types) around Australia every year to obtain better water quality. We’ve been doing this for years.

But if there’s one thing I’ve learned about water over the last 10 years of studying it – it is an incredibly complex substance.

For example, It’s the only molecule that has a localized density inversion, allowing its solid state (ice) to float on its liquid form (an iceberg for example) between 0°C and 4°C.

penguins on an iceberg
Water floats on itself, even Penguins know that’s weird. Photo by Danielle Barnes on Unsplash

Water also has a ridiculously high energy density – so much so that, if it weren’t for its associative hydrogen bonding strength, it should boil at about -112°C. That’s over 220°C below where it actually boils. How cool is that?

The point I’m trying to make here is that, besides being a super funky molecule, engineering water to do what you want is not trivial. So, when I read articles about dosing Epsom salts into water enhance the extraction of coffee flavours, you’ll forgive me if I treated it with some scepticism.

But being sceptical doesn’t do anyone any good. Doing something about it is better!

While my expertise lies in pure water properties and organic solubility, I have some understanding of how to go about quantifying this water-espresso quality conundrum.

To do this, I decided it would be worthwhile collaborating with people who know how to engineer water best at the University of New South Wales. We set up two, year-long research projects, with the objectives to:

  1. Identify what the optimum water quality was by analysing the chemistry of espresso extracted
  2. Identify the best ways of engineering the water to achieve that optimum quality.

It is the results from this first project that we will delve into in this post, though with some findings sprinkled in from Project 2!

How did we do the tests?

water testing instruments on bench
Hanna Instruments, Multi-parameter Water Quality Meter

To do the tests (without going into extraneous detail here), we did a few things.

  1. We did a lot of measuring. In field and with engineered ‘designer’ waters.
  2. We did a lot of espresso extractions
  3. We did a lot of repeats
  4. We tested the extremes. Much higher acid/base content than normal, higher/lower TDS than in the common range, higher hardness etc. – just to more readily identify if trends were there. Or not.

To analyse the chemistry, we used some of the finest analytical techniques known to modern science, including:

  • Gas chromatography with mass spectroscopy analysis (GC-MS), with authentic calibration standards run of ‘chocolate’ ‘malt’ ‘bitter’, caffeine, ‘caramel’, ‘strawberry’ and ‘vanilla’ (vanillin) compounds within the method. We used this to identify and quantify flavours in the extracted espressos.
  • Inductively coupled plasma mass spectrometry (ICP-MS) to quantify and then analyse the chemistry of the water; both pre and post-treated.
  • Hanna instruments multiparameter multiprobe to measure real-time water quality changes (e.g. pH, hardness, alkalinity, TDS, etc.)

The nice things about these techniques; they can be translated into easy-to-digest results. Delicious.

The Results - Hardness

Water hardness refers to the magnesium and calcium minerality of water (aka multivalent cation concentration). Hard water has many of these cations, soft water doesn’t have many cations.

Why are they important? Well, for coffee, if water is too hard, these cations can crystallize in your boiler and within the pipework, blocking flow and causing corrosion. Unfortunately for your boiler, and much like Luke Skywalker in The Empire Strikes Back, it will meet its destiny much too early.

Conversely, if water is too soft, history and our taste buds have told us that your coffee won’t taste so good. And now, the chemistry tells us that too! Behold the graph below.

Interactive | Click the tabs below to see how water hardness changes flavour in coffee

The 'flavour' categories described here are the following compounds measured using gas chromatography with mass spectroscopy analysis (GC-MS): 'Nutty/Roast/Chocolate' = 2-Methylpyrazine, 'Fruity' = Furnaeol, 'Vanilla/Caramel' = Vanillin, 'Caffeine (Bitter)' = Caffeine. Flavour intensity values are measured in micrograms per millilitre (µg/mL)

What this shows is how the flavour of furaneol in coffee (sometimes referred to as ‘strawberry furan’… for, perhaps, an obvious reason – and a compound we’d look for in our Wilde blend) changes with changing hardness. Indeed, soft water (water below 30ppm CaCO3 hardness) yields low flavour extraction. Hard water, on the other hand, extracts much more. However, there exists a convenient trade-off point. When one hits 60ppm hardness, the rate of rise of flavour relative to the rise of hardness decreases. Indeed, to get a significant increase in extracted flavours, we need to go above the machine manufacturers safe limit for hardness.

So, what do we conclude here?

We believe 50-80ppm hardness is BEST for brewing espresso.

This matches up nicely with the SCA recommended ideal of 60ppm and also Hendon et al. recommendation of 68ppm hardness. We agree here.

However, we do not agree with the SCA or Hendon that the upper limits of flavour delightfulness may be found beyond 90ppm (although this particular study was limited to the interaction of ‘hard’ cations with acids in coffee). At least, we haven’t found that from our experiments on over 7 significant and common flavours in the coffee we tested.

One non-conforming molecule to hardness trends was caffeine. Caffeine extraction tended to decrease with higher calcium content. We believe this is within the margin of error in the tests, as the variance of caffeine extraction yields at the extremes of the test was 6%, compared to the result 95% confidence interval of 7%. We therefore conclude here that caffeine is unaffected by hardness levels under typical espresso conditions. This still bucks the typical trend, though. Why is difficult to say without further experiments.

The Results - pH

pH is the indicator of acidity or alkalinity. pH below 7 indicates an acidic solution. pH greater than 7 indicates an alkaline solution.

The results here are simple. As pH increases, so does the amount of flavour extracted. Unless high pH is the result of high alkalinity (i.e. CaCO3 alkalinity), which can lead to chalky flavours being extracted, we don’t see much reason to be alarmed. Our recommendation here is a pH between 7 and 8.5 is best, but with the caveat that if the coffee starts tasting chalky, trying to install a water softener would be a good start.

Interactive | Click the tabs below to see how water pH changes flavour in coffee

The 'flavour' categories described here are the following compounds measured using gas chromatography with mass spectroscopy analysis (GC-MS): 'Nutty/Roast/Chocolate' = 2-Methylpyrazine, 'Fruity' = Furnaeol, 'Vanilla/Caramel' = Vanillin, 'Caffeine (Bitter)' = Caffeine. Flavour intensity values are measured in micrograms per millilitre (µg/mL)

An anomaly in the results with respect to hardness was vanillin (vanilla/caramel flavour). We found that vanillin had minimum extraction concentration around pH 7.7, and then increased at higher pHs. We don’t think this can be discounted as margin of error for the experiments, as the 95% confidence interval for vanillin was ±7%, or about 1ug/mL for vanillin. It’s possible that vanillin solubility was modified based on the presence of cosolvents extracted at higher pH levels, (as other flavonoids can be considered as cosolvents of vanillin), whereas if vanillin solubility was tested alone, it may have exhibited a negative trend. However, more work is needed to clarify this.

So, from a flavour perspective, we believe high pH has a positive effect on flavour extraction. That said, when we tested pH 12 water in the machine, we observed significant gas formation, which increased extraction times. This is likely due to neutralization between the alkaline water and the acidic water compounds, forming more CO2 than in the normal range.

Our verdict? We recommend extracting coffee between a pH of 7 and 8.5.

The Results - TDS

TDS is a bandied about acronym that isn’t understood well enough. TDS means total dissolved solids, which means what it says. Total stuff dissolved in the water. All of it. Everything that is in solution. From salts to dissolved organics, alcohols, rogue pharmaceuticals, and such. And, depending on how it’s measured, can include non-dissolved things too. So, doing studies on ‘TDS’ will always be a model study, and not always modifiable in a real-world way.

That said, we tried. We added non-hardness-modifying salts to change TDS. This means using monovalent cation salts, like sodium chloride. And, lo, that’s what we used. We used almost none, and we used a truckload to see what changed

What did we find?

We found that adding salt to increase TDS made no difference to extracted flavours.

That’s right, by taking what is commonly present in tap water and increasing that by a factor of 30, there was effectively no change in dissolved flavonoids.

Was that surprising? Actually, yes, a little. But that’s the fun thing with experiments – they can prove you wrong. Or right… But the point is they PROVE.

The Results - Chloride

We didn’t taste stuff with extreme chloride concentrations. Why? Well, we didn’t want to drink chlorine. Nor subject our machines to these extremes. However, we do have recommendations here.

Minimize it. The Specialty Coffee Association (SCA) and many machine manufacturers recommend chloride below 60ppm, and we’d stick with this. Indeed we’d go so far as to say keep it below 30ppm, based on some of our experiments from Thesis the Second. This will prevent corrosion in your stainless-steel boilers, and at higher levels, a malodorous taste in your coffee.

UNSW Seven Miles water collaboration project
Angus (UNSW) & myself (Dr Adam Carr) sharing our love of oversized charts

What is the best water for coffee?

We conclude that hardness and high pH are your friend in terms of maximizing the extracted flavours in your coffee. But, limit your water hardness to within 50 and 80ppm and a pH between 7 and 8.5 to help get consistent extractions and prevent damage to your machines.

TDS as a measure means nothing (unless you’re trying to get it as low as possible through your RO filtration systems). Stop talking about it.

Keep your nose and your machine clean by reducing your chloride to below 60ppm. Less than 30ppm would be ideal.

How do we apply this in the real world?

Clearly, the results presented above show the best water for coffee, but do not advise on how you can achieve it! Well, this is deliberate. It’s no simple matter treating water. The host of engineers at Sydney Water would tell you that. In fact, did you know that Sydney water treatment targets achieve the above chemical balance?

This means that Sydney's tap water, out of the treatment stations already has the ideal chemical balance for a perfect espresso!

That said, local pipework factors, local building works and other such things can affect the water quality, even if it’s coming out of the treatment facilities as glorious as it should. So, what’s right for the café across the road may not be right for you.

We did the second research project to come up with a solution model for all cafes in Sydney – which we now have and are rolling out across Australia. If you’d like help with getting the right systems setup to get your water perfect for coffee, please ask us: We’d love to help!

Thanks to Associate Professor Pierre Le Clech, Angus Shields & Jhonny Zhong Chau of the UNSW UNESCO Centre for Membrane Science and Technology for their time, expertise and fantastic work on the ground. Getting this data was challenging and many lost nights sleep getting things working – we applaud you. We’re looking forward doing more work with you in the future!