Water chemistry and brewing

Two properties, often mistaken for one

Sourced

The World Health Organization's background document on the subject describes hardness as caused by dissolved metal ions, predominantly calcium and magnesium, and as most commonly expressed in milligrams of calcium carbonate equivalent per litre. That unit is a convention for adding the two together; it does not mean the water contains that much calcium carbonate.

The same document reports a common classification, which it attributes to an earlier source: below 60 milligrams per litre is generally considered soft, 60 to 120 moderately hard, 120 to 180 hard and above 180 very hard. These are descriptions of water supplies. They are not grades of suitability for coffee.

Alkalinity is a separate measurement. The researchers whose tasting is described below define it as the buffering capacity provided by carbonate and bicarbonate. It is what resists a change in acidity when an acid is added, and brewed coffee is an acidic drink.

The carbonate system: why alkalinity is not pH

Sourced

A geological survey's reference text on water chemistry defines the alkalinity of a solution as the capacity of what is dissolved in it to react with and neutralise acid. It is found by adding acid until the water reaches a fixed end point and recording how much was needed. In almost all natural waters, the text says, that capacity comes from two forms of dissolved carbon dioxide, bicarbonate and carbonate.

Which form is present depends on the water's own pH. On the constants the text tabulates, dissolved carbon dioxide outnumbers bicarbonate by about a hundred to one at pH 4.4, which is why the titration ends near there: virtually all the bicarbonate has reacted by then. At pH 8.3 bicarbonate outnumbers carbonate by about the same margin. The text's own sentence prints the first of those ratios the other way round; the arithmetic on its constants gives it as stated here. Dissolved carbon dioxide neutralises no acid, so between those points, which covers nearly all tap water, the alkalinity is bicarbonate for practical purposes. That is why coffee guidance and water reports use alkalinity, bicarbonate and carbonate hardness almost interchangeably, and why the tasting study below could raise alkalinity simply by adding a bicarbonate salt.

It is also why pH and alkalinity are different questions. The pH of a water says how acidic it is at this moment. Its alkalinity says how much acid it can absorb before that changes. Two waters can both read pH 7.5 and differ tenfold in alkalinity, and it is the second number that decides what happens when the water meets coffee, which is acidic. A water's pH on its own tells a brewer very little.

The units are a convention worth decoding. Alkalinity, like hardness, is usually stated as the weight of calcium carbonate that would have the same effect, in milligrams per litre; the text notes that it can equally be given in milliequivalents per litre, one of which is fifty milligrams per litre as calcium carbonate. The old national degrees of hardness are the same idea on different scales: one French degree is 10 milligrams per litre as calcium carbonate, one German degree 17.8 and one English degree 14.3.

What happened when they were tested apart

Sourced

Researchers at the University of Copenhagen brewed one Colombian coffee, a medium roast, in French presses with ultrapure water and with the same water after adding measured amounts of calcium, magnesium or bicarbonate. Untrained assessors, 31 in one session and 25 in another, were asked whether they could tell each brew from the one made with ultrapure water. Two panels of ten trained assessors then described the coffees in detail.

No brew made with added magnesium alone was told apart from the reference. Of the calcium brews, one was: the weakest, at 15 milligrams per litre. Stronger ones were not, though the strongest, at 240 milligrams per litre, came close to the threshold the authors used, and they say that isolated result should be read cautiously because the series showed no consistent trend with concentration. A trained panel then found no significant difference between coffee brewed with ultrapure water and coffee brewed with water carrying 132 milligrams of calcium and 80 of magnesium per litre. The authors conclude that hardness alone produced no marked change in flavour, including at 264 milligrams of calcium per litre, which they describe as unrealistically high for drinking water.

Bicarbonate was different. Brews made with 120 and with 240 milligrams of bicarbonate per litre were told apart from the reference, and assessors described them as less acidic. So was one brew that combined 60 milligrams of bicarbonate with the highest level of magnesium, which the authors attribute to the bicarbonate. The second trained panel tasted water with 300 milligrams of bicarbonate per litre and nothing else. Against ultrapure water, on a line 15 centimetres long, acidity fell from 12.0 to 2.9 and fruitiness from 10.2 to 2.4, complexity fell from 8.7 to 4.6, and bitterness and roasted character rose. The authors attribute this to bicarbonate buffering the acids in the brew. At 60 milligrams per litre, added alongside calcium and magnesium, the first trained panel found no significant difference from ultrapure water.

Two waters carried calcium and bicarbonate together. The one with 264 milligrams of calcium and 300 of bicarbonate was still significantly less acidic and fruity and more bitter than the reference, but less changed than bicarbonate alone; the authors measured less bicarbonate in it than they had added and put that down to some of it leaving the water as calcium carbonate. The one with half those amounts differed significantly from the reference only in acidity. The authors also describe a subtle rise in chocolate character and aftertaste where calcium accompanied a moderate amount of bicarbonate, and call calcium's contribution secondary.

What the same study measured in the cup

Sourced

The Copenhagen group also measured the acidity of the brews themselves, with three coffees. Coffee made with ultrapure water averaged pH 4.88. With 300 milligrams of bicarbonate per litre in the water it averaged 5.53; with the same bicarbonate and 264 milligrams of calcium, 5.17; with half of each, 5.02; with calcium alone, 4.71. By the paper's statistics the two high-bicarbonate waters differed from the reference and the other two did not. The pattern was the same in a Colombian, a Brazilian and an Ethiopian coffee.

So the bicarbonate in the water neutralised part of the coffee's acid, as the definition of alkalinity says it should, and the brew that tasted least acidic was the one that measured least acidic. The authors are careful about the rest. Bitterness and roasted character were also stronger in the high-bicarbonate brews, and they write that this may reflect a change in perception or in extraction, with the mechanism still to be clarified.

One detail limits the result. The bicarbonate was added as its sodium salt, so sodium rose with it, and nothing in the paper separates the two. A Japanese paper by one author claims that sodium itself matters: coffee brewed with very pure water from melted ice was given the most marks for a clean taste by six trained volunteers, and adding a little sodium to that water, as its citrate, lowered them, while a taste-sensing instrument read more saltiness. CoffeeHQ does not rely on it. The marks were ticks counted across six people, with no statistical test. The sodium added was about a fifth of a milligram per litre. Coffee brewed with laboratory-pure water, which has no sodium either, was marked no cleaner than coffee brewed with tap water, and the same melted-ice coffee got six ticks for clean in one session and four in the next. It is mentioned because sodium is the ion a household softener adds, and because nothing better was found.

The same paper measured something firmer on the way. In the four brews it analysed, made with waters carrying from under 1 to 30 milligrams of calcium and magnesium per litre, the coffee in the cup held about 80 milligrams of magnesium and 14 of calcium per litre. With the purest water all of that had come out of the coffee, and the water carrying 30 milligrams did not give a cup with more. The author concludes that most of those two minerals in the cup come from the bean and not from the water, which is worth holding beside any claim about what the magnesium in brewing water tastes of.

Do calcium and magnesium change what is extracted?

Sourced

The belief that magnesium pulls more flavour out of coffee, and fruitier flavour, comes from a 2014 calculation of how strongly each ion should bind to compounds found in coffee. A calculation predicts; it does not show that the binding happens in a brewer or that anyone can taste it.

A Swedish group tested the extraction directly. They brewed coffee with magnesium chloride or calcium chloride in the water, and separately added the same salts to coffee already brewed without them, then measured citric, malic, lactic and quinic acid by two instrumental methods. At a concentration typical of drinking water the salts made little difference to the acids. At ten times that, differences were larger. And in most cases it did not matter whether the salt went in before or after brewing: the acids in the cup were similar.

Their reading is that the acids are extracted independently of the water's composition, all of what is available leaving the grounds early in the brew, and that if the ions affect flavour it is more likely through what happens in the cup afterwards, to perception. They name the limits themselves: chloride salts only, no bicarbonate, four acids, and no tasting. They add that repeating the work with carbonate present would most likely give different results.

Put beside the Copenhagen tasting, in which magnesium alone was never told apart from ultrapure water, this is two studies by different methods failing to find the magnesium effect. Neither is espresso and neither looked at the less soluble compounds the Swedish authors suggest the ions might help. The claim that magnesium makes coffee fruitier is, on what has been read, a prediction that two tests did not confirm.

How far that result reaches

CoffeeHQ explanation

It is one coffee, one roast, one brewing method and panels in one laboratory, and the authors confine their conclusion to those conditions. It was not espresso. It also did not measure what was extracted, only what was tasted, and the authors say differences in extraction between the waters cannot be completely excluded.

It does bear on two common beliefs. One is that very soft or purified water brews a flat, under-extracted cup. In this study the reference brew was made with ultrapure water, and brews made with hard water were not reliably distinguished from it. The authors note that older studies from the 1950s and 1970s reported distilled water giving overly sour coffee, and that those studies lacked modern sensory methods.

The other is that magnesium draws more flavour out of coffee than calcium. That idea comes from a 2014 study which calculated how strongly each ion should bind to compounds found in coffee. The Copenhagen authors point out that it was a theoretical result with no tasting behind it, and report that a later experiment found neither ion changed the extraction of the main organic acids at realistic levels. That later experiment is described in the section above.

The recommended range, and what it may be for

Sourced

Coffee guidance often shows a target zone for brewing water. As the Copenhagen paper reports it, the standard most often cited defines that zone as roughly 17 to 85 milligrams per litre of calcium hardness, expressed as calcium carbonate, and 40 to 75 milligrams per litre of bicarbonate alkalinity. The standard itself was not opened for this page; the figures are given as that paper quotes them.

The authors say that, to their knowledge, no earlier study had tested the zone by controlled tasting. Their own conclusion is that the bicarbonate range remains meaningful to taste, that the narrow hardness limits are not supported by their sensory evidence, and that the ranges may reflect the needs of machines more than of flavour. That last point is their suggestion, not a finding.

Scale and corrosion: where hardness does matter

Sourced

Heating is what turns hard water into a maintenance problem. The World Health Organization's document says that, depending on its acidity and alkalinity, hard water can deposit scale in heated water applications, where insoluble metal carbonates form, coat surfaces and reduce the efficiency of heat exchange. A kettle element and the boiler of a coffee machine are such surfaces. Scale needs both the calcium or magnesium and the carbonate, which is why hardness and alkalinity are considered together for this purpose.

The opposite extreme has its own cost. The same document says that soft water which has not been stabilised has a great tendency to corrode metal surfaces and pipes, and that soft water has the benefit of minimal scaling. It is describing water supplies and household plumbing. The Copenhagen paper, citing the coffee industry's guidance, says low alkalinity may lead to corrosion in coffee machines.

What a particular machine tolerates is a question for its maker. Manuals state the water they expect and how to descale, and they differ.

A model: four waters, by hardness and alkalinity

CoffeeHQ explanation

A model for placing a water report. It crosses the two properties and says, for each corner, what the tasting study and the water-chemistry texts would lead one to expect. Real waters sit between the corners, and no boundary is given because none has been established for coffee.

Low hardness, low alkalinity

Purified water and very soft supplies. In the one tasting study this was the reference, and hard waters were not reliably told from it. It leaves no scale. The water texts warn that unstabilised soft water tends to corrode metal, which is a question for a machine's maker.

High hardness, low alkalinity

Calcium and magnesium present mostly with chloride or sulphate, not bicarbonate. In the tasting study, calcium without bicarbonate changed little, and the brew's measured acidity was not significantly different from the reference. With little carbonate to pair with, such water has less to form scale from than its hardness figure suggests.

Low hardness, high alkalinity

What an ion-exchange softener produces from hard water: the calcium and magnesium are swapped for sodium and the bicarbonate stays. On the tasting evidence this is the corner that changes the cup most, towards less acidity and fruit and more bitterness, while a hardness test strip reports it as soft.

High hardness, high alkalinity

Water from limestone country. It forms scale when heated, and some of the bicarbonate leaves the water with the calcium as it does so. In the tasting study the high-calcium, high-bicarbonate water still dulled acidity, though less than bicarbonate alone.

The rest of a water report

Sourced

What the other lines on a supplier's report or a bottle's label mean, and whether anything read connects them to coffee. Most have no coffee evidence at all, and the table says so.

The absence of evidence in this table is a statement about what CoffeeHQ has read, not a finding that these have no effect.

Four questions that get answered as one

CoffeeHQ explanation

Most confusion about water comes from running four separate questions together. A figure that answers one of them is routinely quoted as if it answered another.

What does the water extract?

A chemistry question, answered by measuring what is in the cup. One study, four acids: little effect of calcium or magnesium at ordinary levels.

What can people taste?

A sensory question, answered by panels. One study: alkalinity clearly, hardness hardly.

What does the water do to the machine?

An engineering question about scale and corrosion, answered by the water texts and by each maker for its own machine. Hardness and alkalinity together decide scale. The published target zone for brewing water may owe more to this question than to taste, as the tasting study's authors suggest.

What should be done to the water?

A practical question about filters, softeners and added minerals. It depends on the answers to the other three and on what comes out of the tap, and no treatment was tested in anything read.

Filters, purified water and adding minerals back

Widely documented

Tap water is disinfected, usually with chlorine. Some suppliers use chloramines instead, which the United States Environmental Protection Agency describes as disinfectants most commonly formed when ammonia is added to chlorine, used because they last longer as water travels through pipes. A carbon filter jug is the usual household answer to a chlorine taste; no test of how well such filters remove either disinfectant was read for this page.

Different treatments change different things, and it is worth knowing which. A softening cartridge that exchanges calcium and magnesium for sodium lowers hardness and leaves alkalinity where it was. Reverse osmosis and distillation remove nearly everything, hardness and alkalinity alike. A hand-held meter sold as a TDS meter estimates the total of dissolved material from how well the water conducts electricity, and cannot say how much of that total is hardness and how much is alkalinity. These are descriptions of how the methods work in general, written without a source open.

Some people brew with purified water to which minerals have been added back. This page gives no recipe for doing so. The one controlled tasting read here suggests the property worth attention is alkalinity, and a water supplier's published report is the simplest place to find a tap water's figure.

Reading a cup with water in mind

CoffeeHQ explanation

Coffee that tastes flat and dull whatever the recipe

Usually because: High alkalinity is one possible cause among several; stale coffee and a dark roast taste similar.

Change first: Brew the same coffee once with a bottled water of lower stated bicarbonate.

Water suitability helper

White deposits in the kettle or a machine asking to be descaled often

Usually because: Hard water with enough alkalinity to form scale when heated.

Change first: Follow the maker's descaling instructions, and ask the maker about filtration.

Cleaning and descaling

What this page gives no answer on water for

CoffeeHQ explanation

It names no hardness or alkalinity as right for coffee and gives no mineral recipe. It says nothing about espresso, where water also affects the crema and where no tasting study was read. For sodium, chloride, sulphate and silica it reports only that little or nothing was found. It gives no schedule for descaling and no threshold at which scale or corrosion begins. And it draws nothing about health from the water documents cited, which were read only for their definitions and their account of scale and corrosion.

What to do next

• Water suitability helper — Takes the figures from a water report and says what each one bears on, taste or equipment.
• Cleaning and descaling equipment — What scale and coffee residue each need, read from makers' own instructions.
• Coffee extraction — Strength and extraction, the two things a recipe changes before water comes into it.
• Water hardness test kit — What a test strip or drop kit can and cannot tell you about your tap water.