Does blending water carry oxygen into the spirit? Dissolved gases, deaeration, and the line missing from almost every water specification

Key takeaways

• Blending water carries dissolved gas, and purification does not remove it. Air-equilibrated water at 25 °C holds roughly 8 mg/L oxygen, 13 mg/L nitrogen and 0.05 mg/L argon, and “dissolved oxygen and nitrogen and inert gases will pass straight through reverse osmosis membranes, UV light, micro and ultra-filters, ion exchange resins and electro-deionisation (EDI) units” (ELGA LabWater).
• Carbon dioxide is the exception, and it is why your CoA reads acidic. Deionised water “immediately begins absorbing CO₂ as soon as it comes into contact with air, lowering the pH to slightly less than 7” (Distiller magazine).
• The arithmetic is real but small. Reducing a 63.5% ABV spirit to 40% makes water about 37% of the final volume, so fully air-saturated blending water delivers roughly 3 mg of oxygen per litre of bottled whisky — about one O₂ molecule per 70,000 ethanol molecules.
• Oxygen alone barely touches ethanol. It has to be activated by trace iron and copper via the Fenton reaction before hydroxyl radicals can oxidise ethanol to acetaldehyde (Danilewicz, J. Agric. Food Chem., 2010). A blending water specification that removes those metals removes the catalyst, not just the taint.
• Other categories take it seriously. Wine is usually sparged below 0.5 mg/L dissolved oxygen before bottling (AWRI), and beverage water deaeration equipment is specified in parts per billion (Deutsche Beverage).

Yes — blending water carries dissolved oxygen, nitrogen, argon and carbon dioxide into the tank, because none of the unit operations that make water pure are designed to take gases out. In a standard whisky reduction that amounts to roughly 3 mg of oxygen per litre of finished spirit, which is chemically almost inert at 40% ABV unless catalytic metals are present. It becomes a genuine specification issue in lower-strength, higher-substrate products — RTDs, liqueurs, cream liqueurs and zero-alcohol spirits — and in the very different problem of gas breaking out of solution on a filling line.

What gases are actually dissolved in blending water?

The same ones that are in the air above it, in proportions set by Henry’s law rather than by the atmosphere’s composition. Typical levels at 25 °C in air-equilibrated water are “13ppm nitrogen, 8ppm oxygen and 0.05ppm argon” (ELGA LabWater). Nitrogen dominates by mass; oxygen is the reactive one.

Solubility is strongly temperature-dependent. At 100% air saturation and sea level, fresh water holds about 10.92 mg/L at 4 °C and 9.03 mg/L at 21 °C (Fondriest); at 25 °C, 8.24 mg/L (BioNumbers, Harvard). The USGS publishes the full table and a calculator for any temperature, pressure and salinity (DOTABLES).

Two consequences follow. Cold water holds more gas, so a chilled blending water tank carries a bigger gas load than a warm one — and warming that water, or dropping its pressure downstream of a pump or valve, pushes gas back out of solution. That is where bubbles come from.

Why doesn’t reverse osmosis or deionisation remove them?

Because those processes separate on charge and size, and dissolved gases are small, neutral molecules. RO membranes reject ions well and gases poorly; ion-exchange resins have nothing to trade with an O₂ or N₂ molecule; UV is disinfection, not separation. The gases “pass straight through” all of them (ELGA LabWater) — the same structural blind spot we set out in what reverse osmosis does not remove.

Carbon dioxide is the interesting exception, because it ionises. In water it forms carbonic acid, and an anion resin will strip it out as bicarbonate and carbonate — ELGA note that CO₂ “is removed by ion exchange”, and that if resistivity approaches 18.2 MΩ·cm the CO₂ concentration must be very low (ELGA LabWater). RO on its own does not do this well, which is why RO permeate typically comes out slightly acidic.

That explains a number which puzzles a lot of first-time buyers. DI water “immediately begins absorbing CO₂ as soon as it comes into contact with air, lowering the pH to slightly less than 7 and making it slightly acidic”, and “RO water has a similar pH to DI water” (Distiller). A certificate of analysis reading pH 5.5–6.5 on a near-zero-TDS water is not a contaminated water; it is an unbuffered water that met the atmosphere — as we unpack in pH and TDS specs explained and how blending water is tested.

How much oxygen does proofing water actually put into a bottle?

Enough to calculate, not usually enough to taste. Work it through for a cask-strength malt at 63.5% ABV reduced to 40% ABV. Ignoring the volume contraction that ethanol and water undergo on mixing — covered separately in heat of mixing and temperature — water is about 37% of the final volume. At 8.3 mg/L in the water, that is about 3.1 mg of oxygen per litre of bottled whisky.

Push the same arithmetic to a neutral-spirit category and the water fraction rises sharply: reducing 96.4% ABV neutral spirit to a 40% vodka makes water roughly 58% of the bottle, delivering close to 5 mg/L. That fits the general rule that “proofing water can make up as much as 50–60% (or more, in some cases) of the final bottled product” (Distiller) — expanded on in what is blending water.

Now scale it against the spirit itself. Three milligrams of O₂ is 0.094 mmol; ethanol at 40% ABV is about 6.85 mol/L. That is roughly one oxygen molecule for every 70,000 molecules of ethanol. If every one of those oxygen molecules reacted through to acetaldehyde — the stoichiometric ceiling, not a prediction — it would generate under 10 mg/L of acetaldehyde in a spirit that already carries it as a normal distillation congener.

Two caveats keep this honest. The spirit arriving at the tank is usually air-equilibrated too, so the water is topping up a system already in contact with air rather than contaminating a protected one. And oxygen is more soluble in ethanol than in water, with a decidedly non-linear curve across water–ethanol mixtures (Cargill, J. Chem. Soc., Faraday Trans. 1, 1976, 72, 2296); the finished blend re-equilibrates with whatever headspace it meets.

Does that oxygen actually change the whisky?

Mostly no, and the mechanism explains why. Molecular oxygen is a poor direct oxidant for ethanol at ambient temperature. In the best-characterised system — wine — oxidation is initiated when iron and copper in their lower oxidation states transfer an electron to O₂, producing hydrogen peroxide, which then undergoes the Fenton reaction to yield hydroxyl radicals; those radicals, not oxygen itself, oxidise ethanol to acetaldehyde (Danilewicz, J. Agric. Food Chem., 2010; Frontiers in Chemistry, 2018).

Read that mechanism next to a blending water specification and something clicks into place. The reason a reduction water is held to parts-per-billion on iron, copper and manganese is usually explained in terms of haze, colour and metallic off-notes — all true, and set out in trace metals in blending water. But those same metals are the catalysts that make dissolved oxygen reactive in the first place. A water clean enough on metals largely disarms the oxygen it carries. That is a far more efficient control than trying to remove the gas.

A whisky also has little of what wine has a lot of: no added sulphur dioxide, a far smaller pool of oxidisable polyphenols, and years already spent in an oak cask in deliberate, slow contact with air — the process described in how blended whisky is made. A few milligrams per litre at the proofing stage is a small perturbation on that history.

The bottle bears this out. Oxygen ingress through a screw cap is reported at “below 1 µL/day”, against 2–6 µL/day for a natural cork in its first year (Whisky Science) — and even then, “some oxidation reactions are bound to happen between the spirit and the air of the bottle headspace, but they are hardly significant.” An unopened, well-sealed bottle of whisky is not quietly turning into vinegar.

Where does dissolved gas genuinely matter, then?

In four places, none of which is the flavour of a full-strength whisky.

Lower-strength, higher-substrate products. Patent literature on beverage deaeration is blunt about the category risk: “in liquors such as refined sake, wine, and distilled spirits, there may occur deterioration in flavor, discoloration, and the like due to the effect of oxygen remaining dissolved in the drink” (US 5,522,917). The exposure rises as strength falls and as fruit, dairy, botanical or sugar content rises — which is why we treat dissolved oxygen as a live specification line in RTD and canned cocktail water, liqueur and cream liqueur water and zero-alcohol spirits water.

Filling and fill-level accuracy. Gas breaking out of solution at the filler head causes foaming, inconsistent fills and headspace variation — a mechanical problem, not a chemical one, and worse with warm water, long pipe runs and pressure drops.

In-line measurement. Entrained gas corrupts density-based strength verification. A Coriolis meter measures whatever is in its tubes, so bubbles make it report the density of a gas–liquid mixture rather than of the liquid — an error mode significant enough that the major instrument vendors publish dedicated guidance on it (Emerson Micro Motion; Endress+Hauser). For anyone dosing water in-line rather than batching it — the two approaches compared in batch vs in-line dilution — gas-free water is part of getting the ABV right, and ABV measurement has enough complications already, as how whisky strength is measured sets out.

Storage and handling. Water in a vented tank re-equilibrates with air continuously, so any deaeration paid for upstream is lost unless the tank is blanketed — one more reason storage matters as much as purification (blending water storage and handling).

What do industries that care about oxygen actually do?

Three things, all of which are available to a distillery.

Sparge with an inert gas. Wine practice targets dissolved oxygen “generally < 0.5 mg/L” before bottling, using nitrogen at 0.1–0.3 litres of gas per litre of wine with around 30 seconds’ contact (AWRI). The same source notes wine can pick up as much as 4 mg/L through routine tank movements — handling, not the water, is often the larger contributor.

Use a membrane contactor. Hollow-fibre membrane deaerators with vacuum and a nitrogen or CO₂ sweep gas are the beverage industry standard, “capable of achieving very low dissolved gas levels down to 1 ppb O₂ at outlet with proper system design”, though at 20 °C single-digit ppb is “generally not achievable” with a single contactor (Deutsche Beverage). Note the units: the beverage industry specifies deaerated water in parts per billion, three orders of magnitude below what air-saturated water carries.

Mix with nitrogen instead of air. Some proofing tanks are agitated by pulsed injection of compressed nitrogen rather than an impeller — an inert gas chosen partly to avoid oxidation and partly because it minimises foaming during dilution (Pulsair). It is a neat example of a mixing decision that doubles as an oxygen-management decision.

Should dissolved oxygen be on a blending water specification?

For a full-strength brown spirit, usually not as a numeric limit — the chemistry does not justify the cost, and the metals limits already do the important work. What belongs on the specification is the honest statement that the water is supplied air-equilibrated and that dissolved gas is not controlled, so nobody assumes otherwise.

For RTDs, liqueurs, cream liqueurs, low-ABV and zero-alcohol products — or any line where foaming, fill accuracy or in-line density measurement is a known headache — dissolved oxygen deserves a target, a measurement point and a deaeration step sized for it. The question to ask a supplier is not “is your water pure?” but “at what point is the water last in contact with air, and what happens to it after that?” That is a different question from the ones in demineralised water specifications for spirit dilution, and the one most specifications forget to ask.

Glossary

• Dissolved oxygen (DO) — oxygen present in molecular form in solution, normally reported in mg/L (ppm) or µg/L (ppb). Air-equilibrated water at 25 °C holds about 8.2 mg/L.
• Henry’s law — the principle that the concentration of a dissolved gas is proportional to its partial pressure above the liquid. It is why gas solubility falls as temperature rises and as pressure drops.
• Deaeration — the deliberate removal of dissolved gases from a liquid, typically by vacuum, membrane contactor or inert-gas stripping.
• Membrane contactor — a hollow-fibre device that puts liquid and a vacuum or sweep gas on opposite sides of a microporous membrane, allowing dissolved gas to transfer out without the phases mixing.
• Sparging — bubbling an inert gas (usually nitrogen) through a liquid so dissolved oxygen partitions into the bubbles and leaves with them.
• Blanketing (inerting) — maintaining an inert gas layer above a liquid in a tank or bottle headspace so the liquid cannot re-absorb oxygen.
• Fenton reaction — metal-catalysed reduction of hydrogen peroxide to the hydroxyl radical, the step that makes dissolved oxygen chemically dangerous in a beverage.
• Acetaldehyde — ethanal, the first oxidation product of ethanol; a normal distillation congener at low levels, and a marker of oxidation at high ones.

Frequently asked questions

Does demineralised water contain oxygen? Yes. Unless it has been deliberately deaerated and then protected from air, demineralised water holds essentially the same dissolved oxygen as any other water at that temperature — roughly 8 mg/L at 25 °C. Reverse osmosis, ion exchange, EDI, UV and filtration all leave dissolved oxygen and nitrogen untouched (ELGA LabWater).

Why does my blending water certificate show a pH below 7? Because pure, unbuffered water absorbs atmospheric carbon dioxide, which forms carbonic acid. DI water “immediately begins absorbing CO₂ as soon as it comes into contact with air, lowering the pH to slightly less than 7” (Distiller). With essentially no dissolved solids to buffer it, a small amount of CO₂ moves the reading a long way — and the whisky it is blended into, which sits in a pH band of roughly 3.5–4.7 (Chemosensors), sets the final pH regardless.

Will dissolved oxygen in proofing water oxidise my whisky? To a negligible degree in a full-strength spirit with clean water. Oxygen needs catalytic iron or copper to be converted into the radicals that actually attack ethanol (Danilewicz, 2010), and a blending water held to ppb limits on those metals removes the catalyst. The arithmetic ceiling for a standard reduction is under 10 mg/L of acetaldehyde, and the real figure is far below that.

Does it matter more for vodka or gin than for whisky? The water fraction is larger — around 58% reducing 96.4% ABV neutral spirit to 40% — so the oxygen load per litre is larger. But there is less oxidisable substrate than in a matured spirit, so the practical risk is foaming, fill accuracy and any trace metal carried in, rather than flavour change. See proofing water for gin, vodka, rum and tequila.

What dissolved oxygen level should an RTD or liqueur specify? Follow the packaging category rather than the spirits category. Wine is sparged to below 0.5 mg/L before bottling (AWRI) and beverage-water deaeration systems are specified in ppb (Deutsche Beverage). Set the target against shelf-life testing for your own formulation, and measure at the filler, not at the water skid.

Can you remove dissolved oxygen by boiling the water? Boiling drives out dissolved gases and is a standard laboratory approach, but it is not a production solution: the water re-absorbs oxygen as it cools in contact with air, the energy cost is significant, and a membrane contactor or nitrogen sparge does the same job better and continuously.

Blenders International · Water Science series