Is reverse osmosis water good enough for blending water? Five things an RO membrane leaves behind

Key takeaways

• RO is excellent, and it is not absolute. A properly running system removes “95-99% of dissolved salts (ions), particles, colloids, organics, bacteria, and pyrogens,” and “any contaminant with a molecular weight greater than 200 will likely be rejected” (Puretec). Rejection turns on charge and size — so uncharged, low-mass species are the exceptions.
• Dissolved gases pass almost freely. RO “cannot remove dissolved gases, such as carbon dioxide (CO2), very well because they are not highly ionized (charged) while in solution and have a very low molecular weight” (Puretec). Hence RO is “often followed by ‘Degasification’” (DeLoach Industries).
• Boron is the ion RO is worst at. Boric acid is uncharged below its pKa of 9.2, and membranes are “very efficient at removing charged species like the borate ion rather than neutral molecules like boric acid” — typical removal at pH 8 is 73–90% (Lenntech).
• Small organics are the ones you could taste. In a laboratory trial, RO removed 83.8% of trihalomethanes and 66.0% of total organic carbon, with removal “more for higher molecular weight organics” (Ahmadi et al.) — the small ones being the volatile, odorous ones.
• The fix is a train, not a better membrane. Carbon ahead of the RO, degasification and polishing behind it: “Electrodeionization (EDI) is used after reverse osmosis for polishing of demineralized water to obtain low levels of conductivity and silica” (EUROWATER).

Not on its own, no — and the reason is worth understanding rather than taking on trust. A reverse-osmosis membrane rejects contaminants according to their electrical charge and molecular size, so it is superb against dissolved minerals and nearly blind to anything that is both uncharged and small. For water destined to reduce a matured spirit to bottling strength, that leaves a short, specific list to close out: dissolved gases, boron, part of the silica, the low-molecular-weight organics that carry odour, and the microbiology that grows downstream of the membrane rather than passing through it.

What does reverse osmosis actually remove, and how well?

A great deal, which is why it anchors almost every blending water plant in the industry. The headline figure is that RO removes “95-99% of dissolved salts (ions), particles, colloids, organics, bacteria, and pyrogens from feed water” (Puretec).

The mechanism matters more than the number, because it predicts the exceptions. Two properties govern rejection:

• Charge. “The greater the ionic charge of the contaminant, the more likely it will be unable to pass through the RO membrane” (Puretec). Calcium and magnesium carry two charges each and are rejected almost completely — which is precisely why RO solves the calcium haze problem that makes hard water unusable for reduction.
• Size. “Any contaminant with a molecular weight greater than 200 will likely be rejected by a properly running RO system” (Puretec).

Read those two rules together and the gap becomes obvious. Anything that is electrically neutral and lighter than roughly 200 daltons is, from the membrane’s point of view, close to being water. Carbon dioxide is 44. Boric acid is 62. Chloroform is 119. None of them are minerals, and none of them are what an RO plant was bought to remove.

Why does RO permeate still contain carbon dioxide?

Because CO₂ fails both tests at once. RO “cannot remove dissolved gases, such as carbon dioxide (CO2), very well because they are not highly ionized (charged) while in solution and have a very low molecular weight,” and the consequence shows up in the permeate: “permeate water can have a slightly lower than normal pH level, depending on dissolved CO2 in the feed water since CO2 is converted to carbonic acid” (Puretec). Membrane filtration is simply the wrong tool: “Membrane filtration only removes solids from the water so any free form of gas or residual solid that is converting into a gas will still be present” (DeLoach Industries).

For a buyer reading a certificate, this creates a genuinely misleading measurement. Carbonic acid partially ionises, so dissolved CO₂ raises the conductivity reading — the single number most blending-water specifications are built around. A permeate reading a few microsiemens per centimetre may be carrying almost no mineral content at all, with most of the signal coming from a gas that will contribute nothing whatsoever to the taste of a whisky. The reverse trap is worse: a plant chasing a conductivity target can find itself unable to hit it no matter how good the membrane is, because the number it is fighting was never a salt.

CO₂ also loads the equipment behind the membrane, because “high concentrations of CO2 consume the ion charge within the resins” on any downstream ion-exchange stage. The remedy is a dedicated gas-removal step rather than a better membrane: “a properly designed decarbonator can remove 99.99% of the free carbon dioxide gas that is present in the water stream” (DeLoach Industries).

This is a different problem from the CO₂ a finished water picks up later in a vented tank, covered in our guide to storage and purity drift. This is gas that was never removed in the first place.

Why is boron the ion an RO membrane handles worst?

Because at ordinary pH, boron is not an ion. Boron in water exists as boric acid or as the borate anion, and the split between them is set by pH: “The pKa of H3BO3/H3BO2- is 9.2, therefore the equilibrium H3BO3 ⇌ H3BO2- + H+ is typically towards the left at standard seawater pH 8” (Lenntech). Below about pH 9, in other words, boron is overwhelmingly present as a small neutral molecule.

That is exactly the species a membrane cannot see. “Reverse Osmosis membranes are very efficient at removing charged species like the borate ion rather than neutral molecules like boric acid,” and the resulting performance gap is large: “Typical Boron removal rates at pH 8 are between 73 and 90% for standard High Rejection Seawater Reverse Osmosis membranes, depending on the water temperature” (Lenntech). Against 99%-plus rejection of sodium and chloride on the same membrane, 73% is a different order of performance entirely. The engineering answer is to change the chemistry rather than the membrane — a second-pass RO “with caustic soda addition to raise pH around 9.5” converts boric acid to borate, which the membrane then rejects normally.

For blending water this is a feedwater-dependent concern rather than a universal one. Boron travels with coastal and brackish groundwater; a distillery drawing on inland surface water will rarely see enough of it to matter, while a bottler on a saline-influenced aquifer might. The point for a trade buyer is narrower and more useful than “boron is bad”: if your supplier’s feedwater carries boron, single-pass RO is not the barrier you assumed it was, and the certificate should say so.

What about silica?

Silica sits in between, and it splits into two forms that behave nothing alike: “Silica exists in two forms: dissolved (reactive) and colloidal (precipitated)” (Genesis Water Technologies). Colloidal silica behaves as a particle and is straightforwardly removed. Reactive silica is weakly ionised and much harder work — it “is the primary cause of RO membrane surface fouling and isn’t easily removed by standard pretreatment,” and while RO handles most of it, “some may still pass through.”

Silica has no known flavour impact at the concentrations in question, so this is not a taste argument. It is a specification and reliability argument — and it is why silica appears alongside conductivity as a target for the polishing stage rather than the membrane.

Which organics slip through — and are they the ones that matter?

This is the item on the list with a real flavour consequence, and the arithmetic runs the wrong way for whisky.

RO’s rejection of organics is size-dependent, so the compounds it rejects best are the large, heavy, non-volatile ones — and the compounds it rejects worst are the small, light, volatile ones. In a laboratory-scale study passing chlorinated drinking water through microfiltration, activated carbon and an RO membrane, the RO stage removed 83.8% of trihalomethanes and 66.0% of total organic carbon, with the explicit observation that “the removal efficiency of RO membrane is more for higher molecular weight organics” (Ahmadi et al.). Work on domestic units found chloroform removal reaching only about 80% under the least favourable conditions tested (Iran. J. Environ. Health. Sci. Eng. 2009, 6(4):301–306).

Eighty per cent removal sounds respectable until you set it against the concentrations at which taste-active compounds register. As covered in our review of whether water type changes whisky’s taste, chlorophenols are detectable around 0.1 µg/L — roughly a million times below where calcium becomes tasteable. A contaminant class present at even a few micrograms per litre in the feed, reduced by 80%, is still comfortably above its own sensory threshold in the permeate.

So the defence against odorous organics is not the membrane at all — it is the carbon stage in front of it: “Prior to the reverse osmosis system, chlorine and other organics are removed by a backwashing carbon filter and if the water contains iron/manganese or sulfur, it must also be removed” (US Water Systems). A distillery that treats carbon as membrane protection has understood half of its job. It is also the flavour barrier.

Is an RO membrane a microbiological barrier?

Functionally excellent, formally not absolute — and the distinction matters because the failure mode is downstream rather than through the membrane. Bacterial passage across intact RO membranes is an active research question in its own right (Environ. Sci.: Water Res. Technol., “Integrity of reverse osmosis membrane for removing bacteria: new insight into bacterial passage”), but the practical risk in a bottling hall is regrowth in the permeate loop — pipework, tanks and hoses, fed by the organic carbon the membrane did not reject. That is the argument set out at length in why blending water is pure but not sterile, and it is why the trade puts disinfection at the point of use rather than relying on the membrane: in a typical distillery train “there is an ultraviolet (UV) light in-line to destroy any bacteria” (US Water Systems).

So what closes the gap?

A train, not a better membrane. Each gap above has an established, unglamorous answer:

Two of those rows sit in the same box. “Electrodeionization (EDI) is used after reverse osmosis for polishing of demineralized water to obtain low levels of conductivity and silica,” and the two problems interact — “Removal of CO2 after reverse osmosis and before EDI can improve the performance of the EDI significantly and keep the silica content low” (EUROWATER). The classical alternative is resin: post-treatment “such as mixed bed deionization, can increase RO permeate quality and make it suitable for the most demanding applications” (Puretec). In distilling terms, that is the difference between RO water and finished water — “Some distilleries opt to put the water from the RO system through a deionization (DI) system which results in absolutely pure water with a pH of 7.0” (US Water Systems).

Which is the underlying point. “We have reverse osmosis” describes a stage, not a specification. Whether the plant is yours or a supplier’s — a decision covered in make it or buy it — the questions worth asking are about the stages either side of the membrane.

What should a buyer ask beyond “is it RO water?”

1. Is there carbon pretreatment, and is it monitored? This is the flavour barrier, not the membrane.
2. Is there a polishing stage — EDI or mixed-bed DI — behind the RO? Single-pass permeate is not finished water.
3. Was conductivity measured before or after degassing? The same water gives two different numbers.
4. Does the feedwater carry boron? If it is coastal or brackish groundwater, ask for the permeate figure, not the rejection rate.
5. What is the microbiological control at the point of use? Regrowth happens after the membrane.

Glossary

• Rejection — the proportion of a contaminant a membrane keeps out of the permeate, governed chiefly by ionic charge and molecular size.
• Permeate — the purified water passing through the membrane; the stream carrying the rejected salts is the concentrate or reject.
• Boric acid / borate — the neutral and charged forms of dissolved boron, in equilibrium around pKa 9.2. RO rejects the charged form far better.
• Reactive (dissolved) silica — weakly ionised silica in solution, distinct from colloidal silica; the form that both fouls membranes and partially passes them.
• Trihalomethanes (THMs) — low-molecular-weight chlorination by-products including chloroform, small enough to partially cross an RO membrane.
• Decarbonation / degasification — a dedicated stage stripping dissolved CO₂ (and H₂S), which membranes cannot do.
• EDI (electrodeionization) — a continuously regenerated polishing stage after RO, driving down conductivity and silica without regeneration chemicals.

Frequently asked questions

Is RO water the same as demineralised water? Not quite. RO removes 95–99% of dissolved salts (Puretec), which is most of the way there, but a demineralised specification for spirit dilution is normally met by RO plus a polishing stage — a distinction set out in blending water vs RO vs distilled.

Does the CO₂ left in RO permeate affect a whisky? Not in flavour terms at these concentrations — but it depresses permeate pH and inflates the conductivity reading, so it can make a perfectly good water fail a specification.

Should I worry about boron in my blending water? Only if your feedwater carries it, which mainly means coastal or brackish groundwater. Where it is present, single-pass RO removes 73–90% at pH 8 rather than the 99%-plus you might assume (Lenntech).

If RO does not fully remove chlorine by-products, what does? Activated carbon, positioned before the membrane — the standard arrangement in distillery water treatment (US Water Systems). RO’s organic rejection improves with molecular weight, so the small volatile compounds that matter to aroma are the ones it handles least well.

Does a second RO pass fix all of this? It helps with boron, especially with caustic dosing between passes to raise pH (Lenntech), and it lowers conductivity further. It does not remove dissolved gases — the physics is unchanged, so that still needs degasification or EDI.

Blenders International · Water Science series