What is your blending water touching? Food-contact materials, migration limits, and the seal nobody specifies

Key takeaways

• Migration is a legal limit, not a guideline. The EU caps total transfer from plastics at 10 mg/dm² of contact surface (Regulation (EU) No 10/2011, Article 12); India’s FSSAI sets the same ceiling as 60 mg/kg or 10 mg/dm² tested per IS 9845 (FSS (Packaging) Regulations, 2018).
• Polyethylene does release identifiable compounds into water. A study of drinking-water pipe found 2,4-di-tert-butylphenol (2,4-DTBP) as a major migrant from HDPE, with threshold odour numbers ≥4 in five of seven brands tested (Skjevrak et al., Water Research, 2003).
• The seals matter more than the pipe. Non-target screening of a Copenhagen distribution system detected 133 compounds from HDPE in the laboratory, but in the live system “a larger proportion of the compounds detected were assigned to rubber seals” (Diera et al., Water Research, 2023).
• An NSF/ANSI 61 mark is a health certification, not a flavour certification. The standard’s own purpose clause states that it “does not establish performance, taste and odor, or microbial growth support requirements” (NSF/ANSI 61, §1.1). UK approval under BS 6920 does cover “odour and flavour” (Drinking Water Inspectorate), measured as the threshold numbers of EN 1622.
• Strength changes which rules apply. EU migration testing uses 10% ethanol for aqueous foods but 50% ethanol (simulant D1) above 20% ABV (Annex III) — so the water line and the reduced-spirit line downstream of it are held to different tests.

A blending water specification describes the water: conductivity, total organic carbon, metals, microbiology. It almost never describes what the water is held in. Yet between the polisher outlet and the reduction tank, demineralised water sits in polyethylene or stainless steel, passes through a flexible hose, and crosses perhaps a dozen elastomer seals — and every one of those surfaces is a food-contact material with its own regulated migration limit. This post sets out which law applies, what has actually been measured coming out of these materials, and why the component that matters most is usually the one nobody puts on a drawing.

Why does the container matter more for blending water than for other ingredients?

Because purified water is chemically unsaturated, and that is not a defect but the point of it. Water carrying almost no dissolved matter has spare capacity to take material into solution, so it dissolves more from the surfaces it touches than a mineral-rich water would. We have covered the metal side of this in what storage and pipework do to a blending water specification; this post is about the organic side, which is far harder to see.

The second reason is specific to spirits. A blending water’s entire value proposition is that it contributes strength and nothing else — the argument set out in demineralised water specifications for spirit dilution. Any organic compound picked up from a hose is, by definition, a contribution the water was bought not to make. And because demineralised water has no mineral character to mask it, a trace that a spring water would hide arrives in the tank perfectly audible — the same masking asymmetry described for trace metals.

Which law applies — food-contact law or drinking-water law?

Both, at different points in the system, and the distinction is worth getting right because the two regimes test for different things.

Food-contact law governs materials in contact with food. In the EU the framework is Regulation (EC) No 1935/2004, with plastics handled in detail by Regulation (EU) No 10/2011; in India, the Food Safety and Standards (Packaging) Regulations, 2018 do the equivalent job; in the United States, polymers are regulated as indirect food additives under 21 CFR Part 177, with polyethylene and polypropylene at 21 CFR 177.1520, which sets maximum extractable fractions determined by n-hexane extraction.

Drinking-water law governs materials in contact with water in supply systems, and has recently been consolidated in Europe. Article 11 of the Drinking Water Directive (EU) 2020/2184 sets minimum hygiene requirements for such materials, implemented from April 2024 by six legal acts including Commission Implementing Decision (EU) 2024/367, which establishes the first European positive lists of starting substances. The North American route is NSF/ANSI/CAN 61; the UK route is approval under Regulation 31 using BS 6920.

Blending water sits across the boundary. It arrives as water — where the drinking-water regime and its test method EN 12873-1 apply — and it is consumed as an ingredient of a food, where food-contact law applies. A supplier who can answer for only one of the two has answered half the question.

What do the migration limits actually say?

The headline number is remarkably consistent across jurisdictions.

The EU’s overall migration limit (OML) is stated plainly: plastic materials “shall not transfer their constituents to food simulants in quantities exceeding 10 milligrams of total constituents released per dm² of food contact surface” (Article 12). India’s rule reads almost identically — “All packaging materials of plastic origin shall pass the prescribed overall migration limit of 60 mg/kg or 10 mg/dm² when tested as per IS 9845 with no visible colour migration” (FSS (Packaging) Regulations, 2018).

Alongside it sit specific migration limits (SMLs) for named substances. The Indian regulations set these for seven metals, and the list will look familiar to anyone who has read a water certificate:

Source: FSS (Packaging) Regulations, 2018, Table 1.

The same regulations point at specific BIS material standards in Schedule III — polyethylene to IS 10146, PVC to IS 10151, PET and PBT to IS 12252, polypropylene to IS 10910, polycarbonate to IS 14971, polystyrene to IS 10142.

Read against a drinking-water standard, these limits are generous. IS 10500:2012 permits manganese in drinking water at 0.1 mg/L (IS 10500:2012), while the plastics migration ceiling for manganese is 0.6 mg/kg — six times higher. That is not a loophole; it is the correct reading of two limits written for different purposes. But it means “food-grade” is a floor, exactly as IS 10500 is a floor rather than a proofing specification.

Why does the alcohol strength change the test?

Because ethanol is a better solvent for the substances that migrate out of polymers, and the regulations say so explicitly. Annex III of Regulation (EU) No 10/2011 assigns food simulant A (ethanol 10% v/v) to hydrophilic foods, simulant C (ethanol 20%) to “alcoholic foods with an alcohol content of up to 20%”, and simulant D1 — ethanol 50% v/v — to “alcoholic foods with an alcohol content of above 20% and for oil in water emulsions” (Annex III). A 40% ABV whisky is tested with the strongest aqueous simulant in the set.

For a plant, that has a practical consequence that is easy to miss. The blending water line carries water, so it is assessed as a water-contact material. The line immediately downstream of the reduction tank carries a 40% ABV liquid, and the same grade of hose in that position is being asked to resist a far more aggressive extractant. Materials entirely suitable for the water side are not automatically suitable for the spirit side, and a bill of materials written once for “the whole system” can quietly get this backwards.

What has actually been measured coming out of polyethylene?

Enough to take it seriously, and the most-cited work is nearly a quarter-century old. Skjevrak and colleagues tested HDPE, cross-linked polyethylene (PEX) and PVC drinking-water pipe for migration of volatile organic compounds, and assessed the resulting water using the threshold odour number (TON) concept. A major migrating component from HDPE was 2,4-di-tert-butylphenol (2,4-DTBP) — a known degradation product of phosphite antioxidants such as Irgafos 168 — alongside a range of esters, aldehydes, ketones, aromatic hydrocarbons and terpenoids. Water in contact with five of seven HDPE brands returned TON values of 4 or higher: a significant, panel-detectable odour (Skjevrak et al., Water Research 37(8): 1912–1920, 2003).

The contemporary picture is broader. Diera and colleagues ran a two-tier non-target screening study on HDPE pipe used for drinking water in Denmark, following modified EN 12873-1 protocols with ten-thousand-fold sample enrichment. In the laboratory tier they detected 133 compounds originating from the pipe — 31 by suspect screening, 102 by non-target screening — including hindered amine light stabilisers and the flame retardant and plasticiser tris(2-chloroethyl) phosphate (Diera et al., Water Research, 2023).

A caveat belongs here. These are pipe-grade materials tested under standardised migration conditions, not blending water systems, and detection at these enrichment levels is not the same as an off-flavour. The point is not that polyethylene ruins whisky; it is that the list of things capable of crossing from a polymer into low-conductivity water is long, real, and absent from every blending water certificate we have seen.

Why do the gaskets matter more than the tank?

This is the finding that should change how a specification is written.

In the second tier of the same Danish study, the researchers sampled an actual Copenhagen distribution system fed through newly installed HDPE pipe. They detected 51 compounds in the water, 12 attributable to the HDPE. The surprise was the ranking: “HDPE antioxidants and their degradation products contributed only a relatively small percentage of the total measured compound intensities.” Instead, “a larger proportion of the compounds detected were assigned to rubber seals, used upstream in the water system from the abstraction site to delivery at the consumer tap.” The authors close with an unusually direct sentence for a journal abstract: “Seals are considered trifle in the larger picture of materials in contact with drinking water, however these results may cause a reconsideration of this position” (Diera et al., 2023).

Independent work points the same way. A 2025 study validating an LLE-GC-MS method for eleven nitrosamines in leachates from drinking-water contact materials found that a flexible rubber seal was the material that released them — N-ethyl-N-nitrosobenzenamine at 267 µg/L and N-methyl-N-nitrosobenzenamine at 106 µg/L in first-migration chlorinated leachate — while a bi-component polyurethane membrane gave only trace detections. Crucially, all materials complied: scaled to a consumer tap, the rubber seal’s estimated contribution was 0.070 µg/L against a 0.3 µg/L limit for total nitrosamines (Antunes et al., Molecules, 2025).

That contrast — large numbers in the first laboratory leachate, compliant numbers at the tap — is exactly why elastomers get overlooked. Their contact area is trivial, so nobody specifies them; but their extractable load per unit area is high, and unlike the tank they are replaced at short notice from whatever is in the stores. A pipe is specified once by an engineer. An O-ring is replaced fifty times by whoever is on shift.

Does a certification mark mean the water will not taste of anything?

Usually not, and this is the single most common misreading of a compliance document.

NSF/ANSI/CAN 61 evaluates chemical contaminants leaching from products into drinking water and covers pipes, fittings, hoses, gaskets, adhesives, lubricants, coatings and treatment devices. Its own purpose clause states the exclusion: the standard “does not establish performance, taste and odor, or microbial growth support requirements for drinking water system products, components, or materials” (NSF/ANSI 61, §1.1). A component can be fully NSF 61 certified and still impart a detectable odour, because odour was never what was being assessed.

The British route does test it. Under Regulation 31, non-metallic materials are assessed to BS 6920, whose test methods and acceptance criteria cover “odour and flavour, appearance, growth of aquatic microorganisms, leaching of metals, and cytotoxicity” (Drinking Water Inspectorate), with dedicated parts for hoses and for hoses conveying water used in food and drink preparation. The underlying sensory measurement is EN 1622, which determines the threshold odour and flavour numbers — the dilution ratio at which a sample stops being perceptible — and applies both to drinking water and to “migration waters from materials in contact with waters” (EN 1622:2006).

For a spirits producer, the organoleptic test is the one that matters. A health-effects certification answers whether the water is safe; a threshold odour number answers whether it is neutral — which, in a liquid bought precisely for its neutrality, is the whole specification. It is the same logic that underpins sensory testing of blending water, and the same reason a parts-per-trillion taint such as a chlorophenol can pass every chemical limit on a certificate and still be obvious in the glass.

What should a buyer or plant engineer actually ask for?

Six questions, all answerable from documents a competent supplier already holds:

1. A wetted-materials list, end to end. Tank, pipe, pump head, valve seats, hose, gaskets, O-rings, sample valve, sight glass — the actual polymer and elastomer grades, not “food-grade throughout”.
2. Which regime each item is certified under. Food-contact (EU 10/2011, FSSAI 2018 / IS 10146, 21 CFR 177) or water-contact (NSF/ANSI 61, Regulation 31 / BS 6920, the EU positive lists under Directive 2020/2184), and ideally both.
3. Evidence of organoleptic testing, not just health-effects testing. A BS 6920 odour-and-flavour result or an EN 1622 threshold number, because NSF 61 explicitly excludes taste and odour.
4. The elastomers by name. Given the Copenhagen result, seals deserve the same scrutiny as the tank — including a rule about what grade may be fitted during unplanned maintenance.
5. Total organic carbon, measured at the fill point. TOC is the one routine parameter that responds to organic migration; conductivity does not. Sampling point matters as much as the number, as set out in how blending water is tested.
6. A flushing and first-use protocol. Migration is highest on first contact and falls with successive exposures — which is why a newly installed line or a freshly replaced seal is the moment of maximum risk, not steady-state operation.

None of this is exotic. It is the same discipline already applied to microbiology and particulates, extended to the one part of the system a certificate of analysis structurally cannot see: the certificate describes the water, and the materials act on it after the certificate was written.

Glossary

• Food-contact material (FCM) — any material or article intended to come into contact with food, regulated for what it may transfer into that food.
• Migration — the transfer of substances from a material into the food or water it contacts. Overall migration is the total mass transferred; specific migration is the amount of one named substance.
• Overall migration limit (OML) — the regulatory ceiling on total transfer, set in the EU and India at 10 mg/dm² of contact surface (or 60 mg/kg).
• Food simulant — a standardised liquid used in place of real food for migration testing. Simulant A is 10% ethanol, C is 20% ethanol, D1 is 50% ethanol, D2 is vegetable oil.
• Extractables and leachables — extractables are what a material can be forced to release under aggressive laboratory conditions; leachables are what it actually releases in normal service.
• 2,4-DTBP (2,4-di-tert-butylphenol) — a phenolic compound formed when phosphite antioxidants used to stabilise polyolefins degrade; a documented migrant from HDPE pipe.
• Threshold odour number (TON) — the dilution factor at which a water sample’s odour stops being perceptible to a panel; higher means more odorous. Measured under EN 1622.
• NSF/ANSI/CAN 61 — the North American standard for health effects of drinking-water system components. Covers chemical contaminants; excludes taste, odour and performance.
• BS 6920 — the British test suite for non-metallic products in contact with drinking water, covering odour and flavour, appearance, microbial growth, metals and cytotoxicity.

Frequently asked questions

Is a “food-grade” plastic tank good enough for blending water? It is the starting point, not the answer. Food-grade certification confirms the material passes an overall migration limit — 10 mg/dm² in the EU (Article 12), the same figure in India (FSSAI, 2018). It says nothing about whether the material imparts an odour, because overall migration is a mass measurement and odour thresholds sit orders of magnitude below it.

Do plastic pipes really change the taste of water? They can. Five of seven brands of HDPE drinking-water pipe produced water with a threshold odour number of 4 or higher, with 2,4-di-tert-butylphenol identified as a major migrant (Skjevrak et al., 2003). Whether that reaches a glass of whisky depends on contact time, surface-to-volume ratio and flushing — but the mechanism is established, not hypothetical.

Does an NSF 61 certificate mean my water will be flavour-neutral? No. The standard “does not establish performance, taste and odor, or microbial growth support requirements” (NSF/ANSI 61, §1.1). For flavour neutrality you need an organoleptic result — BS 6920 odour and flavour testing, or an EN 1622 threshold number.

Is stainless steel the safe default? For organic migration, largely yes — it contributes no polymer additives. But it is not a complete answer, because a stainless system still contains elastomer seals at every joint, and because low-conductivity water is measurably more aggressive toward metal than ordinary water is, which is the subject of what storage and pipework do to a blending water spec.

Does the same material specification work for the spirit side of the plant? Not automatically. EU migration testing moves from 10% ethanol to 50% ethanol once a product exceeds 20% ABV (Annex III), reflecting how much more a strong spirit extracts. Materials qualified on the water side should be re-qualified against the appropriate simulant before they are used downstream of the reduction tank.

Blenders International · Water Science series