QR codes, GS1, DIDs: choosing a unique identifier for your battery passport

From 18 February 2027, every EV battery, LMT battery and industrial battery above 2 kWh placed on the EU market must carry a QR code giving access to its digital passport. Regulation 2023/1542 spends more than 90 data attributes on what that passport must contain, and only a few sentences on the identifier that holds it all together. Yet the identifier is the spine of the system: every state-of-health update, ownership transfer and recycling record over a 15 to 20 year battery life hangs off the string you choose before the first pack ships. It is also the one design decision you cannot patch later, because it gets engraved into metal.

What does the regulation actually require?

Strip away the commentary and the legal core is three requirements sharing one deadline:

• A QR code on the battery. From 18 February 2027, Article 13(6) requires batteries to be marked with a QR code as specified in Annex VI, Part C, printed or engraved "visibly, legibly and indelibly" on the battery itself. Only where the nature and size of the battery make that impossible may the code move to the packaging and the accompanying documents.
• A unique identifier behind it. Under Article 77, the passport is accessed through that QR code, which links to a unique identifier attributed to the battery by the economic operator placing it on the market. For the passport-scope categories this is a per-unit record: two packs of the same model carry two different identifiers.
• Compliance with ISO/IEC 15459. The QR code and the unique identifier must comply with the ISO/IEC 15459 series (parts 1 to 6) or an equivalent. This family does not prescribe a format. It defines how registered issuing agencies hand out namespaces, so that identifiers minted by different companies can never collide.

Everything else is an implementation choice: which issuing agency, whether the identifier is a bare string or a web URI, what domain it resolves on, how redirection works when the pack changes hands. The regulation is deliberately scheme-neutral, and that neutrality is both a freedom and a trap.

Is the QR code the identifier?

No, and keeping the two apart drives every good decision that follows. The QR code is a data carrier, standardised in ISO/IEC 18004: a convention for printing bits so a camera can read them. The identifier is the payload inside the carrier. A carrier can be reprinted, resized or supplemented with an NFC tag or a data matrix on inner modules. The identifier can never change, because passport history, warranty claims and recycling documentation all reference it.

The practical consequence: the engraved code should contain a stable pointer, never the data. State of health changes weekly, recycled-content shares change per audit, and the regulation's three access tiers (public, legitimate interest, authorities) must all enter through the same physical mark and see different views after authentication. All of that lives behind the pointer, on infrastructure you can upgrade. What is burned into the pack must be the one thing that never needs upgrading.

Which identifier schemes are on the table?

The wider Digital Product Passport framework under the ESPR is converging on the same question. The CEN and CENELEC joint committee JTC 24 has published horizontal DPP standards, and its identifier standard (EN 18219) recognises several product identifier schemes side by side, including web URIs of the GS1 Digital Link type, IEC 61406 identification links and W3C Decentralized Identifiers. For a battery team, three candidates dominate the shortlist:

• GS1 Digital Link. A GTIN identifies the maker and model, a serial number identifies the unit, and both are expressed as an ordinary HTTPS address, for example https://id.maker.eu/01/04012345678901/21/000482917. GS1 is a registered ISO/IEC 15459 issuing agency, so the compliance question answers itself, logistics scanners already parse GTINs, and the GS1 resolver standard lets the same link redirect to different services over the battery's life.
• W3C Decentralized Identifiers (DIDs). A W3C Recommendation since 2022. A DID has no central registrar: control is proven cryptographically, and it pairs naturally with verifiable credentials for claims like recycled-content certificates. That makes DIDs attractive in data ecosystems such as Catena-X. The catch is that a bare did: string is not something a phone camera can open, so most designs wrap it in an HTTPS link anyway, and resolution infrastructure is still young.
• Plain UUIDs or internal serials. Free to mint and instantly available. But global uniqueness then rests on your internal discipline alone, the "equivalent to ISO/IEC 15459" argument is yours to make in front of a market surveillance authority, and the number means nothing to any system outside your own.

These options are not mutually exclusive. A common architecture puts a GS1 Digital Link on the label for universal scannability and anchors DID-based credentials underneath it for cross-company verification. What matters is deciding the layering before the label artwork is frozen, not after.

How do you mark a code that must outlive the pack?

"Indelibly" is a strong word for an object that will see 15 to 20 years of vibration, thermal cycling, road salt and workshop handling, followed by a second life in a storage container. Printed polymer labels fade and peel; laser engraving and anodised metal plates survive, which is why the regulation's printed-or-engraved wording matters at design time, not at the print shop.

The carrier needs engineering margin too. ISO/IEC 18004 error correction at its highest level keeps a code readable with up to about 30% of the symbol damaged, and a generous module size with a clean quiet zone is what lets a recycler's intake station scan a dusty pack from a distance, with gloves on. Placement is a lifecycle decision: a code on a cosmetic cover disappears at the first repair, so durable marking belongs on the structural pack, and module-level marks help whoever dismantles it. And the pointer must outlive your web presence: the passport has to stay accessible for the battery's whole service life, long after the domain in your current label artwork has survived two rebrands, or has not.

Which traps should you catch before printing a million labels?

The expensive mistakes are all decided before the first label is printed. The recurring ones:

• A marketing domain in the URI. www.brand.com/batteries/... breaks at the first website redesign. Use a dedicated identifier domain with a resolver you control, and treat it as 20-year infrastructure.
• Data in the code instead of a pointer. Encoding chemistry or test results into the QR freezes them at print time and bloats the symbol. Engrave the identifier; serve the data.
• Serial collisions across plants and suppliers. Two factories, a contract manufacturer and an acquisition later, "unique within our ERP" is no longer unique. Issuing-agency namespaces exist precisely to make collisions structurally impossible.
• Ignoring the ESPR trajectory. Batteries are the first regulated product passport, not the last. Aligning with the JTC 24 horizontal standards now means one identifier infrastructure can serve the product categories that follow.
• One code, one audience. The same engraving must serve a consumer, a second-life buyer with legitimate-interest access and a market surveillance authority. If your scheme cannot resolve one identifier into tiered views, you will end up printing two codes, and regretting both.

Conclusion: engrave a pointer, not a decision

The identifier will outlive the ERP that issued it, the domain it first resolved on and possibly the business unit that chose it. The regulation asks for little: a QR code, a unique identifier, ISO/IEC 15459 discipline. Everything else is architecture, and the sound architecture is boring: an issuing-agency-backed identifier, a redirectable resolver, and nothing in the engraving that can ever go stale.

Passoria's regulation-ready passport engine puts a compliant per-unit identifier and the full Annex XIII record behind every QR code, with Catena-X and Gaia-X interoperability built in. If 2027 labels are on your roadmap, our pilot program is open to a limited number of manufacturers, resellers and recyclers.