Green hydrogen certification, compared: RFNBO, 45V and GHCI

RFNBO is a legal classification under the EU Renewable Energy Directive. It determines whether a fuel counts towards EU renewable fuel obligations, which is what gives it value to a European buyer who has one.

Section 45V is a production tax credit in the US Internal Revenue Code. It pays a producer for hydrogen made in the United States below a lifecycle emissions threshold. It certifies nothing about the hydrogen and it does not travel with the molecule.

The Green Hydrogen Certification Scheme of India sits under the National Green Hydrogen Mission. It defines green hydrogen by an emissions threshold and issues certificates against it, and for most Indian producers it is not optional.

The three ask different questions. RFNBO asks where the electricity came from, when, and from how far away. 45V asks what the carbon intensity is and whether the facility is American. GHCI asks whether the carbon intensity meets an Indian threshold. A project can satisfy one and fail the others on facts that have nothing to do with how well the plant is run.

Commission Delegated Regulation (EU) 2023/1184 sets out when electricity used to make hydrogen counts as fully renewable, and which test applies depends on the route. A direct line to the generator is Article 3. Grid-taken electricity has four routes under Article 4, and they are not equally demanding: a bidding zone above 90 per cent renewable in the previous calendar year needs none of the three conditions below, subject to a cap on hours; a bidding zone below 18 gCO2eq per MJ needs a power purchase agreement plus temporal and geographic correlation but not additionality; electricity taken during downward redispatch is treated separately; and everything else falls to the residual route where all three apply. A project siting in a Nordic zone is answering a different question from one siting in India, and holding renewable certificates satisfies none of the three on any route.

Additionality. The renewable installation has to have come into operation no more than 36 months before the hydrogen installation, and it should not have received operating or investment aid unless that support was repaid in full. Support for land, and support for a grid connection, do not disqualify. There is a transitional provision that a large share of current project timelines rests on, and it is routinely described wrongly: a hydrogen production installation coming into operation before 1 January 2028 is relieved of the 36-month test and the aid test until 1 January 2038. That is a fixed date, not a rolling ten years from each plant's own commissioning, so a plant starting in 2026 gets about eleven and a half years and one starting in 2024 got more. It does not extend to capacity added after 1 January 2028, which matters for a phased build. And it relieves only those two tests: the producer still has to own equivalent renewable generation or hold power purchase agreements covering at least the volume claimed, and temporal and geographic correlation are untouched by it.

Temporal correlation. Until 31 December 2029, hydrogen has to be produced in the same calendar month as the renewable generation matched to it. From 1 January 2030 the matching period becomes one hour. A Member State may bring the hourly rule forward to 1 July 2027 by notifying the Commission, so the applicable date is not the same across the EU. There is also a price-linked route: production during periods when the market clearing price in the relevant bidding zone is at or below EUR 20 per MWh, or below 36 per cent of the emission allowance price, is treated as satisfying the correlation.

Hourly matching changes what a plant's output is. If eligible renewable electricity exists for ten hours of a twenty-four hour run, the hydrogen attributable to those ten hours is RFNBO and the remainder is not. An annual renewable percentage cannot produce that split and a monthly one will not either, once the rule turns over.

Geographic correlation. Generation and consumption have to sit in the same bidding zone, or in zones connected in a way the regulation recognises. This is the condition Indian export projects find hardest, because the Indian grid does not map onto EU bidding-zone logic and long-distance deliverability is difficult to evidence. Co-location or a direct line removes most of the argument before it starts.

Commission Delegated Regulation (EU) 2023/1185 sets the calculation. The requirement is a 70 per cent greenhouse gas saving against a fossil comparator of 94 gCO2e per MJ, which for hydrogen works out at roughly 3.4 kg CO2e per kg.

The boundary is wide. Electricity generation, conversion, compression, transport and storage are all in it, and so are the process inputs a plant tends to forget: water treatment, purification and drying. The figure is defensible only if the emission factors are sourced, the boundary is written down, and the calculation is version controlled so that the same inputs reproduce the same result in front of an auditor a year later.

Passing one test and failing the other is a fail. A project with an excellent carbon intensity and a renewable asset commissioned in 2015 is not RFNBO, and no amount of emissions performance repairs the sourcing side.

Commission Delegated Regulation (EU) 2025/2359, adopted on 8 July 2025 and in force since 11 December 2025, sets a separate methodology for low-carbon hydrogen. It applies a 70 per cent saving test to routes RFNBO does not cover: steam methane reforming with carbon capture, and hydrogen made from electricity that does not satisfy the RFNBO sourcing rules.

It is a different category under a different name. A buyer holding an RFNBO obligation cannot discharge it with low-carbon hydrogen, and describing a project as low-carbon compliant to a counterparty who asked for RFNBO is the kind of imprecision that surfaces at contract stage. One thing it does not create is a nuclear power purchase agreement pathway. Electricity that is not fully renewable is scored by one of four grid-based methods in the annex, and the Commission is required only to assess, by 1 July 2028, whether a dedicated route for nuclear electricity should exist.

Section 45V pays a credit on hydrogen produced at a qualified US facility with a lifecycle emissions rate of not more than 4 kg CO2e per kg. The credit is tiered on that rate: an applicable percentage of 20 per cent between 2.5 and 4 kg, 25 per cent between 1.5 and 2.5, 33.4 per cent between 0.45 and 1.5, and 100 per cent below 0.45.

Those percentages apply to a base of USD 0.60 per kg, inflation adjusted from a 2022 baseline, and the result is multiplied by five where the prevailing wage and apprenticeship requirements are met. That multiplication is where the headline figure of up to USD 3.00 per kg comes from, and a project that does not meet those labour conditions is working from one fifth of it.

Two things matter more than the tiers. The emissions rate is computed with the 45VH2-GREET model, or through the provisional emissions rate process where a pathway is not represented, and the model version is part of the evidence rather than a footnote. And the 2025 amendments moved the deadline: a facility now has to begin construction before 1 January 2028, against a previous sunset five years later.

For an Indian producer the position is simpler than it looks. The credit is tied to production in the United States or a US territory, so Indian hydrogen is not eligible and should not be described as though it were. A US buyer may still want a carbon intensity figure, a lifecycle assessment, evidence of the electricity attribute and independent verification. That is buyer-driven documentation under a contract, and presenting it as a 45V claim is a misstatement rather than a shortcut.

India defines green hydrogen at not more than 2 kg CO2e per kg, taken as a twelve-month average, on a well-to-gate boundary that includes water treatment, electrolysis, purification, drying, compression and onsite storage. MNRE is the nodal ministry, a designated Implementing Agency runs certification, and a Technical Committee chaired by the National Green Hydrogen Mission director approves issuance. The Bureau of Energy Efficiency accredits the validation and verification agencies, which is a narrower role than running the scheme.

The scheme runs through four certificate levels: concept, facility, provisional and final. The final certificate gates government incentives and domestic sale, and it is mandatory in four cases: a producer receiving any central or state incentive or subsidy for green hydrogen production, expressly including SIGHT awardees; a producer intending to sell or use the hydrogen in India; a producer receiving any exemption or concession; and a producer with an Indian consumer for part of its output and export for the balance. The export relief is narrower than it is usually described. It applies to a producer with one hundred per cent export capacity that takes no central or state incentive or concession at all, and that producer still has to report quantity and emissions against the importing country's standard. A project holding a SIGHT award and exporting everything is not relieved. Facilities producing ten tonnes a year or less are outside the scheme.

What GHCI does not do is create RFNBO eligibility. The thresholds differ, the averaging period differs, and the electricity sourcing tests are simply absent from the Indian test. An exporter needs both engines running against the same operating data, and the EU one is the stricter of the two by a distance.

An I-REC, a European guarantee of origin and an Indian REC all do the same job. They record that a megawatt hour of renewable electricity was generated and that its attribute was transferred to somebody. That supports a market-based Scope 2 claim and it is legitimate evidence for what it covers.

It is not a hydrogen claim. A certificate does not show that the generating asset is recent enough for additionality, that generation and consumption fell in the same matching period, or that the two are geographically correlated. RFNBO asks all three and a certificate answers none of them. Physical delivery is a separate matter: outside the direct-line route the grid routes are contractual, so what RFNBO wants is a power purchase agreement with correlation evidence rather than proof that your electrons arrived. Indian RECs sit further out again, since they are instruments of domestic power-sector regulation and are not designed for international acceptance.

The rule that catches projects is double counting. The same megawatt hour cannot be sold as a certificate to one party and claimed as green electricity by another. Whatever system holds the records has to lock each attribute to one claim, one batch and one reporting period, and refuse to release it a second time.

Most of what an auditor asks for can be assembled late, badly and at cost. Three things cannot be assembled at all.

Interval metering. If generation and consumption were recorded as monthly totals, no later work produces the hourly series an hourly claim needs. Interval data has to be recorded from commissioning, at the generator and at the electrolyser, and retained for longer than the reporting period.

Attribute exclusivity. If the renewable attributes were sold elsewhere, or the power purchase agreement is silent about who owns them, the position at the time of production is what an auditor examines. A later assignment does not repair a period that has already closed.

Clock alignment. Electricity metering, production metering and any grid or certificate record have to sit on a synchronised clock. Two systems drifting apart produce a reconciliation that fails and cannot be re-run, because the period is gone. This is the same failure set out in chain of custody audit evidence, and it tends to be discovered during an audit rather than before one.

Decide the target regime before the metering design rather than after it. Where the EU is the market, co-locate or take a direct line, because it removes the geographic argument entirely rather than arguing it. Put attribute exclusivity in the power purchase agreement. Record commissioning dates with evidence, since additionality is a documented fact and not an intention. Fix the system boundary and the emission factor sources in writing and version the calculation. Engage a scheme and an auditor before production, because a pre-assessment catches structural failures while they are still cheap to fix.

And keep the interval record from the first day, whichever regime you are aiming at. Every one of these frameworks is moving towards finer time resolution rather than away from it, and a plant that has the interval data can answer a rule that tightens. A plant that has monthly totals cannot.

For a lender or an offtaker this is the same discipline as post-COD reporting. The evidence has to exist while the plant runs, produced by the plant's own systems, rather than assembled after the question is asked by the party the question is about.

Every figure above carries its instrument, because these change. This page was checked on 28 August 2026 and should be read against the notified text rather than in place of it.

EU electricity sourcing rules: Commission Delegated Regulation (EU) 2023/1184, consolidated, which is the text as amended in June 2024. EU greenhouse gas methodology: Commission Delegated Regulation (EU) 2023/1185. EU low-carbon hydrogen: Commission Delegated Regulation (EU) 2025/2359. US credit: 26 U.S. Code Section 45V and the Department of Energy 45V resources covering the 45VH2-GREET model. India: the Green Hydrogen Certification Scheme of India, notified April 2025 under the National Green Hydrogen Mission. Read the notified scheme rather than the September 2024 draft; they differ on who runs certification.

Nothing here is legal, tax or certification advice. The applicable scheme document, the certification body's own guidance for the period being claimed, and your advisers govern any actual claim.