Hydrogen Purity Standards: What 99.999% H2 Really Specifies

By Raymond Xie, Director, Hovogen R&D Division

Bottom line first: "99.999%" tells you only one thing — that the total non-hydrogen content is at or below 10 ppm. It says nothing about which 10 ppm. A cylinder at 5N carrying 10 ppm of water and a cylinder at 5N carrying 10 ppm of hydrocarbons will damage your GC baseline in completely different ways, and both are legitimately sold as "99.999%". A real hydrogen purity specification is not a single number; it is a grade plus an impurity table plus a test method plus a measurement location. This guide sets out the standards that actually govern the claim — the N-grade system used by the gas industry, the impurity limits in ISO 14687-2, the ASTM analytical methods that produce the numbers, and the certificate language that separates a verifiable spec from a datasheet line.

1. What the N-Number System Does — and Does Not — Tell You

Industrial and laboratory gas suppliers describe purity with an "N" count, where the number of nines after the decimal is the number of nines in the purity figure, and the digits after the N indicate how far short of the next nine the value falls.

• 4N: 99.99% — ≤ 100 ppm — Industrial combustion, rough carrier gas

• 5N: 99.999% — ≤ 10 ppm — GC-FID fuel gas, GC carrier gas, ICP-MS reaction-cell gas

• 6N: 99.9999% — ≤ 1 ppm — High-sensitivity mass spectrometry, trace analysis

• 7N: 99.99999% — ≤ 0.1 ppm — Semiconductor processes, ultra-high-sensitivity MS

The critical limitation is structural, not incidental: an N-grade is a sum, not a composition. Total impurity ≤ 1 ppm at 6N leaves the buyer choosing how those 1 ppm are distributed among water, oxygen, carbon monoxide, carbon dioxide, total hydrocarbons, nitrogen and argon — and the analytical consequences of each differ by orders of magnitude.

A published vendor grade table makes the point concretely. A 6.0 hydrogen grade is typically specified as H₂O < 0.5 ppm, O₂ < 0.2 ppm, N₂ < 0.2 ppm, total hydrocarbons < 0.1 ppm, CO + CO₂ < 0.1 ppm, total impurities < 1 ppm [1]. Note that "total < 1 ppm" is the ceiling of the individual rows, not a replacement for them. A supplier quoting only the total has quoted nothing actionable.

2. ISO 14687-2: The Reference Point for Fuel-Cell-Grade Hydrogen

For hydrogen entering a PEM fuel cell — vehicle or stationary — the governing document is ISO 14687-2, which classifies hydrogen by type and grade and then caps each individual contaminant [2][3].

ISO 14687-2:2019 — key limits (Type I / Type II)

• Hydrogen fuel index (min. mole fraction): 99.97% — Calculated by subtraction

• Water (H₂O): 5 ppm — ASTM D7653 / D7649

• Total hydrocarbons (C₁ basis): 2 ppm — ASTM D7675

• Oxygen (O₂): 5 ppm — ASTM D7649

• Helium: 300 ppm — ASTM D1945

• Nitrogen + argon: 100 ppm — ASTM D7649

• Carbon dioxide (CO₂): 2 ppm — ASTM D7649 / D7653

• Carbon monoxide (CO): 0.2 ppm — ASTM D7653

• Formaldehyde (HCHO): 0.01 ppm — ASTM D7653

• Formic acid (HCOOH): 0.2 ppm — ASTM D7550 / D7653

• Ammonia (NH₃): 0.1 ppm — ASTM D7653

• Total sulphur compounds (H₂S basis): 0.004 ppm — ASTM D7652

• Total halogenated compounds: 0.05 ppm — ASTM D6366 (work item 23815)

• Particulate concentration: 1 mg/kg — ASTM D7650 / D7651

Two observations matter for anyone specifying equipment.

First, ISO 14687-2 is far stricter than a GC carrier-gas spec — and that is correct. It targets the platinum catalyst in a fuel cell, where CO at 0.2 ppm and sulphur at 0.004 ppm are catalyst poisons, not nuisance impurities. Published risk analyses of ISO 14687-2 contaminants trace their origin to the production route: SMR with PSA purification, chlor-alkali, and PEM electrolysis with TSA each generate a different contaminant profile, which is why the standard specifies individual species rather than a single assay figure [2].

Second, achieving these limits analytically is non-trivial. Total sulphur at 0.004 ppm sits barely above the detection limit of modern sulphur-chemiluminescence methods, and halogenated compounds at 0.05 ppm required a dedicated ASTM work item to characterise [3]. Vendors therefore need multi-train analytical infrastructure — GC, GC-MS, IC, FTIR, SCD and methanizer configurations working from one data system — to issue a compliant certificate [4]. That infrastructure is a reasonable proxy for whether a supplier's purity claim is real.

For laboratory hydrogen — GC carrier gas, FID fuel, ICP-MS collision/reaction-cell gas — ISO 14687-2 is not the applicable document. It is nonetheless the right reference to read, because it defines what "trace" means and it separates fuel-cell-grade hydrogen (5N and above, impurity-controlled) from industrial-grade hydrogen (3N–4N, sold by assay only). If a supplier quotes ISO 14687-2 for a bench generator, ask whether they are quoting the standard or quoting their certificate.

3. Which Impurity Actually Hurts — and What It Does to Your Instrument

The commercial question behind "what does 99.999% specify" is: which impurity will cost me a weekend of troubleshooting? The answers differ sharply by application.

Water. The universal offender. In a GC capillary, excess moisture damages the stationary phase, raises bleed, and shifts retention times; in a FID it destabilises the flame. Hovogen's GC-FID selection guidance specifies moisture < 1 ppm (dew point below −60 °C) for laboratory hydrogen [5]. Note the physical constraint this implies: measuring sub-ppm water requires either a heated sampling line or a cold-trap mirror hygrometer — a room-temperature sample line will not report what the instrument is actually delivering.

Total hydrocarbons. The FID-specific problem. Residual hydrocarbons in hydrogen elevate baseline noise, raise detection limits, and produce false positives in trace analysis, which is why GC-grade hydrogen is specified at 6N (total < 1 ppm, THC < 0.1 ppm) rather than 5N for sensitive work [1][6]. Methane is the usual culprit, because it is the hardest hydrocarbon to remove downstream of a reformer — and the easiest to remove upstream of one.

Oxygen. Two distinct risks. Chemically, O₂ in hydrogen is a combustion hazard in any confined volume. Analytically, oxygen can cause premature stationary-phase degradation and, in methanizer-equipped FIDs, degrade the methanizer itself. Practical laboratory specs put O₂ at < 1 ppm [5].

CO and CO₂. Low relevance for GC-FID, decisive for fuel cells and for any hydrogen used in a catalyst or synthesis train. CO₂ at 2 ppm and CO at 0.2 ppm are the ISO 14687-2 figures [2].

Nitrogen and argon. Chemically inert, but they accumulate in the carrier stream and dilute the analyte, shifting linear range and detection limits. For trace work, N₂ < 5 ppm is the practical target [5].

Particulates. Routinely ignored and occasionally decisive — a 10 µm particle downstream of a generator can lodge in a capillary restriction and produce erratic flow. ISO 14687-2 caps particulates at 1 mg/kg with a maximum particle size of 10 µm [3].

The general rule: specify by application, verify by certificate, re-verify on your own gas path. For the full breakdown of how a PEM generator's purification chain produces these numbers — deoxo, PSA, deep drying — see our guide to lab hydrogen generator purity grades and selection; for the flow and pressure side of the same decision, see the complete 2026 GC-FID selection guide.

4. What a Verifiable Certificate Actually Says

Most datasheets stop at the N-number. A certificate of analysis — ideally from an accredited third-party laboratory, not the manufacturer's own spec sheet — should carry six things:

1. Grade claimed, in both notations (e.g. "5.0 / 99.999%") plus the underlying assay method.

2. Individual impurity table with numeric limits, not "total" alone.

3. Test method per impurity (ASTM D7649, D7653, D7675, D7652, D7550…), because a limit without a method is not verifiable.

4. Measurement location — production outlet, after final purification, or at the delivery point. A certificate measured at the plant is a statement about a different gas than the one at your instrument.

5. Date, sample conditions, and lab accreditation (ISO/IEC 17025 for the testing laboratory).

6. Traceability statement referencing the certified reference materials used for calibration.

Two failure patterns to watch for. First, a datasheet listing "99.999%" with no moisture figure at all — a unit at 5N total purity with 10 ppm moisture will cause baseline drift and column degradation [6], exactly the failure the headline number was supposed to rule out. Second, a certificate valid only at commissioning. Ask whether purity is maintained continuously across the unit's operating life, and whether online monitoring (dew-point display, purity output) exists so that a drifting unit is visible before it corrupts a data set.

5. Specifying for On-Site Electrolytic Hydrogen

If the hydrogen is made on site by PEM electrolysis rather than delivered in cylinders, the impurity profile is set by the plant design, and the specification conversation moves upstream:

• Feed water sets the floor. Deionised water at ≥ 1 MΩ·cm (≤ 1 µS/cm) is the input that protects membrane life and holds the purity certificate. Conductivity above ~5 µS/cm accelerates membrane fouling, raising impurity levels and maintenance frequency [6].

• Oxygen is inherent, not accidental. PEM electrolysis vents oxygen at the anode. It is kept out of the hydrogen circuit by the membrane itself, but any seal or membrane degradation shows up immediately as an O₂ excursion — which is why O₂ belongs on the in-line measurement list, not just the annual certificate.

• Pressure and flow stability are part of the purity story. Unstable delivery pressure produces flow instability, which corrupts retention-time reproducibility independently of gas composition. A purity certificate does not certify pressure stability; that is a separate specification.

• Scale changes the answer. Where a bench generator serves one or two instruments, 5N–6N is sufficient and proportionate. Where the same quality system has to cover a larger laboratory or extend toward a process duty, the same membrane architecture scales up — Hovogen's PEM electrolyzer platform is the relevant reference for that transition, and the scientific hydrogen generator series covers the laboratory end.

6. Five Specification Mistakes

1. Buying on the headline number. 5N total says nothing about composition. Demand the impurity table.

2. Wrong standard for the duty. ISO 14687-2 is a fuel-cell standard; quoting it for a GC carrier gas either means over-specification or a misread certificate. Know which document governs your application.

3. Limits without methods. "CO < 0.2 ppm" is only meaningful next to the test method that produced it and the calibration standard behind it.

4. Certificate without location. Plant-outlet purity is not delivered-purity. Ask where the gas was measured.

5. One-time verification. On-site generators drift as consumables age and DI resin degrades. Continuous monitoring plus periodic third-party analysis is the only version of the claim that stays true.

FAQ

What does 99.999% hydrogen purity mean in ppm? It means total non-hydrogen content ≤ 10 ppm. It does not specify which impurities make up that 10 ppm, which is why a full impurity table is required.

What is the difference between 5N, 6N and 7N hydrogen? 5N = 99.999% (≤ 10 ppm total), 6N = 99.9999% (≤ 1 ppm), 7N = 99.99999% (≤ 0.1 ppm). Analytical labs typically run at 5N; 6N is for high-sensitivity MS; 7N is semiconductor-grade territory [1][6].

Is ISO 14687 the same as ISO 14687-2? ISO 14687-1 covers gaseous hydrogen generally; ISO 14687-2 covers hydrogen quality for fuel-cell applications, and is the source of the CO 0.2 ppm / total sulphur 0.004 ppm limits [2].

What purity does GC carrier gas need? GC-grade hydrogen is specified at 99.9999% (6N) with hydrocarbons < 0.1 ppm, moisture < 1 ppm, O₂ < 1 ppm and N₂ < 5 ppm — verified by certificate of analysis rather than datasheet [1][6].

Does PEM electrolysis produce 99.999% hydrogen? PEM electrolysis is the cleanest route — no liquid electrolyte carryover, no hydrocarbons from a reformer — and its natural output profile is naturally well suited to 5N laboratory duty once deoxo/PSA and deep drying are fitted. The purity claim still needs a certificate against a stated method, not a nameplate.

Does a certificate of analysis guarantee continuous purity? No. It certifies a sample at a location and a moment. For on-site generators, ask for continuous monitoring (dew point, purity output) plus a periodic third-party schedule.

Conclusion

"99.999%" is the beginning of a hydrogen specification, not the end of one. The number sets a budget of 10 ppm and tells you nothing about its composition; ISO 14687-2 shows what full composition control actually costs, down to 0.004 ppm total sulphur; and the ASTM method list shows which of those limits a supplier can genuinely defend. Specify by application, require the impurity table with methods and measurement location, and treat "5N" as the entry ticket rather than the answer. For the flow, pressure and sizing decisions that sit alongside purity in any laboratory hydrogen specification, start with the 2026 GC-FID selection guide, and for the detection-side context see how a flame ionization detector uses hydrogen.

Next step: Contact the Hovogen scientific team for a site-specific purity specification and third-party certificate of analysis for your instrument set.

References

[1] Messer, "Gas Purity and How It Is Expressed" — N-grade notation, 6.0 hydrogen impurity table (H₂O < 0.5 ppm, O₂ < 0.2 ppm, N₂ < 0.2 ppm, THC < 0.1 ppm, CO+CO₂ < 0.1 ppm, total < 1 ppm). http://www.messerco2.cn/m/xxyxz/103.html

[2] "Probability of occurrence of ISO 14687-2 contaminants in hydrogen: Principles and examples from steam methane reforming and electrolysis (water and chlor-alkali) production processes" (Int. J. Hydrogen Energy) — ISO 14687-2 Table 1 contaminant thresholds and production-route risk analysis (SMR+PSA, chlor-alkali, PEM+TSA). https://www.sciencedirect.com/science/article/abs/pii/S0360319918308450

[3] Instrumart, "Measurement of Trace Impurities in UHP Hydrogen for Fuel Cells" — ISO 14687-2 limits mapped to ASTM D7649/D7653/D7675/D7652/D7550/D7650/D7651 methods and detection limits. https://www.instrumart.com/assets/54027E-V3_PST_Hydrogen.pdf

[4] Shimadzu, "Hydrogen Energy Analysis Solutions for Quality Control of Hydrogen" — ISO 14687-2019 compliant analyser configuration and multi-system data integration. https://www.shimadzu.co.uk/industries/new-energy/fuel-cells/hydrogen-energy/index.html

[5] Hovogen, "Hydrogen Generator for GC-FID: The Complete 2026 Selection Guide" — laboratory impurity limits (moisture < 1 ppm, hydrocarbons < 0.1 ppm, O₂ < 1 ppm, N₂ < 5 ppm), Type I DI water ≥ 1 MΩ·cm, flow sizing. https://www.hovogen.com/post/hydrogen-generator-for-gc-fid-the-complete-2026-selection-guide

[6] Hovogen, "Lab Hydrogen Generator: The Complete Guide to PSA Purification, 5N/7N Purity Grades & Selection" — grade-to-application mapping, purification chain, certificate requirements. https://www.hovogen.com/post/lab-hydrogen-generator-guide-purity-5n-7n-en

About the author

Raymond Xie is Director of the R&D Division at Hovogen, where he leads development of PEM electrolysis and hydrogen generation systems for analytical laboratories. His work spans stack degradation, gas purity, and lifetime economics — the three parameters that decide whether a laboratory hydrogen supply stays inside specification over a five-year service life.