top of page

Oxygen Removal After PEM Electrolysis: Deox Stages

4 hours ago
9 min read

By Raymond Xie, Director, Hovogen R&D Division

Bottom line first: a PEM electrolyzer does not make pure hydrogen — it makes wet hydrogen with a small but persistent oxygen carryover. That oxygen is chemically harmless to the stack yet fatal to some downstream uses: a hydrogen-cooled ECD, a fuel-cell test bench, or a materials furnace can all be compromised by single-digit ppm of O₂. The fix is a dedicated deox stage — a catalytic recombiner that converts trace O₂ back into water, followed by drying to remove the water it just made. This guide walks through why oxygen appears in PEM output, how much is too much for each application, how a deox stage actually works, where it sits in the purification train, and how to verify the result with real measurements.

1. Why PEM Electrolysis Output Carries Oxygen

In a PEM electrolyzer, hydrogen evolves on the cathode and oxygen on the anode, separated by a proton-exchange membrane. The membrane is an excellent but imperfect gas barrier. Three mechanisms push oxygen into the hydrogen stream:

  • Anodic crossover. Dissolved O₂ permeates the hydrated membrane from the anode side to the cathode side, where it is swept into the evolving hydrogen flow. Permeability rises with temperature and membrane thinning — the same degradation that drives stack replacement economics also slowly raises crossover oxygen.

  • Entrainment at the separator. The cathode exit is a two-phase mix of hydrogen and misted water. Droplets and bubbles can carry anode-side gas if the gas–liquid separator or demister underperforms.

  • Startup and transient purge. During shutdown, purge, and pressure-equalization events, anode and cathode manifolds briefly share gas pathways; each transient leaves a small oxygen footprint.

For a healthy industrial PEM stack in normal operation, the oxygen carryover at the cathode outlet is typically in the tens to a few hundred ppm range — small in absolute terms, but far above what fuel-cell-grade or analytical applications tolerate. This is a systematic feature of the technology, not a defect. Even the best-designed electrolyzer ships PEM electrolysis output that needs a final purification stage before it is "5.0-grade" hydrogen.

2. How Much Oxygen Is Too Much

The tolerable oxygen ceiling depends entirely on the load:

  • PEM fuel cells (vehicle-grade hydrogen). ISO 14687:2019 and SAE J2719 cap oxygen at 5 ppm for fuel-cell-grade hydrogen. Fuel-cell catalysts tolerate moderate O₂, but the limit protects against excursions that oxidize separator materials and upset stack control assumptions.

  • GC carrier gas and make-up gas. Laboratory hydrogen generators for gas chromatography specify oxygen below roughly 1 ppm for 5.0-grade output, because oxygen bleeds columns, shifts baselines, and shortens detector life. Oxygen is especially hostile to electron-capture detectors, where trace O₂ quenches the standing current and destroys sensitivity.

  • Hydrogen furnaces and sintering lines. Reducing-atmosphere processes usually need O₂ in the low-ppm range; excess oxygen oxidizes workpieces instead of reducing oxides.

  • Hydrogen blending and energy uses. Combustion applications are tolerant, but even here, upstream users of the same supply chain may impose the tighter limit.

One nuance matters for buyers: the oxygen number on a gas quality certificate is measured after purification, at the point of use — not at the electrolyzer outlet. A specification that quotes "O₂ < 1 ppm" is implicitly promising a purification train that includes oxygen removal, not merely a better electrolyzer. This is the same principle that governs hydrogen purity standards: a purity grade is a system property, not a property of the electrolysis cell alone.

3. The Deox Stage: Catalytic Recombination

The workhorse of trace oxygen removal is the catalytic deox reactor. Hydrogen itself is the reductant: over a platinum- or palladium-based catalyst, oxygen reacts with hydrogen to form water:

2 H₂ + O₂ → 2 H₂O (exothermic, ΔH ≈ −484 kJ per mol O₂)

The reaction is fast, complete, and self-driving as long as the feed is hydrogen-rich — which it always is, because the oxygen carryover is only ppm-to-percent scale. Key engineering parameters:

  • Catalyst. Pt/Pd on alumina or a proprietary support, operated at moderately elevated temperature (typically 80–200 °C) to keep the reaction complete even at ppm inlet concentrations.

  • Heat release. The exotherm is the main design constraint. On a hydrogen-rich feed, each 1% of oxygen in the inlet gas raises the catalyst bed temperature by roughly 150–170 °C. At ppm-level carryover this is negligible; it only becomes a design issue if a separator fault sends bulk oxygen into the deox inlet — which is why inlet oxygen limits and interlocks exist (Section 5).

  • Exit performance. With a healthy bed, catalytic deox routinely brings oxygen below 0.1 ppm — comfortably inside both the 1 ppm analytical target and the 5 ppm fuel-cell target.

  • Life and poisoning. Liquid slugs and silicone/oil vapors are the classic killers. A demister and a coarse dryer upstream protect the catalyst; a well-protected bed commonly runs for years before replacement.

The water formed in the reaction joins the moisture already present, which leads directly to the placement question: a deox stage always creates a downstream drying requirement.

4. Placement in the Purification Train

A practical on-site system chains the stages in a fixed order. Each stage protects the next:

  • Gas–liquid separation. Bulk liquid water and electrolyte mist drop out at the electrolyzer outlet. This is a mechanical stage — demister or separator drum — and it shields everything downstream.

  • Coarse drying. A condenser or first desiccant bed removes most of the saturated moisture. This is not the polish stage; it exists to keep liquid water out of the catalyst bed and to cut the load on the final dryer.

  • Deox reactor. The catalytic stage converts trace O₂ to water. It must be located before final drying, because the reaction generates new moisture.

  • Final drying. An adsorption dryer (typically twin-bed, automatically regenerated) brings the dew point down to specification — for analytical grades, well below −60 °C.

  • Polishing and analysis. Particulate filtration, then a trace oxygen analyzer and dew-point monitor at the point of use close the loop (Section 6).

Two placement mistakes show up repeatedly in the field. The first is putting the deox reactor after the final dryer: the reaction water then re-contaminates a gas that was already at specification. The second is omitting the coarse dryer to save cost: liquid carryover attacks the catalyst, and the "cheap" system pays for repeated bed replacement. Installation and layout constraints — ventilation, relief paths, code compliance — follow the same logic as any lab hydrogen system and are covered in the installation and safety codes guide.

Twin-tower adsorption dryer installation with filtration housings and control panel

Twin-tower adsorption dryers with upstream filtration — the final drying stage that follows the deox reactor. Image: Didier Homna / Wikimedia Commons, CC BY-SA 4.0.

5. Operating Limits and Safety

A deox reactor is safe when its inputs are controlled, and the control logic is simple:

  • Hydrogen-rich operation. The feed must stay far on the hydrogen side. At ppm-level oxygen carryover the mixture is nowhere near any flammable regime — the reactor runs, in effect, as a very lean trace-oxidation bed inside a hydrogen atmosphere.

  • Inlet oxygen cap. Designs specify a maximum inlet O₂ (commonly around 1–2%) beyond which the exotherm could overheat the bed. Interlocks typically watch deox inlet oxygen or the upstream separator level and shut the electrolyzer down on excursion.

  • Temperature monitoring. One or more thermocouples on the bed catch runaway behavior early. A trip on bed temperature is the standard backstop.

  • Pressure relief. The reactor is a pressure vessel; it carries its own relief device sized for the upstream supply — the same philosophy as buffer-tank protection in storage and refueling sizing.

None of these constraints is exotic. They are the ordinary discipline of catalytic gas processing, applied to a stream whose oxygen content is normally measured in ppm. The safety case is thus dominated not by the deox reactor itself but by the upstream system — separator health, purge procedures, and ventilation.

6. Verifying the Result: Oxygen Analysis

Purification claims must end in a measurement. The standard tools:

  • Electrochemical trace oxygen analyzers. The workhorse for ppm-level O₂ in hydrogen; inexpensive, sensitive to sub-ppm levels, and easy to integrate into a generator's control loop. The sensor cell is consumable and needs periodic replacement.

  • Zirconium-oxide analyzers. Fast response, robust at elevated temperature; common in continuous process monitoring.

  • Sampling discipline. The analyzer sits at the point of use, after the final dryer, with a leak-tight line — ambient air in-leakage at a fitting produces a false oxygen reading that looks exactly like a deox fault.

  • Method framework. ISO 21087:2019 catalogues the analytical methods for hydrogen fuel quality, giving laboratories a common vocabulary for what "verified" means.

For a laboratory hydrogen generator, the practical verification loop is continuous trace-O₂ readout plus dew-point monitoring, with alarms set inside the application limit (for example, alert at 0.5 ppm O₂ for a 1 ppm application). Buyers should insist on seeing the analyzer specification — sensor type, range, and calibration traceability — rather than accepting a "low oxygen" datasheet claim.

Process gas chromatograph instrumentation used for continuous gas quality monitoring

Process gas-analysis instrumentation: continuous trace-oxygen monitoring at the point of use closes the purification loop. Image: Linda Bartlett / U.S. National Cancer Institute, public domain.

7. Buyer Checklist for Deox-Equipped Systems

  • Ask where the deox stage sits. Correct answer: after coarse drying, before final drying — never at the very end of the train.

  • Ask for exit performance at end-of-life. "Below 0.1 ppm with a fresh bed" is easy; the useful number is performance after the catalyst's warranted lifetime.

  • Ask what protects the catalyst. Demister plus coarse dryer upstream, inlet oxygen interlock, bed thermocouples with trip — each item is cheap insurance against a full bed replacement.

  • Ask how the oxygen number is measured. Analyzer type, sampling point, calibration traceability, and alarm thresholds, per ISO 21087.

  • Match the limit to the load. Fuel-cell work needs ≤5 ppm; GC carrier gas and ECD make-up need ≤1 ppm; blending applications can relax. Buying one spec for all uses either wastes money or under-specifies the lab.

FAQ

Q: Why does a PEM electrolyzer produce oxygen in the hydrogen stream at all? A: The membrane that separates the anode (oxygen) and cathode (hydrogen) is not a perfect gas barrier. Dissolved oxygen permeates through it, and separator/transient events can add small amounts. Tens to hundreds of ppm at the outlet is normal system behavior.

Q: Can I just dry the gas instead of using a deox stage? A: No. Dryers remove water, not oxygen. Oxygen crosses the membrane chemically dissolved and leaves the cathode as a gas; only a catalytic recombination stage (or an entirely different purification technology) removes it.

Q: What oxygen limit applies to fuel-cell-grade hydrogen? A: ISO 14687:2019 and SAE J2719 set 5 ppm O₂ for fuel-cell grades. Analytical applications are usually tighter — GC-grade hydrogen targets about 1 ppm or less.

Q: Does the deox reaction add water to the gas? A: Yes — that is the reaction product. Each mole of oxygen removed produces two moles of water vapor, which is exactly why the deox reactor must be placed upstream of the final drying stage.

Q: How hot does the deox catalyst get? A: At normal ppm-level carryover, barely above its operating setpoint. The design constraint is the fault case: roughly 150–170 °C of bed temperature rise per 1% inlet oxygen on a hydrogen-rich feed, which is why inlet oxygen interlocks exist.

Q: How do I verify the deox stage is working? A: Continuous trace oxygen analysis at the point of use — electrochemical or zirconium-oxide sensors — plus dew-point monitoring after the final dryer. ISO 21087 provides the method framework, and the sampling line must be leak-tight so air in-leakage is not mistaken for a deox fault.

Conclusion

Oxygen carryover is a built-in characteristic of PEM electrolysis, and it is manageable with a mature, well-understood purification stage. The engineering is not exotic: protect the catalyst from liquid, run hydrogen-rich with an inlet oxygen cap, place the reactor before final drying, and verify the result with a trace oxygen analyzer at the point of use. What separates a robust system from a fragile one is not the deox chemistry — it is the placement discipline, the interlocks, and the willingness to publish measured numbers instead of datasheet claims. Buyers who ask those four questions will find that most of the answers are visible in the equipment's process diagram, before any purchase order is signed.

References

  • [1] ISO 14687:2019, Hydrogen fuel quality — Product specification, International Organization for Standardization (oxygen limit for fuel-cell grades).

  • [2] SAE J2719, Hydrogen Fuel Quality for Fuel Cell Vehicles, SAE International (fuel-cell-grade impurity limits).

  • [3] M. Carmo, D. L. Fritz, J. Mergel, D. Stolten, "A comprehensive review on PEM water electrolysis", International Journal of Hydrogen Energy, 38 (2013) 4901–4934. https://doi.org/10.1016/j.ijhydene.2013.01.151

  • [4] A. L. Kohl, R. B. Nielsen, Gas Purification, 5th ed., Gulf Publishing Company, 1997 (catalytic oxidation and gas purification practice).

  • [5] ISO 21087:2019, Gas analysis — Analytical methods for hydrogen fuel, International Organization for Standardization.

  • [6] Hovogen, "Hydrogen Purity Standards: What 99.999% H₂ Really Specifies" — grade system, impurity tables and certificate language. https://www.hovogen.com/post/hydrogen-purity-standards-what-99-999-h2-really-specifies

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 purification, and lifetime economics — the parameters that decide whether on-site hydrogen stays inside specification over a five-year service life.

 
 
 

Comments


HOVOGEN (Headquarter)

Dongguan, China

Floor 4, Room 411, No. 10, Keji 2nd Road, Zhongke Innovation Plaza, Songshan Lake High-Tech Industrial Development Zone, Dongguan City, Guangdong, China 523000
Tel/WhatsApp: +86-15916927868 Email: sales@hovogen.com

HOVOGEN (Manufacture Base) Zhuzhou, China

Add:Unit 1208, Building C1, Xinma Jingu Phase III, No. 266 Xinma South Road, Majiahe Street, Tianyuan District, Zhuzhou City, Hunan

HOVOGEN (Office)

HongKong, China

Add:Unit 18, 10/F

Fortune Commercial Building

362 Sha Tsui Road

Tsuen Wan 

Hong Kong

Tel: +852-8402 1602

HOVOGEN (Japan Branch)

Tokyo, Japan

東京都八王子市長沼町104-2ヒルサイドテラス1-4

bottom of page