Bottom line first: A directly oil-immersed hydrogen sensor removes the “degassing” step, but not the “conversion” step. What the sensor measures is the activity / partial pressure of hydrogen in the oil. Turning that into the ppm reading you are used to still requires a coefficient strongly tied to oil type and temperature. That coefficient — the Ostwald coefficient — is itself one of the largest sources of uncertainty in the whole industry.
Reading “no degassing” as “no conversion” is one of the easiest mistakes to make at selection time.
1. What the Ostwald coefficient is
Gas dissolved in oil distributes between the oil and gas phases in proportion:
K (Ostwald coefficient) = concentration in oil / equilibrium concentration in gas phase
The larger K is, the more a gas “prefers” the oil and the harder it is to drive out.
| Gas | K (approx., 50 °C, mineral oil) | Extraction difficulty |
|---|---|---|
| H₂ | 0.06 | Easiest |
| CH₄ | 0.39 | Medium |
| C₂H₂ | 1.02 | Harder |
| C₂H₄ | 1.46 | Hard |
| C₂H₆ | 2.30 | Hardest |
This is also why headspace recovery looks so low: most of the ethylene and ethane stays in the oil. In a headspace sample the C₂H₆ recovery can be around 1%, while H₂ can reach about 8%.
2. Trap one: the coefficient tables are not consistent
Different standards, temperatures and laboratories give Ostwald coefficients that do not agree with one another:
- IEC 60599 provides values for 20 °C and 50 °C; GB/T 17623 provides one set at 50 °C — the numbers themselves differ.
- In measurement, the direction of H₂’s K with temperature can even oppose theory (rising from 0.056 to 0.074 between 25 °C and 70 °C).
- Deviation between sources: >20% at 70 °C, about 30% at 25 °C.
The industry’s own words: “coefficients from different laboratories are not reliable and not reproducible.”
Every route that back-calculates through a coefficient — headspace, membrane extraction, and a directly immersed palladium-alloy sensor — hits its accuracy ceiling here.
3. Trap two: the coefficient changes completely with oil type
The ppm ↔ %RS conversion holds only for mineral oil.
Take the saturated water content at 20 °C as an example:
| Oil type | Saturated water content | Ratio to mineral oil |
|---|---|---|
| Mineral oil | ~50 ppm | 1× |
| Natural ester | ~1100 ppm | ~22× |
| Synthetic ester | ~2200 ppm | ~44× |
Ester-based oils have solubility behavior completely different from mineral oil. Apply a mineral-oil coefficient to an ester-oil transformer and the error can reach tens of times.
4. So what does “directly immersed, no degassing” actually win?
Not “no coefficient” — it wins on these points:
- No degassing hardware: no vacuum pump, membrane or headspace chamber — no moving parts, no consumables, longer life.
- No degassing error: the “incomplete extraction” uncertainty of vacuum and membrane methods simply disappears.
- Faster response: no waiting for gas-liquid equilibrium; hydrogen enters the sensor directly.
- Zero maintenance: no carrier gas, no span gas, no calibration.
But be honest with the customer: the ppm reading still requires a coefficient correction for the medium and temperature. Anyone who claims that “no degassing means no conversion at all” is overstating the case.
5. The right questions to ask at selection
| Don’t ask | Ask instead |
|---|---|
| “Is it degassing-free?” | “What oil and what temperature coefficient is the ppm conversion based on?“ |
| “What is the accuracy?” | “What is the accuracy measured in mineral oil, and in ester oil, separately?“ |
| “Does it need calibration?” | “Must the coefficient be recalibrated after switching oil type?“ |
Solubility data cited from IEC 60599, GB/T 17623 and standard laboratory measurements; the divergence in Ostwald coefficients from published literature reviews.
William Xiong — Shenzhen Feso Automation Technology Co., Ltd (PAS DGA) · DGA-500 palladium-alloy dissolved-hydrogen sensor: directly oil-immersed, no degassing, no consumables
inquiry@pasdga.com · https://pasdga.com
If you would like to discuss conversion methods for dissolved hydrogen and how they are benchmarked in the field, we welcome the conversation.