A high-purity gas is delivered in a package, not in isolation. The cylinder, valve and transfer path can introduce moisture or other impurities after the gas has been produced. A cylinder that looks clean outside, or has been emptied, is not automatically ready for a demanding new filling.
Thermal-vacuum preparation addresses part of this problem by combining controlled heat with evacuation. Its value lies in an appropriate, verified process for the actual package and gas, not in a single impressive temperature or vacuum reading.
What remains after a cylinder is emptied?
Residual free gas is only one possible source of contamination. Molecules can remain attached to surfaces, be retained in deposits or non-metallic components, or, for certain gas/material combinations, enter the material itself. Later release can affect the next filling. The useful term “gas memory” therefore covers several different mechanisms.
- Surface retention
- Adsorption retains molecules at a surface. Desorption releases them. Exposure to ambient air can introduce surface moisture even when the cylinder's previous gas was dry.
- Material inventory
- Absorption concerns entry into a material; diffusion concerns movement within it. These processes depend on the gas, material, temperature and history.
- Transport through a material
- Permeation is passage through a material, not a synonym for every kind of residual contamination.
Does the old gas penetrate the cylinder wall?
It can for particular gas/material combinations, but “every gas soaks into steel” is an incorrect generalization. Hydrogen is an important example: Sandia's reference for 304/304L stainless steel discusses solubility, diffusion and permeability, as well as material-condition effects. That evidence cannot be transferred unchanged to helium, neon, krypton, xenon or an unidentified cylinder alloy.
The source is not necessarily the previous filling. NIST's research on stainless-steel vacuum components (PDF) distinguishes water associated with surface exposure from hydrogen held within the metal. Its findings explain why removing free gas does not remove every gas source. They do not prescribe a treatment for commercial cylinders.
For procurement, the right question is therefore specific: what was the previous service, which materials contact the gas, what contamination or damage is plausible, and how was suitability for the proposed service established? A claimed penetration depth or universal cleaning time is not meaningful without that context. Degassing also must not be presented as a repair for corrosion, cracking or hydrogen-related damage.
What controlled heat and vacuum do
Heating can accelerate the release of retained species. Evacuation removes released gas and vapour from the connected internal space. The resulting gas load, surface condition and limitations of the complete assembly matter together. “Degassing” describes removal of gas from the material or assembly; it is broader than surface desorption alone.
There is an industrial cylinder-preparation context for this approach. Messer's description of gas-mixture characterization connects mixture stability with cylinder pretreatment, heated purge/evacuation cycles and conditioning. Linde's specialty-gas quality guidance separately identifies pretreatment and passivation in maintaining mixture stability. These are supplier examples, not evidence of a particular SG Trading facility or process.
Cleaning, drying, evacuation, thermal-vacuum treatment and passivation are not interchangeable services. Removing oil or reaction products is a different task from reducing retained moisture. Surface conditioning for a reactive calibration mixture has a different purpose again. The selected process must address the actual contamination and intended product.
No universal preparation recipe: allowable temperature, duration, vacuum and acceptance criteria depend on the approved procedure, cylinder material, valve, seals and gas service. This guide is not an instruction to heat a filled cylinder, vent unknown contents or perform preparation without qualified facilities. It does not replace inspection, compatibility assessment or oxygen-service cleanliness requirements.
Why 0.1–0.5 ppm is not a boiling temperature
Water vapour is already in the gas phase. Boiling concerns a liquid-to-vapour transition; a trace-water specification instead describes how much water is present. To relate that amount to condensation, specify pressure and whether the saturation boundary is liquid water or ice. A dew point over liquid water and a frost point over ice are not identical conventions below freezing.
Also distinguish ppmv or mole fraction from ppm by mass. Under an ideal-gas approximation, 0.1 ppmv corresponds to a water mole fraction of 0.1 × 10−6. Vaisala's humidity fundamentals explain the connection between partial pressure, saturation and pressure-dependent dew/frost point.
For an explicitly defined example at 1.000 bar absolute (100,000 Pa), use pH₂O ≈ xH₂O × P. Inverting the IAPWS ice-sublimation relation gives:
| Water content | Water partial pressure | Frost point over ice |
|---|---|---|
| 0.1 ppmv | 0.010 Pa | −89.8 °C |
| 0.2 ppmv | 0.020 Pa | −85.9 °C |
| 0.5 ppmv | 0.050 Pa | −80.6 °C |
Calculated from IAPWS R14-08(2011), equation 6, with ideal-gas partial pressure and no mixture enhancement correction. Rounded to 0.1 °C for readability, not as an analyser-accuracy claim. The values are not a direct conversion for a high-pressure cylinder.
Changing the pressure changes the conversion. Ask whether an analyser reports the value at its actual sample pressure, a pressure-corrected process value or a stated reference pressure. Comparing two numbers without that basis can create an apparent disagreement.
A very low frost point does not prove a clean wall
The table concerns water in the gas phase at equilibrium. It says nothing by itself about the total inventory retained on a surface or how quickly that inventory may be released. A dry sample can coexist with a material history that still needs assessment. The temperature and time required for removal cannot be read from the frost-point table.
Likewise, total vacuum pressure does not identify the remaining species. A low reading is not a final-product moisture analysis. Preparation should be assessed against defined criteria, and final gas quality should be measured using a suitable method and sampling arrangement.
As a simple quantity illustration, 100 mol of total gas at 0.1 ppm water contains approximately 0.18 mg of water; at 0.5 ppm, approximately 0.90 mg. This follows from nH₂O = xH₂O × n and a rounded water molar mass of 18 g/mol. It shows the scale of trace-water specifications, not the amount on any actual cylinder wall. A visually clean surface is not a measurement at that scale.
What evidence should a buyer request?
- Package history: dedicated service or assessed change of gas, material and identifiable cylinder or batch.
- Preparation scope: what risks were addressed, and whether cleaning, drying or gas-specific conditioning was required.
- Traceable records: how the preparation and release record connect to the supplied package.
- Acceptance basis: relevant checks and approved criteria, without asking for disclosure of a proprietary operating recipe.
- Analytical evidence: the required impurity limits, units, reporting limits and sample conditions in the agreed quality documentation.
- Storage and use conditions: any limitations relevant to maintaining the specified product over its stated validity period.
These are practical procurement questions, not a claim that all suppliers use one identical release system. The quality specification, package requirements and applicable SDS should be reviewed together.
Connect preparation with identification and analysis
Use cylinder markings and labels to connect the physical package to its documents. Review a proposed change of gas service before assuming that evacuation alone is sufficient. For the measurement side, see PDD, laser and coulometric gas analysis.
In an RFQ, name the gas, critical impurities, packaging, ownership, known previous service and required evidence. “Vacuum dried” can be a useful process description, but it is not a substitute for an agreed product specification and traceable quality results.
Scope of this guide
Prepared by SG TRADING L.L.C-FZ. Sources reviewed 23 October 2026. General educational and procurement guidance, not an operating procedure or a determination that a particular item is approved for service. No specific cylinder markings or customer equipment have been decoded or certified in this article.