Outgassing: P ∝ t−1 — surface desorption, continuous power-law decay.
Virtual leak: trapped volume empties through a high-impedance path — temporary plateau then resumes.
Real Leak
Pressure hits a permanent plateau and never improves, no matter how long you pump. Rate-of-rise test shows a linear pressure increase when the valve is closed. Baking does not help.
Action: He leak check all joints, flanges, welds
Outgassing
Continuous slow descent, never truly flat. Dominates after 24h+ of pumping. Accelerated dramatically by baking. Dominant species: H₂O, CO, H₂.
Action: Continue bake, monitor RGA spectrum
Virtual Leak
Temporary plateau that eventually breaks through. Pressure descent is irregular — stalls, resumes, stalls. Often appears after mechanical work (tightening, welding).
Action: Check trapped volumes, blind holes, double seals
Linear rise = Real Leak
Constant slope — gas flows in at a fixed rate from outside. dP/dt = Qleak/V = constant.
Decelerating rise = Outgassing
Slope decreases over time as re-adsorption competes with desorption. Surfaces re-equilibrate.
Fast then plateau = Virtual Leak
Trapped volume equalizes quickly, then pressure stops rising. The trapped gas is finite.
- Has the pressure reached a true plateau, or is it still descending slowly? Plateau → leak | Still falling → outgassing
- Is the rate-of-rise linear? Linear → real leak confirmed
- Is there M32 (O₂) in the RGA spectrum? M32 present → air leak confirmation
- Did the problem start after a mechanical intervention? Post-intervention → suspect virtual leak
- Has baking improved the situation? Improved → outgassing | No change → leak
✅ Good Setup
Turbo 300 L/s, wide tube (10 cm × 30 cm), 50 L chamber, low outgassing. Textbook pumpdown.
🚨 Pinched Tube
Turbo 300 L/s but through a 1 cm × 100 cm tube. Watch your expensive pump become useless.
⚠ Leaky System
Good setup but with a real leak. Watch the pressure plateau and refuse to go lower.
Q = V · dP/dt [Torr·L/s]
V = system volume, dP/dt = rate-of-rise slope. For a real leak, Q is constant.
q(t) = q0 · (t / t0)−α
α ≈ 0.7–1.2 for H₂O on stainless steel. q0 ≈ 10−8 Torr·L/s/cm² after 1h pump.
C = 12.1 · (D³ / L) · √(T / M) [L/s]
D = diameter (cm), L = length (cm), T = temperature (K), M = molecular mass (AMU). Valid for L/D > 10.
τ = Vtrapped / Cpath
Vtrapped = trapped volume, Cpath = conductance of the leak path. Typical: 0.01 cm³ through a 1µm gap → τ ~ minutes to hours.
| Scenario | Typical Q (Torr·L/s) | Rate-of-rise shape | α | Bake effect |
|---|---|---|---|---|
| Real leak (flange) | 10−7 – 10−5 | Linear | — | None |
| Real leak (micro) | 10−10 – 10−8 | Linear | — | None |
| H₂O outgassing (SS, 1h) | 10−8 /cm² | Decelerating | 0.7–1.0 | Strong |
| H₂ outgassing (SS, baked) | 10−12 /cm² | Decelerating | 0.5 | Moderate |
| Virtual leak (O-ring groove) | 10−6 initial | Fast → plateau | — | Partial |
| Virtual leak (blind tapped hole) | 10−8 initial | Fast → plateau | — | Partial |
[2] K. Jousten (ed.), Handbook of Vacuum Technology, 2nd ed., Wiley-VCH, 2016 — Ch. 6: Leak Detection.
[3] B. Henrist et al., "Outgassing rate measurements of stainless steel and copper vacuum chambers," Vacuum 60 (2001) 27–34.
[4] N. Hilleret et al., "The secondary electron yield of technical materials and its variation with surface treatments," EPAC 2000.
[5] SLAC-TN-23-003, "Vacuum bake-out procedures for LCLS-II cryomodules," 2023.