Reference

RGA Spectrum Reading Abacus

Every peak, every signature, every ratio. What a residual gas spectrum is telling you, and what to do about it.

Identifying a single mass? RGA Abacus →

Worked example: Air leak (N2/O2)

Preloaded from the reference base: the ratio that calls an air leak, and the two that confirm it. Same base as the tables below.

TestRatioExpectedToleranceReading
Air leak (N2/O2) M28 / M32 3.73 ±0.3 If matched = air leak
Air leak (Ar/N2) M40 / M28 0.012 ±0.002 Confirms air leak
Air leak (Ar/O2) M40 / M32 0.044 ±0.005 Third confirmation
  • M28N₂⁺/CO⁺Big one - see diagnostics
  • M32O₂⁺AIR LEAK if present
  • M40Ar⁺Argon - 0.93% of air

Source: the RGA reference base of this document. All sources for this reference base

Part 1: Mass table

What every peak means, in an accelerator UHV context. Click a row to expand. Click a header to sort. Filter by mass, ion, or text.

Part 2: Diagnostic signatures

A single peak is never enough. The pattern makes the diagnosis. Click a row to expand.

2.1: The 12 key signatures

2.2: Fluorine decision tree: "I see M69, what is it?"

Part 3: Diagnostic ratios

The numbers that decide.

3.1: Air leak vs outgassing

M32 + M40 in atmospheric ratio = leak. M28 without M32 and M40 = CO (normal outgassing). Edge case: scroll-pump backdiffusion shows M32 but O₂/N₂ < 0.20.

3.2: Separating N₂ from CO at M28

M14/M28 > 5% = significant N₂ (leak). M12/M28 > 2% = significant CO. In a clean baked system with no leak, M28 is essentially 100% CO.

3.3: HC vs natural isotope at M29

M29 is always present (~1.1% of M28 = ¹³CO). Only a problem if >> 1.1%. Confirm HC with M15 and M43.

3.4: PFPE ratios (Fomblin)

M69 rising over days/weeks = active backstreaming (turbo bearings failing). M69 constant = legacy contamination.

3.5: PDMS ratios (silicone)

3.6: Methanol vs O₂ at M32

M32 = O₂ is the default assumption. After cleaning with methanol, M32 can be methanol parent. If M31 is strong and M40 absent: methanol, not a leak.

Part 4: Rapid decisions

4.1: "Is the spectrum clean?" 7-point checklist

4.2: "Pressure is too high" 60-second triage

4.3: HC envelope: 14 AMU pattern (CₙH₂ₙ₊₁ series)

Each 14 AMU apart (CH₂ unit). The further the envelope extends to high mass, the heavier the source (oil > solvent > fingerprint). Second series CₙH₂ₙ₋₁: M27, M41, M55, M69, M83. Note: M69 in this HC series is C₅H₉⁺, not CF₃⁺.

Part 6: Reference spectra

6.1: Clean baked system (target)

 
H₂ towers over everything. M28 is 10-100x lower than M2, and it is CO (verified by absence of M32/M40). M32, M40, M69 absent. Everything above M47 at the noise floor. This is what GO looks like.
 
6.2: Contaminated system (common problems)
 
 
M18 dominant = water. M32 + M40 = air leak. M69 = fluorine. HC peaks beyond M47 = contamination. M2 not dominant = dirty. Fix in order: leak first, then bake (water), then contamination.
 
 
 
Part 7: Cracking patterns
 
Fragment fingerprints at 70 eV electron impact. Fixed by physics, independent of RGA brand.
 
 
7.9: RGA sensitivity factors
 
 
True partial pressure = displayed PP / sensitivity factor. H₂ is always underrepresented (reads 0.44x), CO₂ overrepresented. Most RGA software does not correct automatically.
 
 
 
Part 8: Advanced diagnostics
 
8.1: Quantitative N₂/CO separation
 
f(N₂) = (R − 0.009) / 0.063, where R = M14/M28 measured.
 
 
8.2: ESD artifact protocol
 
 
Degas protocol: run filament degas at high emission (5-10 mA) 30-60 min, wait 10 min, rescan. Peaks that decreased = ESD artifact (ignore). Peaks that stayed = real chamber gas. Always degas before a qualification scan.
 
8.3: Ion pump specific signatures
 
 
M40 alone (without M32) is not an air leak. It is almost certainly ion-pump Ar re-emission. Confirm by checking M32.
 
8.4: Expected baseline partial pressures (clean baked system)
 
 
 
 
Sources · 22
  1. [A]NIST Chemistry WebBook, Mass Spectra (70 eV electron impact).
  2. [A]NIST Electron-Impact Cross Section Database.
  3. [A]J. O’Hanlon, A User’s Guide to Vacuum Technology, 3rd ed., Wiley, 2003: cracking patterns, fragment ratios, sensitivity factors, outgassing theory.
  4. [A]K. Jousten (ed.), Handbook of Vacuum Technology, 2nd ed., Wiley-VCH, 2016: quadrupole MS, ionization cross-sections, ion-pump physics.
  5. [A]A. Roth, Vacuum Technology, 3rd ed., North-Holland, 1990: permeation theory, outgassing rates, virtual-leak signatures, conductance.
  6. [A]L.N. Rozanov, Vakuumnaya Tekhnika, Vysshaya Shkola, Moscow: getter and ion-pump discharge physics, outgassing data for stainless steels.
  7. [A]CERN Accelerator School: Vacuum Technology (CAS proceedings).
  8. [A]P. Chiggiato, CERN: ion-pump noble-gas behavior, Ar-instability mechanism.
  9. [A]N. Marquardt, Principles of Vacuum Physics, CERN-99-05.
  10. [A]BINP Novosibirsk (Anashin, Malyshev): PSD yields, outgassing for SS/Al/Cu.
  11. [A]KEK (SuperKEKB), SPring-8, IHEP, SSRF: beam-induced desorption, NEG performance, long-term RGA surveys.
  12. [A]NIST atomic weights and isotopic compositions.
  13. [A]Y.-K. Kim & M.E. Rudd, Phys. Rev. A, 1994: electron-impact ionization cross-sections.
  14. [B]MKS Instruments, MicroVision 2 Residual Gas Analyzer : Operating Manual, 2019.
  15. [B]Solvay, Fomblin PFPE product documentation (M69 / M119 / M169 triplet).
  16. [B]DuPont / Chemours, Viton and Kalrez guides (thermal decomposition, M100 marker).
  17. [B]Dow Corning / Shin-Etsu, PDMS data (M73 / M147 / M207 triplet, silicone migration).
  18. [B]Gamma Vacuum / Agilent, Ion Pump manuals (noble-gas pumping, cathode sputtering).
  19. [B]Stanford Research Systems (SRS), RGA Application Notes (AN100 series).
  20. [B]Inficon Transpector gas analysis reference.
  21. [B]MKS / Granville-Phillips RGA technical notes.
  22. [B]Pfeiffer Vacuum mass spectrometer reference (QMG/QMA).
Rated [A] primary or standard, [B] manufacturer documentation. 70 eV cracking patterns are nominal and vary 10-20% between instruments : validate against a known calibration gas before quantitative use. Facility-specific qualification specifications (Part 5 of the field edition) are omitted from this public document.