Silicon K-fluorescence yield: 0.93% (G4EMLOW) vs 4.5% (literature)

[montarou]

---

• Geant4 version: 11.3.1 (patch-01)
• G4EMLOW version: 8.6.1
• OS: Ubuntu 24
• Physics List: FTFP_BERT + G4EmLivermorePhysics
• Analysis: ROOT 6.x

Summary
The K-shell fluorescence yield for Silicon (Z=14) in G4EMLOW8.6.1 is 0.93%,
compared to the experimental reference value of 4.5% (Krause 1979),
representing a systematic underestimation by a factor of 4.8.

This affects simulations of Si detectors (SDD, Si(Li))

Geometry:
• Source: Eu-152 (14 keV gamma)
• Detector: Silicon cylinder (450 µm thick, 7.98 mm diameter)
• Simplified SDD geometry with Be window

Physics Configuration:
• ReplacePhysics(new G4EmLivermorePhysics)
• RegisterPhysics(new G4RadioactiveDecayPhysics)
• em->SetFluo(true)
• em->SetAuger(true)
• em->SetAugerCascade(true)
• em->SetDeexActiveRegion(“DefaultRegionForTheWorld”, true, true, true)
• em->SetDeexcitationIgnoreCut(true)
• Production cuts: gamma=0.01µm, e-=0.01µm

Log verification confirms:
✓ Fluorescence enabled: 1
✓ Auger electron cascade enabled: 1
✓ De-excitation module ignores cuts: 1
✓ Process “phot” using LivermorePhotoElectric with Fluo

Simulation Setup:
• 1000 photons at 13.95 keV incident on Si (450 µm)
• All fluorescence flags enabled and verified in logs
• PhysicsList correctly configured with SetDeexcitationIgnoreCut(true)

Results:
• 3015 photoelectric interactions (99.9% - correct)
• 28 K-alpha photons at 1720 eV created
• Observed yield: 28/3015 = 0.93%

G4EMLOW Analysis:
• File fl-tr-pr-14.dat sum: 0.93% ✓ (matches simulation)
• Literature (Krause M.O., J. Phys. Chem. Ref. Data 8, 307, 1979): 4.5%
• Discrepancy: Factor 4.8×

G4EMLOW Analysis:
• File fl-tr-pr-14.dat sum: 0.93% ✓ (matches simulation)
• Literature (Krause M.O., J. Phys. Chem. Ref. Data 8, 307, 1979): 4.5%
• Discrepancy: Factor 4.8×

Extended tosome other light elements
Similar underestimation for Z < 20:
• C (Z=6): 0.10% vs 0.27% (factor 2.7×)
• O (Z=8): 0.30% vs 0.76% (factor 2.5×)
• Al (Z=13): 1.50% vs 3.64% (factor 2.4×)
• Si (Z=14): 0.93% vs 4.50% (factor 4.8×)
• S (Z=16): 2.80% vs 6.59% (factor 2.4×)

Good agreement for Z > 25 (Fe, Cu, etc.)

Impact on my simulation
For Si detectors: escape peaks underestimated 5×

QUESTIONS

1. Is this a known limitation of G4EMLOW for light elements?
2. Are there plans to update with more complete fluorescence data?
3. Should we apply correction factors for Si detector simulations?

[civanch]

Hello,

thank you very much for clear report. Your report is copied to the geant4 Bugzilla bug report system:

2683 – Silicon K-fluorescence yield: 0.93% (G4EMLOW) vs 4.5% (literature)

Please, visit and follow the evolution.

VI