Benchmarking Geant4 hadronic models for prompt-γ monitoring in carbon ion therapy

The real-time monitoring of the spread-out Bragg peak would allow the planned dose delivered during treatment to be directly verified, but this poses a major challenge in modern ion beam therapy. A possible method to achieve this goal is to exploit the production of secondary particles by the nuclear reactions of the beam with the patient and correlate their emission profile to the planned target volume position. In this study, we present both the production rate and energy spectra of the prompt-γ produced by the interactions of the ¹² C ion beam with a polymethyl methacrylate (PMMA) target. We also assess three different Monte Carlo models for prompt-γ simulation based on our experimental data.

The experiment was carried out at the GSI Helmholtz Centre for Heavy Ion Research, Darmstadt, Germany with a 220 MeV/u ¹² C ions beam impinging on a 5× 5× 20 cm³ polymethyl methacrylate beam stopping target, with the prompt-γ being detected by a hexagonally-shaped barium fluoride scintillator with a circumscribed radius of 5.4 cm and a length of 14 cm, placed at 60° and 90° with respect to the beam direction. Monte Carlo simulations were carried out with three different hadronic models from the Geant4 code: binary ion cascade (BIC), quantum molecular dynamics (QMD), and Liege intranuclear cascade (INCL⁺⁺ ).

An experimental prompt-γ yield of 1.06 × 10^-2  sr^-1 was measured at 90°. A good agreement was observed between the shapes of the experimental and simulated energy spectra, especially with the INCL⁺⁺ physics list. The prompt-γ yield obtained with this physics list was compatible with the measurement within 2σ, with a relative difference of 26% on average. BIC and QMD physics lists proved to be less accurate than INCL⁺⁺ , with the difference between the measured and simulated yields exceeding 100%. The differences between the three physics lists were ascribed to important discrepancies between the models of the physical processes producing prompt-γ emissions.

In conclusion, this study provides prompt-γ yield values in agreement with previously published results for different carbon ions energies. This work demonstrates that the INCL⁺⁺ physics list from Geant4 is more accurate than BIC and QMD to reproduce prompt-γ emission properties.