Validation of Geant4 fragmentation models for broad beam heavy-ion therapy

Objective.Heavy-ion therapy (HIT) is valued for its superior dose localisation and biological effectiveness; however, nuclear fragmentation remains a major challenge, creating a complex secondary radiation field. Monte Carlo simulations are widely used to study this radiation field, which can be used to estimate healthy tissue risks, making it important to assess how accurately fragmentation models reproduce measurements. Previous validation studies have predominantly focused on millimetre-scale pencil-beam geometries, whereas clinical irradiation fields are often centimetre-scale. This difference in measurement geometry may influence the apparent agreement between simulation and experiment.Approach.Four Geant4 nuclear fragmentation models were evaluated: BIC, INCL, QMD and LiQMD. Passive broad-beam measurements of 290 MeV/uC and 400 MeV/uNe ions, each with a diameter of, were compared with simulations using fragment build-up curves and primary-beam attenuation. Additional comparisons were performed for 400 MeV/uC and 670 MeV/uC,N andO pencil beams, including angular-distribution data where available.Main results.For theC passive broad beam, LiQMD showed the best agreement with experiment, with mean absolute percentage errors belowfor all fragments except lithium. LiQMD also gave the best agreement for theC ion pencil-beam fragment-yield comparison. For theNe ion passive broad beam, no single model consistently outperformed the others, although LiQMD remained among the most accurate overall. TheN andO ion pencil-beam comparisons were less definitive for LiQMD because much of the 670 MeV/u dataset lies above its 500 MeV/u upper energy range. Overall, pencil-beam comparisons generally yielded better agreement than passive broad-beam comparisons, indicating that angular distributions and measurement geometry can strongly influence apparent model agreement.Significance.Conclusions drawn from pencil-beam fragmentation benchmarks do not necessarily translate directly to passive broad-beam geometries. Broad-beam measurements provide a complementary test of Geant4 fragmentation models because they additionally constrain fragment angular distributions, lateral transport and detector acceptance.