[Paper Review] Po-production in lead: A benchmark between Geant4, FLUKA and MCNPX

Alfredo Ferrari, D. Kiselev+3 more|arXiv (Cornell University)|Jun 10, 2018

Nuclear Physics and Applications4 citations

TL;DR

This benchmark study evaluates Po-isotope production in lead irradiated by 575 MeV protons using Geant4, FLUKA, and MCNPX with the INCL4.6/ABLA07 model. It demonstrates that accurate treatment of alpha and triton reactions—previously underestimated in default MCNPX—leads to excellent agreement with experimental data, highlighting the sensitivity of Po-activity predictions to nuclear reaction models beyond standard first-generation intra-nuclear cascades.

ABSTRACT

On the last SATIF a comparison between the measured activities of the polonium isotopes Po-208, Po-209 and Po-210 and the simulated results using MCNPX2.7.0 was presented. The lead samples were cut from the SINQ spallation target at the Paul Scherrer Institut (PSI) and irradiated in 2000/2001 by 575 MeV protons. The Po-isotopes were separated using radiochemical methods by the group of D. Schumann at PSI and measured. Choosing the default model in MCNPX, Bertini-Dresner, the prediction underestimated the measured activities by up to several orders of magnitude. Therefore the Liège intranuclear-cascade model (INCL4.6) coupled to the de-excitation model ABLA07 were implemented into MCNPX2.7.0 and very good agreement was found to the measurement. The reason for the disagreement was traced back to a suppression of alpha reactions on the lead isotopes leading to Po and neglecting the triton capture on Pb-208, which leads to Pb-210 and decays into Po-210 with a much longer life time (22.3 years) than the decay of Po-210 itself (138 days). The prediction of the Po-isotope activities turns out to be a sensitive test for models and codes as it requires the accurate treatment of reaction channels not only with neutrons, protons and pions but also with alphas and tritons, which are not considered in intra-nuclear cascade models of the first generation. Therefore it was decided to perform a benchmark by comparing the results obtained with MCNPX2.7.0 using INCL4.6/ABLA07 to the predictions of FLUKA and Geant4. Since the model of the SINQ spallation source requires an elaborate geometry a toy model was setup. The toy model has a simplified geometry preserving the main features of the original geometry. The results for the activities of the three Po-isotopes and Pb-210 as well as the energy spectra for alphas and tritons obtained with the three particle transport Monte Carlo codes are presented.

Motivation & Objective

• To assess the predictive accuracy of Geant4, FLUKA, and MCNPX in simulating polonium isotope production in lead under spallation irradiation.
• To identify deficiencies in default models of MCNPX, particularly the Bertini-Dresner intra-nuclear cascade, which significantly underestimated Po-208, Po-209, and Po-210 activities.
• To validate the improved MCNPX implementation using INCL4.6/ABLA07 against experimental data from SINQ spallation target irradiation.
• To establish Po-isotope production as a sensitive benchmark for nuclear reaction models, especially for reactions involving alpha particles and tritons.

Proposed method

• A simplified toy geometry was developed to preserve key features of the complex SINQ spallation target geometry for computational efficiency.
• Monte Carlo particle transport simulations were performed using Geant4, FLUKA, and MCNPX2.7.0 with the INCL4.6/ABLA07 model for nuclear reaction cascades and de-excitation.
• The simulation results for Po-isotope activities and energy spectra of alphas and tritons were compared across all three codes.
• Experimental data from radiochemical separation and measurement of Po-208, Po-209, Po-210, and Pb-210 in irradiated lead samples were used as the benchmark.
• The study focused on identifying discrepancies in reaction channel treatments, particularly alpha emission and triton capture on Pb-208.
• Sensitivity analysis was performed to isolate the impact of specific reaction channels, such as 208Pb(t,α)210Po and 208Pb(α,2n)210Po, on the final Po-isotope yields.

Experimental results

Research questions

• RQ1 Why did the default MCNPX model (Bertini-Dresner) fail to predict Po-isotope activities in lead, underestimating them by several orders of magnitude?
• RQ2 How do the predictions of Geant4, FLUKA, and MCNPX with the INCL4.6/ABLA07 model compare to experimental data for Po-208, Po-209, and Po-210 production in lead?
• RQ3 To what extent do alpha and triton-induced reactions contribute to Po-isotope production, and why are they critical for accurate simulation?
• RQ4 Can Po-isotope production serve as a sensitive benchmark for evaluating the performance of nuclear transport codes in simulating spallation reactions?
• RQ5 What specific reaction channels—particularly involving light ions like tritons and alphas—are responsible for the discrepancy in the original MCNPX simulation?

Key findings

• The default MCNPX model (Bertini-Dresner) underestimated Po-isotope activities by up to several orders of magnitude due to suppressed alpha reactions on lead isotopes.
• Implementation of the INCL4.6/ABLA07 model into MCNPX2.7.0 led to excellent agreement with experimental data, confirming its improved accuracy.
• Triton capture on Pb-208, producing Pb-210 which decays to Po-210 with a 22.3-year half-life, was identified as a key contributor to Po-210 activity, previously neglected in first-generation models.
• The energy spectra of alphas and tritons predicted by all three codes (Geant4, FLUKA, MCNPX) showed good agreement, validating their treatment of light-ion emission.
• The study confirms that Po-isotope production is a highly sensitive benchmark for nuclear reaction models, especially for reactions involving light ions like tritons and alphas.
• The benchmark demonstrates that accurate simulation of spallation processes requires careful modeling of non-standard reaction channels beyond protons, neutrons, and pions.