A study of Type B uncertainties associated with the photoelectric effect in low-energy Monte Carlo simulations.

The goal of this manuscript is to estimate Type B uncertainties in absorbed-dose calculations arising from the different implementations in current state-of-the-art Monte Carlo codes of low-energy photon cross-sections (<200 keV). Monte Carlo simulations are carried out using three codes widely used in the low-energy domain: PENELOPE-2018, EGSnrc, and MCNP. Mass energy-absorption coefficients for water, air, graphite, and their respective ratios; absorbed dose; and photon-fluence spectra are considered. Benchmark simulations using similar cross-sections have been performed. The differences observed between these quantities when different cross-sections are considered are taken to be a good estimator for the corresponding Type B uncertainties. A conservative Type B uncertainty for the absorbed dose (k=2) of 1.2%-1.7% (<50 keV), 0.6%-1.2% (50-100 keV), and 0.3% (100-200 keV) is estimated. The photon-fluence spectrum does not present clinically relevant differences that merit considering additional Type B uncertainties except for energies below 25 keV, where a Type B uncertainty of 0.5% is obtained. Below 30 keV, mass energy-absorption coefficients show Type B uncertainties (k=2) of about 1.5% (water and air), and 2% (graphite), reaching values about 1% (40-50 keV) and 0.5% (50-75 keV). Type B uncertainties for the water-to-graphite ratios are observed for energies below 30 keV, about 0.7% (k=2). In contrast with the intermediate (about 500 keV) or high (about 1 MeV) energy domains, Type B uncertainties due to the different cross-sections implementation cannot be considered subdominant with respect to Type A uncertainties or even to other sources of Type B uncertainties. Therefore, the values reported here should be accommodated within the uncertainty budget in low-energy photon dosimetry studies.