Comparison of proton energy loss in thick absorbers in terms of a reduced calibration curve

Monte Carlo simulations are essential for the support of particle experiments and developments of novel particle registration systems ranging from detectors developed for high-energy physics experiments at CERN to those for medical tomography. For proton beams, popular Monte Carlo codes like TRIM/SRIM, MCNPX and GEANT4 generate very similar final energy spectra for relatively thin absorbers, with differences unlikely to be detected in experiments. For thick absorbers, however, the disagreement is much larger, even for a moderate energy resolution. The reason for this is unclear because the actual overall accuracy of the proton stopping power in the Bethe–Bloch domain is known to be about 1%. One approach to investigate these differences is to compare, for example, the data from the NIST PSTAR and the SRIM reference data tables with the output of the Monte Carlo codes. When the various codes are validated against these tables, the differences in the simulated spectra mainly reflect the differences in the reference tables. Of more practical interest is the validation of the codes against experimental data for thick absorbers. However, only few experimental data sets are available here, and the existing data have been acquired at different initial proton energies and for different absorber materials. In order to compare the results of Monte Carlo simulations with existing experimental data, we applied the so-called reduced calibration method. This reduced calibration curve represents the range–energy dependence normalizing the range scale to the full projected range (for a given initial proton energy in a given material), and the proton energy scale to the given initial proton energy. The advantage of this approach is that the reduced calibration curve is nearly energy and material independent, and, thus, experimental, simulated and published reference data obtained at different energies and for different materials can be compared in one graph.