I am currently working on calculating the fluence-to-personal dose equivalent conversion coefficients for Hp(10) by definition using the ICRU slab geometry. Specifically, my goal is to reproduce the values reported in ICRU Report 57.
Here’s what I’m doing:
I implemented a slab with dimensions 30 cm × 30 cm × 15 cm, composed of unit density tissue with the elemental composition specified by ICRU: H 10.1%, C 11.1%, N 2.6%, O 76.2% by weight.
The slab is irradiated by a parallel, uniformly expanded neutron beam over its front surface.
The medium between the source and the ICRU slab is assumed to be vacuum.
The scoring volume is a small cylinder located within the slab, 1 cm from the front surface, made of the same composition as the slab.
Dose is scored using the USRBIN DOSEQLET option with type region.
According to the manual, the output of the USRBIN from the ASCII file is given in GeV·cm³/g, so I convert this to Sv by multiplying by the conversion factor 1.6 × 10⁻⁷ and dividing by the volume of the scoring region. To get the conversion coefficient, the dose in Sv is then divided by the original fluence.
I have performed this evaluation for monoenergetic neutron beams at various energies, but I am finding discrepancies compared to the values reported in ICRU 57, particularly for neutrons above 20 MeV.
I have also used the LOW-PWXS card to activate point-wise cross sections for low-energy neutrons, specifically employing the recent ENDF-VIII.0 data.
I would appreciate any feedback on whether my approach is correct, as well as any tips to improve the accuracy of my calculations.
reading the description of your calculation method, all looks correct to calculate Hp(10) for neutrons (or any other particle for which the quality factor Q(L) \neq 1.
I checked ICRU 57, and it gives values for Hp(10) only up to 20 MeV. Values at higher energy, up to 200 MeV are given for H*(10), which is calculated in 10 mm depth in the ICRU-sphere, made from ICRU 4-component tissue with 30 cm diameter.
Small discrepancies (a few percent) with tabulated data may have different resons:
different MC code used
what happens when you change the ENDF V!!!.0 cross sections against others ? Note that the calculations in ICRU 57 are nearly 30 years old and neither the codes nor the data were as good as today.
target has a different extension: how thick (in beam-direction) s your scoring region? What diameter ? The size of the scoring region is not prescribed in the ICRU report, and every author uses a different size. I did calculation for photons and electrons and I had a 1 mm thick target, 1 cm in diameter, extending from 0.95 cm to 1.05 cm depth, I had deviations of < 1 % from the tables.
I have performed simulations for 50 MeV neutrons to calculate H*(10), using the ICRU sphere geometry. The scoring region I used is a cylinder 1 cm in diameter and 1 mm thick, extending from 0.95 cm to 1.05 cm depth, exactly the configuration you mentioned for your photon and electron simulations.
Using the ENDF-VIII.0 cross section library, I obtain a fluence-to-ambient dose equivalent conversion coefficient of approximately 456 pSv·cm². However, ICRU Report 57 gives a reference value of 400 pSv·cm². Additionally, the conversion coefficients provided by Pelliccioni (attached pdf file) give a value of 359 pSv·cm² for the same energy. ReferencePelliccioni.pdf (230.8 KB)
Given these discrepancies, I would like to ask: which reference should be considered most appropriate to validate my FLUKA simulations?
As you suggested, I have also tested older nuclear data libraries such as ENDF-VI, but the results are very similar.
Moreover, I have another question regarding the use of older libraries: ENDF-VI lacks cross section data for certain isotopes, such as oxygen-18.In this case, would it be acceptable to supplement missing data with cross sections from newer libraries like ENDF-VIII.0?
the data in ICRU Report 57 are an “average” of available publications at that time. “Average” was not in the strict statistical sense but more in the sense of producing “smooth lines”. For H*(10) of neutrons for E > 20 MeV, two publications were available, Sannikov and Savitskaya 1993 and Nablessi and Hertel, 1994. You find this in Table 6 of the ICRU Report. When you compare the data in the original publications with each other and with those in the report, you see sizeable differences, like those between your own results and the reference values.
Sannikov and Savitskaya explain that they used a partial kerma approximation in which after an interaction of the neutron in the phantom only protons and pions are transported, all other particles deposit their energy on the spot. Pelliccioni on the contrary transports all secondaries, this leads necessarily to a lower dose in the target volume. The authors also explain how they obtain their quality factors Q(L), this is different between each other and possibly different from the inner workings of the DOSEQLET weighing.
As you see, the conversion coefficients in ICRU Report 57 are a “committee average”, using the best available data at the time, judging their quality and then publishing a reasonable “average”. One has to read the original papers in detail to understand exactly how the results were obtained. By the way, it is not different for the neutron cross section libraries: the E in ENDF stands for evaluated , by a committee.
It will be difficult to reproduce the reference values exactly with a modern code and neutron library. You can validate your own calculations by explaining exactly what you did and trying to explain the differences to earlier calculations.