Assessment of cross-section library impact on neutron beam quality in AB-BNCT monte carlo studies

Abstract

This study investigated the impact of nuclear reaction cross-sections data from various nuclear data libraries on neutron parameters in Monte Carlo simulations using the PHITS code for accelerator-based BNCT (AB-BNCT). Specifically, it evaluated the effect of different proton data libraries-ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0-on the neutron yields from the ⁷Li(p, n)⁷Be reaction in a natural lithium target. The results showed that the calculated target neutron yields differed by up to 16.86%, with notable variations in the target neutron energy spectra. Furthermore, the study assessed the impact of neutron libraries-ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0-on neutron beam parameters at the BSA exit, based on the neutron spectrum at the lithium target. It was found that the epithermal neutron intensity and fast neutron component calculated with the ENDF/B VIII.1 neutron library were 5.51%-8.61% and 19.04%-63.94% higher, respectively, than that obtained with the other neutron libraries, due to a more prominent neutron flux peak. The fast neutron component, calculated using the JENDL-5.0 neutron library, was 23.42%-63.94% lower compared to other libraries, while the gamma-ray component derived from the JEFF-3.3, CENDL-3.2, and JENDL-5.0 neutron libraries was 6.79%-10.08% higher than that from the ENDF/B series. Further analysis revealed that the discrepancies were primarily caused by differences in the neutron cross-section data of F, Pb, Mg, Cu, and H in specific neutron libraries. In particular, the differences in the neutron cross-sections of F and Pb accounted for the variations in the obtained BSA outlet spectra and neutron parameters when using the ENDF/B VIII.1 and JENDL-5.0 neutron libraries. The elevated gamma-ray component observed in the JEFF-3.3 and CENDL-3.2 neutron libraries, compared to the ENDF/B series, was largely due to the different cross-sections of Mg, Cu, and H. These findings underscored the considerable discrepancies arising from variations in nuclear data libraries and emphasized the critical importance of appropriate selection of nuclear data libraries for BNCT applications.

Introduction

Boron neutron capture therapy (BNCT) is a specialized form of radiotherapy that has garnered increasing attention in recent decades for its potential to treat intractable cancers¹. The BNCT differs from conventional radiotherapy techniques through its unique mechanism, which leverages the nuclear reaction between ¹⁰B and thermal neutrons to selectively target cancer cells at the cellular level. A significant challenge in BNCT is creating a neutron beam with the specific properties necessary for effective therapy. Clinical BNCT requires a neutron beam with an energy spectrum tailored to optimize the boron-neutron capture reaction in tumors while minimizing damage to surrounding healthy tissues. The optimal neutron spectrum for BNCT typically comprises a high flux of epithermal neutrons (0.5 eV to 10 keV), which can penetrate tissue to reach tumors embedded in deeper structures². Upon entering the tissue, epithermal neutrons slow down and thermalize, enhancing their probability of being captured by boron atoms.

The quality of the neutron beam in BNCT is determined by a beam shaping assembly (BSA), which conditions the neutron beam generated by a primary neutron source-typically a proton accelerator striking a target material such as beryllium or lithium, through the ⁹Be(p, n)⁹B and ⁷Li(p, n)⁷Be reactions. The BSA modifies the beam’s spectral composition, fluence rate, and spatial profile to meet therapeutic requirements. It consists of a carefully designed combination of moderators, filters, and reflectors that reduce the neutron energy spectrum to the desired epithermal range while filtering out fast neutrons and gamma radiation³. Designing an effective BSA requires accurate modeling and optimization, typically achieved using Monte Carlo simulations^3–5. This modeling depends on reliable nuclear data, particularly reaction cross-sections from nuclear data libraries.

Many nuclear data libraries provide cross-section data for neutrons and other particles, including evaluated nuclear data file, version B (ENDF/B)⁶, joint evaluated fission and fusion file (JEFF)⁷, Japanese evaluated nuclear data library (JENDL)⁸ and TALYS-based evaluated nuclear data library (TENDL)⁹. Each of these libraries has distinct methods of data acquisition, evaluation, and presentation, leading to differences in cross-section values and, consequently, in simulation outputs. Variations across libraries are particularly relevant for BNCT, as obtaining the neutron beam parameters at the BSA outlet requires precise neutron production and transport modeling within the BSA. Therefore, choosing a nuclear data library can have substantial implications for BSA design. Studies have shown that variations in cross-section values can affect Monte Carlo simulations and result in differences in predicted neutron fluence, dose deposition and energy spectra¹⁰. These differences are magnified in the BSA, where the neutron moderation, scattering, and filtering processes are highly sensitive to the accuracy of cross-section data, which is crucial for achieving the desired beam characteristics. Current radiotherapy standards require a less than 5% deviation between measured and calculated doses¹². Consequently, selecting nuclear data libraries closely aligned with experimental measurements is crucial for meeting clinical requirements.

This study was motivated by the observation of significant discrepancy in the neutron energy spectrum generated by proton bombardment of a lithium target when using the ENDF/B VIII.0 proton library compared to the result from the ENDF/B VII.1 library. After consulting relevant literature and communicating with the contributors to the ENDF/B VIII.0 library, it was found that the proton library for ⁷Li in ENDF/B VIII.0 underwent substantial modifications⁶, which were subsequently carried over into the latest ENDF/B VIII.1 library. Given that ⁷Li is the key nuclide in the ⁷Li(p, n)⁷Be reaction and plays a pivotal role in the design and simulation of AB-BNCT systems, it is essential to comprehensively evaluate the impact of different nuclear data libraries on these processes. This study aimed to thoroughly evaluate the influence of different nuclear data libraries on the simulation processes of AB-BNCT devices. By comparing and analyzing the discrepancies in computational results, it seeks to quantify the impact of library selection on BSA output parameters and assess the implications of cross-section discrepancies for clinical BNCT.

Materials and methods

Target and BSA model

The target and BSA models used in this study are derived from a previous research⁵. As illustrated in Fig. 1, all components are arranged along the z-axis. The BSA has a cylindrical structure. A 2.8 MeV proton beam with a 20 mA current strikes a natural lithium target, which has a radius of 6 cm and a thickness of 140 μm. Neutrons are generated through the ⁷Li(p, n) reaction. To dissipate heat, a copper tube is positioned behind the lithium target, with circulating water for cooling. The moderator layer consists of magnesium fluoride (MgF₂) with a radius of 25 cm and a thickness of 31.66 cm. Behind the moderator, a 0.049 cm thick lithium fluoride (LiF) thermal neutron filter layer and a 0.2 cm thick lead (Pb) gamma-ray filter layer are arranged sequentially, both with a radius of 50 cm. The collimator, made of a 10.0 cm thick lithiated polyethylene (PE) composite, has an entrance radius of 11 cm and an exit radius of 5 cm. A lead reflector surrounds the target and moderator radially.

Fig. 1.

Target and BSA models utilized in this study.

Monte carlo simulations

A two-step Monte Carlo simulation approach was employed to evaluate the impact of different nuclear data libraries on the neutron beam parameters at the BSA exit. In the first step, simulations focus solely on the lithium target to analyze how various proton data libraries influence the target neutron energy spectrum generated by proton bombardment. In the second step, these target neutron energy spectra were used as the primary neutron sources, and different neutron data libraries were used to simulate the transport of neutrons within the BSA and calculate the neutron parameters at the BSA exit. All simulations were conducted using PHITS code (version 3.330)¹¹.

The proton-target reactions

This section presents the simulation of neutron production through the proton beam bombardment of a lithium target using the PHITS program. A 2.8 MeV, 20 mA proton beam with a radius of 5 cm is directed onto a natural lithium metal target, 6 cm in radius and 140 μm thick. Since the threshold energy of the ⁷Li(p, n)⁷Be reaction is 1.88063 MeV, the maximum energy of neutrons produced by this reaction will be less than 1.1 MeV. The 0-1.1 MeV energy range was divided into 110 bins to obtain the complete neutron energy spectrum. The number of particle histories is 2 × 10¹⁰ to achieve a maximum statistical error of less than 0.5% in each bin. The angle between the particle motion direction and the z-axis was divided into 19 bins, ranging from 0° to 180° to better preserve the variation of the energy spectrum with angle.

Neutron transportation within BSA

In this part, various nuclear reactions occurring within the BSA were simulated. Five parameters proposed in the IAEA report: the epithermal beam intensity , the ratio between the total neutron current and the total neutron flux , the thermal-to-epithermal flux ratio , the fast neutron component , and the gamma-ray component , were employed to quantify the neutron beam characteristics at the collimator exit, with their recommended values being 1.0 × 10⁹, 0.7, 0.05, 2.0 × 10^–13, and 2.0 × 10^–132. The decision not to adopt the recommended values from the latest IAEA report stems from a notable reduction in the updated values¹³, opting to maintain consistency with the parameters used in our previous research⁵.

The energy of neutrons at the BSA outlet was divided into 200 bins, with a lower cutoff energy set at 1.0 × 10^–10 MeV. The number of particle histories is 1.2 × 10⁹ to achieve a maximum statistical error of less than 1%. The neutron and photon doses were calculated using flux-to-dose conversion factors based on KERMA values. The KERMA values for neutrons were obtained from the ICRU-44 report¹⁴, while those for photons were derived from the relationship between photon energy and mass energy absorption coefficient provided in the NISTIR-5632 report¹⁵.

Nuclear cross-section libraries

This study requires two types of nuclear data libraries to obtain the final neutron parameters at the BSA exit. The first type is the proton library used to calculate the neutron yields from the proton bombardment of a lithium target. The second type is the neutron library, which simulates the interactions of the neutrons generated from the target with the moderator, filter, collimator, and reflector within the BSA. We selected four proton libraries and six neutron libraries as the data sources for this study to comprehensively assess the impact of different data libraries.

Proton libraries

The lithium target consists primarily of the isotopes ⁶Li and ⁷Li, with ⁷Li playing the dominant role in neutron production through nuclear reactions with protons in this study, for the threshold energy of the ⁶Li(p, n)⁶Be reaction is 6.0 MeV. We collected ⁷Li cross-section data from several commonly used proton nuclear libraries and analyzed their sources and energy ranges, as summarized in Table 1. The ⁷Li cross-section data in these proton libraries originate from three institutions: LANL, JAEA, and Lawrence Livermore National Laboratory (LLNL). The nuclear libraries ENDF/B VII.0, ENDF/B VII.1, JEFF-3.3, CP2020, and TENDL-2023 adopted the ⁷Li(p, n)⁷Be reaction cross-section data from LANL. Among these, all except CP2020 share the same cross-section data. In the ENDF/B VIII.0 proton library released by the National Nuclear Data Center (NNDC) in 2018, David Brown from Brookhaven National Laboratory and Petr Navratil from Tri-University Meson Facility (TRIUMF) re-evaluated the reaction cross sections for the ⁷Li(p, n)⁷Be reaction⁶. The energy range of incident protons has been significantly expanded from 1.881 to 10.0 MeV in ENDF/B VII.0 and VII.1 to 1.88–20.0 MeV. The data for this reaction cross-section has been retained in the ENDF/B VIII.1 release published in 2024. The same reaction cross-sections released by JAEA cover an energy range of 1.88–200.0 MeV, both in JENDL-4.0/HE and JENDL-5.0, making it suitable for simulations involving protons across a broad spectrum of energies. The data library was also adopted by the FENDL-3.2c library¹⁶, which was released by the International Atomic Energy Agency (IAEA) in June 2024. In addition, the LANL was used to release the CP2020 charged particle transport library in 2021, which includes a ⁷Li proton library with an energy range of 1.88-3.0 MeV¹⁷. By comparing and analyzing these libraries, we ultimately selected four libraries-ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0-as the candidate libraries for their different incident proton energy ranges. Since ENDF/B VIII.1 is a newly released library with no corresponding ACE-format ⁶Li and ⁷Li proton cross-section files for Monte Carlo simulations, we have used NJOY99.304 to generate the corresponding ACE cross-section files. Prior to this, we used NJOY99.304 to generate the ENDF/B VII.1 proton library and performed a consistency comparison with the ENDF/B VII.0 proton library¹⁸ released by LANL (as the ⁶Li and ⁷Li cross-section data are identical in the ENDF/B VII.0 and ENDF/B VII.1 proton reaction libraries). The results demonstrated that the lithium target neutron spectra and neutron yield calculated using both proton libraries were completely identical, confirming the accuracy of the ENDF/B VIII.1 proton cross-section library processed in this study.

Table 1.

The energy range and data release information of ⁷Li(p, n)⁷Be reaction in commonly used nuclear cross-section libraries.

Library Name	Released Time	Organization	Data Source	Energy (MeV)
ENDF/B VII.0	Dec 2006	NNDC	LANL	1.881 ~ 10.0
ENDF/B VII.1	Dec 2011	NNDC	LANL	1.881 ~ 10.0
JENDL-4.0/HE	Nov 2015	JAEA	JAEA	1.88 ~ 200.0
JEFF-3.3	Nov 2017	NEA	LANL	1.881 ~ 10.0
ENDF/B VIII.0	Feb 2018	NNDC	LLNL	1.88 ~ 20.0
CP2020	Jul 2021	LANL	LANL	1.88 ~ 3.0
JENDL-5.0	Dec 2021	JAEA	JAEA	1.88 ~ 200.0
TENDL-2023	Dec 2023	IAEA	LANL	1.881 ~ 10.0
FENDL-3.2c	Jun 2024	IAEA	JAEA	1.88 ~ 200.0
ENDF/B VIII.1	Oct 2024	NNDC	LLNL	1.88 ~ 20.0

To compare and analyze the differences among different proton libraries, we used the ACER module of the NJOY program to extract the ⁷Li(p, n)⁷Be reaction cross-sections from four ACE-format proton libraries. Based on the extracted cross-section data, the variation of the neutron production cross-section with incident proton energy was plotted, as presented in Fig. 2. Starting from ENDF/B VIII.0, the re-evaluated ENDF/B database introduced reaction channel MT 51 to describe the ⁷Li(p, n)⁷Be* reaction⁶, where ⁷Be* represents the excited state of ⁷Be. In contrast, other cross-section libraries only considered the ⁷Li(p, n)⁷Be reaction (MT 50), which produces ground-state ⁷Be. The excited state ⁷Be* decays to the ground state by emitting a gamma photon. Therefore, when simulating neutron production from a 2.8 MeV proton beam striking a lithium target using the ENDF/B VIII.1 ACE-format proton library, both the ⁷Li(p, n)⁷Be and ⁷Li(p, n)⁷Be* reactions are taken into account. Notably, the threshold energy of the ⁷Li(p, n)⁷Be* reaction is 0.4291 MeV higher than that of the ⁷Li(p, n)⁷Be reaction, but they share the same neutron energy-angular distribution¹⁹. Figure 2(a) demonstrates the variation of neutron production cross-section of the ⁷Li(p, n)⁷Be and ⁷Li(p, n)⁷Be* reaction across the energy range of 1.5–10 MeV for all four cross-section libraries, while Fig. 2(b) presents a zoomed-in view of the energy range from 1.8 to 3.0 MeV. These two graphs indicate significant differences in the ⁷Li neutron production cross-sections corresponding to the four cross-section libraries. The four libraries show distinct resonance peaks around 2.3 MeV of the incident proton energy, highlighting the impact of different evaluation methods on nuclear reaction cross-sections.

Fig. 2.

The cross sections of ⁷Li(p, n)⁷Be and ⁷Li(p, n)⁷Be* reactions from ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0 proton reaction libraries with the proton incident energy in the range of 1.5–10.0 MeV (a) and 1.8-3.0 MeV (b). The labels “gs” and “ex” in parentheses after “ENDF/B VIII.1” represent the ground state and the excited state, respectively.

Neutron libraries

Six neutron libraries were selected to comprehensively evaluate the impact of different libraries on the BSA simulation results, representing diverse regions and released periods: ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2²⁰, and JENDL-5.0. These libraries are widely used in various fields and are currently the most mature and well-established libraries.

The study simultaneously employed ENDF/B VII.1, ENDF/B VIII.0, and ENDF/B VIII.1 libraries for the following reasons. Initially, the ENDF/B series libraries are universally acknowledged as the most dependable and accurate nuclear data libraries. Second, ENDF/B VIII.0 and ENDF/B VIII.1 incorporate significant updates to the nuclide data relevant to this study compared to ENDF/B VII.1. A comparative analysis of these libraries enables a more thorough evaluation of the impact of different nuclides on neutron parameters. In addition to the ENDF/B series, JEFF-3.3, CENDL-3.2, and JENDL-5.0-developed by European, Chinese, and Japanese research teams, respectively-were included to expand the range of data sources and mitigate potential biases arising from reliance on a single library. The NJOY99.304 program was also employed to generate the necessary ACE-format neutron libraries for the ENDF/B VIII.1 library. However, since NJOY99.304 cannot recognize the new ENDF cross-section format for the isotopes of Cu, O, Si, and Pb in the ENDF/B VIII.1 neutron library, NJOY2016.65 was used to process the ACE-format cross-section files for these nuclides. The reason for not using NJOY2016.65 to process all nuclides in the ENDF/B VIII.1 neutron library is the significant discrepancy observed in the energy range of 1.0 × 10⁻³ to 1.0 × 10⁻¹ MeV. This discrepancy emerged when comparing the neutron flux calculated using NJOY2016.65-processed ENDF/B VIII.0 and ENDF/B VII.1 neutron cross-section libraries with that obtained from the LANL-provided ACE-format neutron cross-section library²¹ in the BSA simulation of this study. In contrast, results obtained using the NJOY99.304-processed cross-section library were more consistent with those from the LANL-provided ACE-format neutron cross-section library, exhibiting a deviation of less than 0.5% in all neutron beam parameters during the same simulation. Although NJOY2016.65 was used to process certain nuclides in the ENDF/B VIII.1 neutron library, their impact on the final results is minimal, as will be explained in the following sections of this paper.

Results and discussion

The neutron energy spectra at the lithium target

Figure 3 indicates the neutron energy spectra produced by proton beam bombardment of the lithium target, calculated using four different libraries. Specific neutron yields are presented in Table 2. As depicted in Fig. 3, the neutron energy spectrum, calculated using the ENDF/B VIII.1 library (green solid line), exhibits a distinct single-peak structure (approximately 0.32 MeV). In contrast, the neutron energy spectra obtained using the ENDF/B VII.1, CP2020, and JENDL-5.0 libraries display a characteristic double-peak structure (approximately 0.1 and 0.45 MeV, respectively), demonstrating a clear divergence between the two types of spectral profiles.

Fig. 3.

The neutron energy spectra of a 2.8 MeV proton beam bombed a 140 μm thick natural lithium target simulated by PHITS with four nuclear libraries.

Table 2.

Total neutron yields of lithium target calculated based on four nuclear cross-section libraries.

Parameters	ENDF/B VII.1	ENDF/B VIII.1	CP2020	JENDL-5.0
Total neutron yields (n·mC−1)	1.3051E + 12	1.2684E + 12	1.3872E + 12	1.1871E + 12
Relative deviation	9.94%	6.85%	16.86%	-

Figure 4 presents the neutron flux distribution near the metal target. It is evident that the neutron flux calculated using the ENDF/B VIII.1 library in Fig. 4(b) exhibits a distinct inward shrinkage compared to the flux calculated with the other three databases, particularly with a significant shift in the position of the 5.6 × 10⁸ cm⁻²·s⁻¹ dose contour as indicated by the bold black arrow.

Fig. 4.

The neutron flux distribution of a 2.8 MeV proton beam bombed a 140 μm thick natural lithium metal target simulated by PHITS based on four nuclear libraries: (a) ENDF/B VII.1, (b) ENDF/B VIII.1, (c) CP2020, and (d) JENDL-5.0.

Since PHITS was used as the simulation tool in this study, the results calculated with the proton library from JENDL-5.0 were used as reference values and compared with those obtained using the other three libraries. As depicted in Table 2, the total neutron yield calculated using the JENDL-5.0 proton library indicates the most significant relative deviation of 16.86% compared to the result based on CP2020. In contrast, the slightest relative deviation of 6.85% is observed when compared with the results based on ENDF/B VIII.1, corresponding to the similarity in cross-section trends revealed in Fig. 2. The relative deviation with the results calculated using the ENDF/B VII.1 library is 9.94%. Overall, significant differences in neutron yields are observed depending on the choice of the proton library.

Neutron parameters at BSA outlet

Using the PHITS program, neutron energy spectra from four proton libraries-ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0–were combined with six neutron libraries-ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0. These combinations were employed to simulate the neutron interactions with various BSA structural components in Fig. 1. The simulations ultimately provided neutron beam parameters at the BSA collimator outlet.

Five neutron beam parameters obtained from all combinations of four proton libraries and six neutron libraries were compared with the IAEA recommended values to evaluate the overall agreement of them. The relative deviations were calculated using the formula: (calculated value-recommended value)/recommended value×100%. The results are presented in bar charts in Fig. 5. From the results indicated in Fig. 5(a), epithermal beam intensity calculated using all library combinations significantly exceeds the IAEA recommended value of 1.0 × 10⁹ n·cm⁻²·s⁻¹, with excess values ranging from 43.18% (ENDF/B VIII.0 + ENDF/B VIII.0) to 84.78% (CP2020 + ENDF/B VIII.1). Notably, the group computed using the CP2020 proton library indicates the most significant deviation, with an average excess of 73.38%. This is followed by the group based on the ENDF/B VII.1 proton library, which exceeds the reference value by an average of 63.13%. Combinations using the JENDL-5.0 and ENDF/B VIII.1 proton libraries exhibit relatively lower deviations, with an excess of 48.45% and 46.25%, respectively. Additionally, the bar charts for each group in Fig. 5(a) reveal that within each dataset, calculated with the ENDF/B VIII.1 neutron library is consistently the highest, exceeding the reference value by approximately a range of percentage values between 11% and 14% more than the group’s average for other combinations. The IAEA-TECDOC-1223 report recommends a minimum value of 0.7 for the ratio between the total neutron current and the total neutron flux . As depicted in Fig. 5(b), the calculated values for all combinations are slightly higher than the recommended value, with the percentage excess remains below 0.5%, except for CP2020 + ENDF/B VIII.1 and ENDF/B VII.1 + ENDF/B VIII.1, where the former is slightly lower than the reference value of 0.7, while the latter is exactly 0.7. The recommended value for the thermal-to-epithermal flux ratio is ≤ 0.05. As indicated in Fig. 5(c), all calculated results for different combinations meet this criterion, with values ranging from 18.70–24.52% below the recommended threshold. Figure 5(d) reveals the bar chart of relative deviations for the fast neutron component and gamma-ray component , compared to the recommended values (both ≤ 2.0 × 10⁻¹³ Gy·cm²). The fast neutron component is marked with the letter ‘n’ following the neutron library name in the legend, while the gamma-ray component is marked with the Greek letter ‘’. From Fig. 5(d), it can be observed that for the fast neutron component, the results based on the ENDF/B VII.1, CP2020, and JENDL-5.0 proton libraries-except when paired with the ENDF/B VIII.1 or JENDL-5.0 neutron libraries-are over 10% higher than the recommended value. When paired with the ENDF/B VIII.1 neutron library, all four proton libraries produce significantly higher fast neutron components than other combinations within the same group, exceeding the IAEA recommended value by 25.69–42.57%. In contrast, the cases using the JENDL-5.0 neutron library are all more than 10% lower than the recommended value, with the calculation based on the ENDF/B VIII.1 + JENDL-5.0 combination even being 21.86% lower than the recommended value. Additionally, the fast neutron component , calculated using the ENDF/B VIII.1 proton library paired with the other four neutron libraries, is less than 2% higher than the recommended value. For the gamma-ray component, all calculation results are below the recommended value of 2.0 × 10⁻¹³ Gy·cm². Among them, the three ENDF/B series libraries indicate a more significant deviation, which is over 20% below the recommended value. However, the results from the other three neutron libraries deviate from the recommended value by less than 20%.

Fig. 5.

The relative deviations of five neutron beam parameters at the BSA collimator exit compared with IAEA recommended values, based on neutron energy spectra obtained using PHITS with proton libraries (ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0) and neutron libraries (ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0). (a) The epithermal beam intensity (b) the ratio between the total neutron current and the total neutron flux, (c) the thermal-to-epithermal flux ratio, and (d) the fast neutron component and the gamma-ray component.

Figure 6 presents the neutron beam energy spectra at the collimator outlet, calculated using various proton and neutron library combinations. These combinations include ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, and JENDL-5.0, where each library’s proton and neutron data were paired. Additionally, the CP2020 proton library paired with the neutron libraries from ENDF/B VII.1 and ENDF/B VIII.1. The inclusion of ENDF/B VIII.0 in the analysis is noteworthy, as its proton library is identical to that of ENDF/B VIII.1 in this study. However, these two versions differ in their neutron libraries, with ENDF/B VIII.1 incorporating updated neutron cross-section data for several isotopes. For this reason, we replaced the target spectrum obtained using the ENDF/B VIII.0 proton library with that from ENDF/B VIII.1 to derive the results for ENDF/B VIII.0 + ENDF/B VIII.0 in Fig. 6; Table 3. Figure 6 also presents the neutron energy spectra calculated using the CP2020 proton library paired with the neutron libraries from ENDF/B VII.1 and ENDF/B VIII.1. The reason for this inclusion is that the neutron flux calculated with the CP2020 proton library paired with ENDF/B VIII.1 neutron library is the highest among all configurations, while the results based on the ENDF/B VII.1 neutron library provides the most consistent with those from the other non-ENDF/B neutron libraries. The results in Fig. 6 indicate significant differences in the neutron beam energy spectra obtained using different libraries at the BSA exit. This disparity is particularly pronounced in the epithermal neutron energy region, where all spectra exhibit a distinct gradient in neutron flux peaks near 1 × 10⁻³ MeV. As depicted in Fig. 6, the neutron energy spectrum derived from ENDF/B VIII.1 deviate considerably from those of ENDF/B VIII.0 but are more similar to the results from ENDF/B VII.1.

Fig. 6.

Neutron spectra at BSA exit simulated by PHITS based on the neutron libaries ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1 and JENDL-5.0 with the target neutron energy spectra calculated by ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, CP2020 and JENDL-5.0, respectively.

Table 3.

Five neutron beam parameters at the BSA collimator exit were calculated using specific combinations of proton and neutron libraries.

Proton/neutron libraries		(n·cm−2·s−1)			(Gy·cm2)	(Gy·cm2)
ENDF/B VII.1	ENDF/B VII.1	1.6011E + 09	0.7012	0.03879	2.2678E-13	1.4962E-13
ENDF/B VIII.0	ENDF/B VIII.0	1.4318E + 09	0.7019	0.03907	2.0352E-13	1.5099E-13
ENDF/B VIII.1	ENDF/B VIII.1	1.5538E + 09	0.7003	0.03889	2.5138E-13	1.5337E-13
CP2020	ENDF/B VII.1	1.7023E + 09	0.7011	0.03864	2.2731E-13	1.4851E-13
	ENDF/B VIII.1	1.8478E + 09	0.6998	0.03774	2.8513E-13	1.5085E-13
JENDL-5.0	JENDL-5.0	1.4959E + 09	0.7032	0.04003	1.7279E-13	1.6322E-13
Recommend values		≥ 1.0E + 09	≥ 0.7	≤ 0.05	≤ 2.0E-13	≤ 2.0E-13

The five neutron beam parameters calculated for these combinations are summarized in Table 3. The results calculated based on these combinations align with the recommended values in the IAEA report, with only being slightly higher than the recommended value. However, we can also observe that, except for and , which exhibit relatively small intergroup deviations; the other three parameters reveal significant intergroup deviations. Among these parameters, shows a relative deviation of 29.05% between the minimum value (ENDF/B VIII.0 + ENDF/B VIII.0) and the maximum value (CP2020 + ENDF/B VIII.1). Similarly, exhibits a substantial relative deviation of 65.02% between the minimum value (JENDL-5.0 + JENDL-5.0) and the maximum value (CP2020 + ENDF/B VIII.1). Additionally, depicts a relative deviation of 9.91% between the minimum value (CP2020 + ENDF/B VII.1) and the maximum value (JENDL-5.0 + JENDL-5.0). Among these combinations, and calculated using the CP2020 + ENDF/B VIII.1 pairing are the highest due to the highest neutron yield obtained based on the CP2020 library. Conversely, the values of these two parameters calculated using the JENDL-5.0 + JENDL-5.0 pairing are significantly lower, with being slightly higher than that of the ENDF/B VIII.0 + ENDF/B VIII.0 combination. This result aligns well with the low ⁷Li(p, n)⁷Be reaction cross-section of JENDL-5.0 indicated in Fig. 1 and low neutron yields in Table 2.

To emphasize the differences in results obtained by using the same proton library paired with different neutron libraries, this research was used to select the calculations listed in Table 3 as the reference values for each proton library: ENDF/B VII.1 + ENDF/B VII.1, ENDF/B VIII.1 + ENDF/B VIII.1, CP2020 + ENDF/B VII.1, and JENDL-5.0 + JENDL-5.0. These were chosen as the reference for neutron spectra generated using the ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0 proton libraries, respectively, when applied to other BSA simulations. The relative deviations between these reference values and other simulation results were calculated and are presented in Table 4. The data in Table 4 organize all calculation results into four groups based on combining each proton library with five different neutron libraries. The results in Table 4 reveal several notable trends. Initially, calculations using the ENDF/B VIII.1 neutron library consistently yielded higher than other neutron libraries. When combined with the ENDF/B VIII.1 proton library, the calculated values exceed those from other neutron libraries by 5.51%−7.85%. When paired with the other three proton libraries, derived from ENDF/B VIII.1 surpasses the reference neutron libraries by 5.63% (JENDL-5.0 + ENDF/B VIII.1), 8.61% (ENDF/B VII.1 + ENDF/B VIII.1), and 8.55% (CP2020 + ENDF/B VIII.1). This trend is also evident in the epithermal neutron flux curves revealed in Fig. 7, where the flux peak for ENDF/B VIII.1 (gray solid line) is significantly higher. Secondly, the value calculated using the JENDL-5.0 neutron library is notably lower than those obtained with other neutron libraries, whereas the value derived from the ENDF/B VIII.1 neutron library is significantly higher. When the JENDL-5.0 neutron library is paired with the ENDF/B VII.1, ENDF/B VIII.1, or CP2020 proton libraries, represents 23.53%, 37.83%, and 23.42% lower than their respective group reference values. When the proton library is JENDL-5.0, is 28.57–30.64% lower compared to other neutron library combinations, except when paired with ENDF/B VIII.1. When the JENDL-5.0 proton library paired with the ENDF/B VIII.1 neutron library, the obtained from the JENDL-5.0 + JENDL-5.0 combination is 63.94% lower than that of the former. This trend is also reflected in Fig. 8, where the fast neutron flux calculated using the JENDL-5.0 neutron library (purple solid line) is lower than in other cases, whereas the fast neutron flux calculated using the ENDF/B VIII.1 neutron library (grey solid line) is significantly higher than in the other combinations. Lastly, JEFF-3.3, CENDL-3.2, and JENDL-5.0 neutron libraries produce slightly larger than the ENDF/B series neutron libraries, with minor variations depending on the selected primary neutron source. For instance, when the primary neutron source is based on the CP2020 library, calculations with CENDL-3.2 and JENDL-5.0 neutron libraries yield gamma-ray components that are 10.08% and 8.93% higher than those obtained with the ENDF/B VII.1 neutron library. Similarly, the JEFF-3.3 neutron library results in a gamma-ray component 7.74% higher than ENDF/B VII.1. Apart from the differences mentioned above, the neutron beam parameters calculated using different neutron libraries based on the same target neutron energy spectrum are generally consistent, with overall deviations not exceeding 5%.

Table 4.

Relative deviations of five neutron beam parameters at the BSA collimator exit were calculated using different combinations of proton and neutron libraries compared to the reference values in Table 3.

Proton/neutron libraries		(%)	(%)	(%)	(%)	(%)
ENDF/B VII.1	ENDF/B VIII.0	−0.34	0.00	−0.46	0.35	−1.28
	ENDF/B VIII.1	8.61	−0.17	−2.37	25.68	0.99
	JEFF-3.3	0.38	0.01	1.08	−0.97	7.38
	CENDL-3.2	−0.16	0.00	−1.01	0.55	9.40
	JENDL-5.0	2.83	0.23	2.68	−23.53	8.80
ENDF/B VIII.1	ENDF/B VII.1	−7.38	0.20	0.54	−19.26	−0.85
	ENDF/B VIII.0	−7.85	0.23	0.46	−19.04	−1.55
	JEFF-3.3	−7.02	0.14	1.83	−20.32	6.92
	CENDL-3.2	−7.51	0.09	1.04	−18.86	7.79
	JENDL-5.0	−5.51	0.44	4.53	−37.83	7.92
CP2020	ENDF/B VIII.0	−0.35	0.00	−0.65	0.18	−0.69
	ENDF/B VIII.1	8.55	−0.19	−2.33	25.44	1.58
	JEFF-3.3	0.36	0.00	1.60	−1.13	7.74
	CENDL-3.2	−0.26	0.06	−0.67	0.50	10.08
	JENDL-5.0	2.80	0.20	2.92	−23.42	8.93
JENDL-5.0	ENDF/B VII.1	−2.51	−0.28	−3.17	29.87	−6.79
	ENDF/B VIII.0	−2.87	−0.26	−3.30	30.24	−8.83
	ENDF/B VIII.1	5.63	−0.44	−4.90	63.94	−7.88
	JEFF-3.3	−2.15	−0.27	−2.55	28.57	−1.28
	CENDL-3.2	−2.66	−0.33	−2.52	30.64	1.10

Fig. 7.

Epithermal neutron flux (0.5 eV–10 keV) at the BSA collimator exit calculated using PHITS with neutron energy spectra obtained from proton libraries: (a) ENDF/B VII.1, (b) ENDF/B VIII.1, (c) CP2020, and (d) JENDL-5.0. The spectra are paired with neutron libraries: ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0.

Fig. 8.

Fast neutron flux (10 keV–1.0 MeV) at the BSA collimator exit calculated using PHITS with neutron energy spectra obtained from proton libraries: (a) ENDF/B VII.1, (b) ENDF/B VIII.1, (c) CP2020, and (d) JENDL-5.0. The spectra are paired with neutron libraries: ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0.

Analysis of the discrepancies

Causes of the discrepancies of primary neutron sources

For the proton libraries ENDF/B VII.1, CP2020, and JENDL-5.0, the differences in the obtained target spectra primarily originate from variations in the nuclear experimental data and evaluation methods used for ⁷Li. Since the fundamental behavior of a given nuclear reaction should not exhibit intrinsic differences due to variations in measurement techniques, the resulting neutron spectra all exhibit a similar double-peak structure. However, the neutron spectrum of the lithium target calculated using the ENDF/B VIII.1 proton library is single-peak, presenting a markedly different profile compared to the other three spectra.

As reported by a previous study⁶, for the ⁷Li(p, n)⁷Be reaction in ENDF/B VIII.0, both the ECPL and ENDF/B-VII.1 cross-sections were excluded, and a new evaluation was developed using spline-interpolated data. The ENDF/B VIII.1 library inherits the cross-section data for the ⁷Li(p, n)⁷Be reaction from the ENDF/B VIII.0 library. According to the information obtained through communication with the library developers, the cross-section for the ⁷Li(p, n)⁷Be reaction may have limitations in its evaluation method due to a lack of data for the outgoing particle distributions. After analyzing the original ENDF-format cross-section file for ⁷Li in the ENDF/B VIII.1 proton library, we identified two major changes compared to the ENDF/B VII.1 version:

(1) ENDF/B VIII.1 accounts for both ground-state and excited-state ⁷Be production in the ⁷Li(p, n)⁷Be reaction, incorporating reaction channels MT 50 and MT 51, whereas ENDF/B VII.1 only considers ground-state ⁷Be production (MT 50);

(2) ENDF/B VIII.1 re-evaluates the neutron energy-angle distribution for the ⁷Li(p, n)⁷Be reaction, assuming identical distributions for both ground-state and excited-state ⁷Be production. However, we found it lacks neutron energy-angle distribution data for ground-state ⁷Be in the energy range of 1.880–2.371 MeV.

To investigate the causes of the single-peak neutron energy spectrum obtained using the ENDF/B VIII.1 proton library, we analyzed the two key changes in the ENDF/B VIII.1 ⁷Li proton cross-section file. Specifically, we supplemented the missing neutron energy-angle distribution data in the 1.880–2.371 MeV energy range and replaced all neutron energy-angle distribution data in the ENDF/B VIII.1 ⁷Li proton cross-section file with the corresponding data from the ENDF/B VII.1 proton library. We then regenerated the ACE-format cross-section files using the NJOY code and conducted simulations with the PHITS program. The resulting neutron energy spectra are shown in Fig. 9. It can be seen that after adding the neutron energy-angle distribution data for the 1.880–2.371 MeV energy range, the neutron energy spectrum calculated using the ENDF/B VIII.1 ⁷Li proton cross-section file changed to a double-peak structure (red solid line labeled ‘R1’). However, the spectrum above 0.6 MeV still retains its original characteristics, showing noticeable differences compared to the results obtained using the ENDF/B VII.1 proton cross-section file. When all neutron energy-angle distribution data were replaced, the resulting neutron energy spectrum nearly overlapped with that obtained using the ENDF/B VII.1 proton library (blue solid line labeled ‘R2’). Therefore, we can conclude that the primary reason for the significant differences between the neutron energy spectrum calculated using the ENDF/B VIII.1 proton cross-section file and those based on the CP2020, ENDF/B VII.1, and JENDL-5.0 databases is the re-evaluation of the neutron energy-angle distribution data for the ⁷Li(p, n)⁷Be reaction in the ENDF/B VIII.1 library, especially the omission of data in the 1.880–2.371 MeV energy range, while the inclusion of the excited-state reaction channel has little to no effect.

Fig. 9.

Comparison of the neutron energy spectrum after supplementing the neutron energy-angle distribution data for the 1.880–2.371 MeV energy range (red solid line) and replacing all neutron energy-angle distribution data for this reaction in the ENDF/B VIII.1 ⁷Li proton cross-section file with the corresponding data from the ENDF/B VII.1 database (blue solid line), alongside the original calculated results and those based on the ENDF/B VII.1 proton cross-section file.

Factors contributing to the variations in BSA exit parameters

To analyze the variations presented in Table 4, this study separately examined the ENDF/B VIII.1, JENDL-5.0, JEFF-3.3, and CENDL-3.2 neutron libraries. Based on the elements that constitute the BSA and their respective volumes shown in Fig. 1, four elements-Mg, F, Pb, and Li-were selected for detailed investigation. Mg and F constitute the moderator, while Li and F form the thermal neutron filter. Additionally, Li is also a component of the collimator, whereas Pb serves as both the gamma-ray filter and the encapsulating material for the entire BSA. For the analysis, the primary neutron sources in all cases were obtained from the CP2020 proton library. Since the ENDF/B VIII.1 neutron library is an updated version based on the ENDF/B VIII.0 neutron library, which in turn originates from the ENDF/B VII.1 neutron library, this study replaced the cross-section data of the four elements with those from the ENDF/B VII.1 neutron library when analyzing the impact of ENDF/B VIII.1 on the neutron beam parameters of the BSA. Similarly, to ensure a consistent reference when analyzing the effects of the other three neutron libraries, the cross-section data from the ENDF/B VII.1 neutron library were also used to replace the corresponding data in these three libraries.

Within the ENDF/B VIII.1 neutron library, the neutron cross-section data of Mg, F, Pb, and Li were individually replaced with those from the ENDF/B VII.1 library. Subsequently, recalculations were performed using PHITS, and the updated results were compared with the original. These findings are summarized in Table 5; Fig. 10. The significantly higher neutron energy spectrum obtained at the BSA exit, calculated using the ENDF/B VIII.1 neutron library compared to the ENDF/B VII.1 library, is primarily attributed to the updated cross-sections of F. Specifically, substituting the neutron cross-section of F led to a 7.19% decrease in , a 21.10% reduction in , and a 2.17% increase in . By comparing and analyzing the neutron cross-section data of ¹⁹F from ENDF/B VIII.1 and ENDF/B VII.1, we found that although the elastic and inelastic scattering cross-sections of ¹⁹F in ENDF/B VIII.1 have undergone certain changes (as shown in Fig. 11), the observed differences primarily originate from the updated neutron energy-angle distribution for the elastic scattering reaction in the ENDF/B VIII.1 ¹⁹F neutron cross-section file. In addition, the replacement of the neutron cross-section of Pb led to a 2.41% decrease in , while the changes in the other four parameters remain below 1%. Considering that the neutron cross-section data for Pb was generated using NJOY2016.65, some degree of inaccuracy is expected. However, as shown in Table 5; Fig. 10, these effects do not compromise the final conclusions of this study, as the impact of replacing Pb’s cross-section on the five neutron parameters is minimal. The decrease in following the replacement of Pb is primarily due to changes in the photon production cross-section of Pb in the ENDF/B VIII.1 neutron library. Compared to F and Pb, replacing the data for Mg and Li had virtually no impact. By simultaneously replacing the cross-section data for both F and Pb elements, the changes in the five neutron parameters observed are almost identical to those obtained when using the ENDF/B VII.1 neutron library. Therefore, it can be concluded that the differences in the results calculated using the ENDF/B VIII.1 neutron library, compared to other libraries, are primarily due to the updates in the neutron cross-sections for F and Pb elements.

Table 5.

Relative deviations of neutron beam parameters at the BSA collimator exit compared to original ones after replacing mg, F, pb, and Li neutron reaction cross-sections in ENDF/B VIII.1 library with corresponding data from ENDF/B VII.1 library.

Neutron libraries		(%)	(%)	(%)	(%)	(%)
ENDF/B VIII.1	ENDF/B VII.1(F)	−7.19	0.23	2.17	−21.10	0.81
	ENDF/B VII.1(Mg)	−0.01	0.00	0.13	0.01	0.03
	ENDF/B VII.1(Pb)	−0.80	−0.03	0.08	0.79	−2.41
	ENDF/B VII.1(Li)	−0.01	0.00	0.03	0.01	0.01
	ENDF/B VII.1(F + Mg)	−7.89	0.19	1.93	−20.36	−1.82
ENDF/B VII.1		−7.87	0.19	2.38	−20.28	−1.55

Fig. 10.

The neutron spectra (0.5 eV-1.0 MeV) at the BSA collimator exit after replacing F and Pb neutron cross-sections in ENDF/B VIII.1 library with corresponding data from ENDF/B VII.1 library.

Fig. 11.

Comparison of elastic (a) and inelastic (b) scattering cross-sections of ¹⁹F in ENDF/B VII.1, ENDF/B VIII.1, and JENDL-5 neutron libraries.

Analysis of the JENDL-5.0 neutron library followed a similar approach, where the cross-section data of Mg, F, Pb, and Li were replaced with the corresponding ENDF/B VII.1 values. As presented in Table 6, substituting the neutron cross-section of ¹⁹F led to a 2.61%, 2.82%, and 5.38% reduction in, , and, respectively. After replacing the ¹⁹F cross-section data, the obtained fast neutron component was nearly identical to the value calculated entirely using the ENDF/B VII.1 neutron library, with relative deviations of 30.63% and 30.59% compared to JENDL-5.0. Therefore, the significantly lower fast neutron component derived from the JENDL-5.0 neutron library, compared to other databases, is also primarily attributed to the cross-section data of ¹⁹F. This trend is clearly depicted in Fig. 12, which compares neutron energy spectra before and after the replacement of the neutron cross-section of ¹⁹F. As shown in Fig. 11, the elastic scattering cross-section of ¹⁹F in the JENDL-5.0 neutron library is significantly lower in the fast neutron region compared to those in ENDF/B VII.1 and VIII.1. Additionally, we also found the cross-section curves exported using the VIEWR module of the NJOY code reveal that the neutron energy-angle distribution for the elastic scattering reaction in JENDL-5.0 differs markedly from those in the other two libraries. Therefore, we hypothesize that these factors contribute to the lower fast neutron component observed in calculations based on JENDL-5.0. Meanwhile, the gamma-ray component was slightly higher than those in other neutron libraries, which is attributed to the combined effects of the cross-sections of F and Pb elements, as illustrated in Table 6. When both elements were replaced simultaneously, the recalculated parameters closely align with the ENDF/B VII.1 library results, with all deviations falling below 0.5%. Therefore, the differences in the five neutron parameters calculated using the JENDL-5.0 neutron library, compared to other libraries, are also primarily attributed to the effects of F and Pb elements.

Table 6.

Relative deviations of five neutron beam parameters at the BSA collimator exit compared to original ones after replacing mg, F, pb, and Li neutron reaction cross-sections in JENDL-5.0 library with corresponding data from ENDF/B VII.1 library.

Neutron libraries		(%)	(%)	(%)	(%)	(%)
JENDL-5.0	ENDF/B VII.1(F)	−2.61	−0.21	−2.82	30.63	−5.38
	ENDF/B VII.1(Mg)	−0.05	0.06	−0.03	−0.05	0.45
	ENDF/B VII.1(Pb)	−0.21	0.00	0.20	0.31	−3.48
	ENDF/B VII.1(Li)	0.08	−0.03	0.33	−0.11	−0.06
	ENDF/B VII.1(F + Pb)	−2.89	−0.26	−2.79	30.88	−8.45
ENDF/B VII.1		−2.72	−0.20	−2.84	30.59	−8.20

Fig. 12.

The neutron spectra (0.5 eV-1.0 MeV) at the BSA collimator exit after replacing F and Pb neutron cross-sections in JENDL-5.0 library with corresponding data from ENDF/B VII.1 library.

For JEFF-3.3 and CENDL-3.2 neutron libraries, distinct variations were found exclusively in the gamma-ray components compared to ENDF/B VII.1 data. To explore these discrepancies, cross-section data for four specific elements in both libraries were individually replaced, and recalculated outcomes are provided in Tables 7 and 8. In JEFF-3.3 library, Pb emerged as a significant factor, with its replacement resulting in a gamma-ray deviation of −3.64%, while the other three elements exhibited negligible effects. However, this alone could not explain the total deviation of −7.18% observed when all cross-sections were replaced with ENDF/B VII.1 data. Subsequent analysis identified copper (Cu) as an additional contributor, accounting for a deviation of −2.90%. The gamma-ray component observed with the CENDL-3.2 library presented a more complex case. Individual replacement of Mg and Pb cross-sections led to unexpected increases in the gamma-ray component by 7.99% and 3.57%, respectively, in contrast to the 9.16% reduction achieved with complete replacement using ENDF/B VII.1 data. Further investigation pinpointed hydrogen (H) as the principal contributor to the elevated gamma-ray component, with its replacement leading to a substantial deviation of −17.67%. These results indicate that when selecting nuclear libraries, not only the elements comprising the BSA moderation layer, filtering layer, reflector, and collimator should be considered, but also auxiliary structures, such as the heat dissipation components of the metal target, may also have a certain impact on the final calculation results.

Table 7.

Relative deviations of five neutron beam parameters at the BSA collimator exit compared to original ones after replacing mg, F, pb, li, and Cu neutron reaction cross-sections in JEFF-3.3 library with corresponding data from ENDF/B VII.1 library.

Neutron libraries		(%)	(%)	(%)	(%)	(%)
JEFF3.3	ENDF/B VII.1(F)	0.08	−0.01	−0.51	0.66	0.06
	ENDF/B VII.1(Mg)	0.01	−0.01	−0.76	0.08	−0.39
	ENDF/B VII.1(Pb)	−0.58	0.06	−1.66	0.52	−3.64
	ENDF/B VII.1(Li)	0.00	0.00	0.03	0.00	0.00
	ENDF/B VII.1(Cu)	0.16	−0.04	0.15	−0.08	−2.90
ENDF/B VII.1		−0.36	0.00	−1.58	1.14	−7.18

Table 8.

Relative deviations of five neutron beam parameters at the BSA collimator exit compared to original ones after replacing mg, F, pb, li, and H neutron reaction cross-sections in CENDL-3.2 library with corresponding data from ENDF/B VII.1 library.

Neutron libraries		(%)	(%)	(%)	(%)	(%)
CENDL-3.2	ENDF/B VII.1(F)	0.05	−0.03	−0.05	0.77	−0.25
	ENDF/B VII.1(Mg)	−0.09	−0.06	0.68	−0.74	7.99
	ENDF/B VII.1(Pb)	0.22	−0.04	0.08	−0.12	3.57
	ENDF/B VII.1(Li)	0.06	0.00	0.23	−0.08	0.57
	ENDF/B VII.1(H)	0.03	0.00	−0.26	−0.01	−17.67
ENDF/B VII.1		0.26	−0.06	0.68	−0.49	−9.16

Discussion

This study investigates the impact of proton and neutron libraries on the neutron beam parameters at the BSA outlet of the AB-BNCT device in two stages. Firstly, we evaluated the neutron energy spectra generated by a 2.8 MeV, 20 mA proton beam bombarding a 140 μm-thick, 6 cm-radius metallic lithium target. Four proton libraries-ENDF/B VII.1, ENDF/B VIII.1, CP2020, and JENDL-5.0-were used to calculate the neutron energy spectra. Subsequently, these spectra were combined with six neutron libraries–ENDF/B VII.1, ENDF/B VIII.0, ENDF/B VIII.1, JEFF-3.3, CENDL-3.2, and JENDL-5.0–to determine the neutron beam parameters at the BSA outlet of an AB-BNCT device. The results were thoroughly analyzed to evaluate the impact of different nuclear data libraries on the final outcomes. In the initial step, noticeable variations were observed in the target neutron energy spectra across the four proton libraries. Specifically, the neutron energy spectrum calculated with the ENDF/B VIII.1 proton library exhibited a distinct single-peak structure, whereas the other three libraries generated double-peak spectra. Through our analysis, we found that in the ⁷Li cross-section file of the ENDF/B VIII.1 proton library released by NNDC, the neutron energy-angle distribution corresponding to the MT 50 reaction channel is missing data in the 1.880–2.371 MeV energy range after the update. This is the fundamental reason why the target spectrum obtained using the ENDF/B VIII.1 proton library exhibits a single-peak. Additionally, after attempting to supplement the missing data in this energy range using the neutron energy-angle distribution data of ⁷Li from the ENDF/B VII.1 proton library, we found that although the resulting target spectrum exhibited a double-peak structure, it differed significantly from the spectra obtained from other proton libraries. This deficiency will limit its applicability in relevant calculations. In addition to the aforementioned differences in spectral profiles, the neutron yield differed by 16.86% between CP2020 and JENDL-5.0. This variation further propagated to the calculated neutron beam parameters at the BSA outlet. Specifically, the group using the target spectrum derived from the CP2020 proton library exhibited a significantly higher epithermal neutron flux compared to those using other target spectra, whereas the group using the target spectrum from the JENDL-5.0 proton library yielded a notably lower epithermal neutron flux. It is worth noting that within the same group using the same target spectrum, the neutron energy spectrum and beam parameters at the BSA outlet obtained from the ENDF/B VIII.1 and JENDL-5.0 neutron libraries exhibit significantly different characteristics compared to those from the other four neutron libraries. The epithermal neutron intensity and fast neutron component calculated using the ENDF/B VIII.1 neutron library were 5.51%−8.61% and 19.04%−63.94% higher, respectively, than those obtained with other libraries due to a more prominent neutron flux peak. In contrast, the fast neutron component calculated using the JENDL-5.0 neutron library was 23.42%−63.94% lower compared to other libraries, with the largest deviation observed when compared to results obtained using the JENDL-5.0 proton library and ENDF/B VIII.1 neutron library combination. Moreover, the gamma-ray component derived from the JEFF-3.3, CENDL-3.2, and JENDL-5.0 neutron libraries was 6.79%−10.08% higher than that obtained with the ENDF/B series. Further analysis revealed that the discrepancies were primarily caused by differences in the neutron cross-section data of F, Pb, Mg, Cu, and H in specific neutron libraries. In particular, the differences in the F and Pb cross-sections accounted for the variations in the obtained BSA outlet spectra and neutron parameters when using the ENDF/B VIII.1 and JENDL-5.0 neutron libraries. The elevated gamma-ray component observed in the JEFF-3.3 and CENDL-3.2 neutron libraries, compared to the ENDF/B series, was largely due to the different cross-sections of Mg, Cu, and H.

This study has certain limitations. When generating the ACE-format neutron library for ENDF/B VIII.1, we encountered difficulties processing some nuclides using NJOY99.304. As a result, we used NJOY2016.65 to process these nuclides instead. However, a comparison between the ACE-format neutron cross-section data processed by NJOY2016.65 and the reference cross-section data provided by LANL revealed significant discrepancies in the calculated results for the cases analyzed in this study. This was the primary reason why NJOY2016.65 was not used to process all nuclides in our research. Nevertheless, when assessing the impact of Pb processed using NJOY2016.65, we found that Pb had a negligible effect on the final results, indicating that its influence does not compromise the conclusions of this study. Furthermore, in our analysis of the influence of specific nuclides on the discrepancies among different neutron libraries, we faced challenges due to the presence of multiple isotopes for certain elements. Given the large volume of data, we did not conduct a detailed analysis of each individual isotope. Consequently, although we identified the impact of Cu, Mg, and Pb, we were unable to precisely quantify the contributions of their respective isotopes.

Conclusion

This study demonstrates that the choice of proton and neutron libraries can significantly influence Monte Carlo simulation results for the AB-BNCT system. Differences were observed in the target spectra produced by the ENDF/B-VII.1, CP2020, and JENDL-5.0 proton libraries, which in turn affected the neutron beam parameters at the BSA outlet. The ENDF/B-VIII.0 and VIII.1 libraries provided less detailed data for the ⁷Li(p, n) reaction, which may limit their applicability in simulations involving metallic lithium targets. For neutron libraries, most versions—ENDF/B-VII.1, ENDF/B-VIII.0, JEFF-3.3, and CENDL-3.2—yielded consistent results. In contrast, ENDF/B-VIII.1 and JENDL-5.0 showed notable spectral deviations, the causes of which remain to be fully understood. As JENDL has been widely applied and validated in BNCT research, further simulations reflecting real device configurations, combined with experimental benchmarking, are essential to support future data library selection.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Hong-Bing Song, Jing-Gang Xu, Xiong Yang, Zhi-Feng Li, Sheng Wang, Chang-Li Ruan, Xiang-Pan Li. The first draft of the manuscript was written by Hong-Bing Song and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Key R&D Program of China (Grant No. 2023YFA1008600) and the National Nature Science Foundation of China (Grant No. 12405393 and 12305340).

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Conflict of interest

The authors declare that they have no competing interests.