Ref.: EURISOL DS/Task5/TN-06-14

High-energy neutron attenuation in iron and concrete Verification of FLUKA Monte Carlo code by comparison with HIMAC experimental results Daniela ENE NIPNE, Bucharest, Romania Abstract Neutron energy spectra penetrated through iron and concrete shields were calculated with FLUKA Monte Carlo code to investigate the capability of the program to be applied in the design of the biological shield of the EURISOL facility. Simulations have been carried out for the experimental arrangements supplied in benchmark problem HIMAC to estimate the neutron attenuation in the energy range from 20 to 800 MeV. The RQMD & DPMJET-II external event generator coupled with FLUKA was used to model the problem since the source of the neutron in the calculation was the 400MeV/nucleon 12C ions bombarding a thick Cu target. Neutron fluence profiles and corresponded derived attenuation lengths are also compared with the experimental data. The calculation results are in fairly good agreement with the measurements except some discrepancies found for the activation detectors set-up that can be reasonably explained. It is concluded from this computational work that FLUKA code is a suitable tool to be used in the accurate determination of the high energy radiation field surrounding the powerful drivers of the EURISOL facility and further planning and optimizing of the installation shielding design concept. 1. Motivation A major objective of the Task#5/subtask_A "Safety & Radioprotection" of the EURISOL DS project consists in developing and studying of methods for shielding against prompt radiation and for the containment of the radioactivity suitable to EURISOL facility design [1-2]. The first step in developing this strategy is the code validation by comparison with experimental data. There are two major problems connected with the simulation of the radiation transport produced by high intensity spallation sources through bulk-biological shields: i) Nuclear data and the reaction models of the existing computer codes are not completely evaluated in their accuracy because of the lack of experimental data at high energy range; ii) Sophisticated variance reduction techniques are needed to obtain particle fluxes and energy spectra with good statistics since the radiation flux attenuation spans over several orders of magnitude. From these reasons the qualification of the methods and tools by comparison with benchmark experimental results, especially at deep penetration is strongly necessary. The aim of this work is to investigate the applicability of the Monte Carlo code FLUKA in simulating high energy neutron deep penetration problem. The reliability of the code has been tested by means the experiment benchmark problem

HIMAC (Heavy-Ion Medical Accelerator in Chiba, Japan) [4]. The neutron energy spectra behind thick iron and concrete shield layers of the HIMAC experimental configurations were calculated with reasonable statistics in the energy range from thermal to 800 MeV and compare with the experiment data to quantify the accuracy of the code estimations. Deeply penetrating high energy neutrons arising from spallation process represents the main shielding problem in the operation of these sources. The secondary protons and photons which are also generated are less important [6] as they can be more easily attenuated through the very thick layers involved in the shields. Even though the neutrons emitted directly from the intra-nuclear cascade represents a small fraction [5] they have a strong forward anisotropy and can reach the incident proton energy. The neutrons created in subsequent evaporation stage and emitted almost isotropically represent a minor shielding concern as they have lower energies and therefore can be attenuated in the most shielding materials. Do to the considerable contribution of the high energy neutrons the maximum of the radiation dose will be in the proton beam direction and therefore the design of the shield for this direction will be of high importance. Accurate calculations are necessary to design an optimal layout of the shielding under ALARA principles reducing consequently the shielding costs that represents the significant part of the total cost of the facility. 2. Geometry and Simulation setup 2.1

Geometry set-up

The test problem supplied neutron energy spectra measured values produced by 400 meV/nucleon C ions at HIMAC. The stopping-length Cu target size was 10 X 10cm and 5cm thick. In the experiment the measurements have been done using two high energy neutron detectors developed in the Japanese laboratory: Self-TOF detector and Ne213 organic liquid scintillator at various depths through thick iron and concrete layer shields were neutrons were penetrated. Figure 1 (a and b) shows the experimental arrangements at the HIMAC beam line using the Self-TOF detector and NE213 detector. Additionally pairs of 209 Bi(n,xn) and 12 C(n,2n)11C activation detectors have been used only for neutrons penetrated through concrete shields, see Figure 1c. The use of the three detector experiment arrangements was necessary due to the detector characteristics: -The Self-TOF detector was used to measure mainly the primary high-energy neutron attenuation through the shields because this detector has low sensitivity to scattered neutrons. -The NE213 organique liquid detector is able to measure both direct and scattered neutrons in the energy range of tens to 800MeV. -Bi and C activation detectors were used to measure neutron fluence distribution in the shield. Gamma ray measurements from the activation detectors have been done with HPGe detectors. Both iron and concrete shield slab has a size of 100 X 100 cm being centered on the beam axis. The thickness of the iron shield assembly was changed to: 20, 40, 60, 80 and 100cm while the neutron spectra penetrating through concrete shield slabs in stacks of 50cm layer have been measured up to 200cm thickness, and respectively 250cm for Bi and C

detectors.

Figure 1: Experimental arrangement at HIMAC of: (a) the Self-TOF detector, (b) an NE213 detector and © the Bi and C activation detectors, respectively.

During the experiments the Self-TOF detector was fixed at the same position (509 cm for iron shields and respectively 506 cm for concrete layers). The iron collimator with an aperture of 10 X 10 cm used in the Self-TOF detector arrangement has the role to shape the neutron almost normally into detector and to avoid the signals induced by the fragment charge particles The NE213 detector was placed in the contact with the shield surface (point A) and far away from the shielding surface (point B) 5m downstream of the copper target. Measuring point B was selected for comparison with the Self-TOF results. The benchmark supplies neutron energy spectra and attenuation profiles through - iron shields of 0 to 100cm thickness [5]; - concrete shields of 0 to 200cm thickness [6]. 2.2 Simulation set-up Attenuation of secondary neutrons produced by 12C heavy ions on a thick (stoppinglength) copper target through iron and concrete shields has been calculated with FLUKA–Monte Carlo interaction and transport code [3]. Simple slab geometry models have been adopted in the simulations to describe the experimental arrangements mentioned above. The RQMD & DPMJET-II external event generator (nucleus-nucleus interaction) coupled with FLUKA. code has been used. Transport of the electromagnetique cascade was switched-off as has been mentioned before photon and proton contributions are a minor concern for the shielding. Neutron were transported down to 19.6MeV the limit below FLUKA uses multigroup cross section library The use of a biasing technique was essential to obtain results with reasonable statistical significance. To apply the Russian roullete and splitting at boundary crossing based on region relative importance the shielding geometry was divided in several layers (of 10 cm thickness). Region-dependent weight window in three energy ranges has been used to minimize weight fluctuations. The densities and atomic compositions of the shield materials i) Iron (Density 7.87 g/cm3), and ii) Concrete (Density 2.27 g/cm3) [Type 02-a, ANL-5800, 660(1963)] were taken from the benchmark problem. The neutron spectra on the front surface of the Self-TOF detector were scored using a surface current estimator “usbrdx”, while for NE213 and activation detectors a track length estimator “usrtrx” has bee used. 3. Results. Discussions 3.1

Comparison of the results for iron shield.

3.1.1

Self-TOF detector arrangement

The neutron energy spectra measured by the Self-TOF detector compared with the results of FLUKA calculations are shown in Figure 2. The Figure presents also the source neutron spectrum at 0 degree. Experimental results are available only for the energy range between 100 to 600MeV, except for 100cm thickness. This can be explained by the insufficient number of neutrons above 600 MeV and respectively by the low efficiency of the Self-TOF detector for neutrons below 100 MeV. As can be seen in the Figure the

Figure 2: Comparison of measured and calculated neutron energy spectra penetrating through iron shields for Self-TOF detector configuration

Figure 3: Comparison of measured and calculated neutron energy spectra penetrating through iron shields for NE-213 detector configuration

shapes of the spectra do not change so much with the shield thickness. The spectra have a broad peak around 200 to 300 MeV with the iron thickness up to 80 cm. The softening of the neutron spectra with increasing of the shielding thickness can not be observed because in this experimental arrangement almost the direct high-energy neutrons penetrated through the shields in the forward direction can be detected. The softening of the spectrum occurs at 100 cm shield layer where the broad peak is not present and the neutron component with energy less than 100MeV can be observed. As can be seen in the Figure the calculated results are in fairly good agreement with the measurements giving harder spectra than the measured results for all cases 3.1.2 NE213 detector arrangement Figure 3 shows comparison between experimental and calculation results for the case of NE213 detector, position A. In this case the neutron energy spectra are available over the whole range (20-800) MeV. The broad peaks around 200 to 300 MeV seen in the Figure 2 have been distributed widely to the lower and higher energy regions. The calculations overestimate the measurements with increasing of the shield thickness in the energy range below 100 MeV and underestimate experimental spectra in the energy range above 400 MeV and are in good agreement with the experiment for (100-400) MeV energy range. One can conclude that in the entire energy range (20-800) MeV the obtained results are in a reasonable good agreement with experiment values.

Table 1: Attenuation profiles of neutron fluences through iron shields for STOF detector configuration x(cm) Experiment FLUKA C/E 0 4.338 6.26 1.44 20 2.329 2.14 0.92 40 0.603 0.627 1.04 60 0.161 0.216 1.35 80 0.045 0.068 1.5 100 0.018 0.027 1.51 Figure 4: Attenuation profiles of neutron fluences through iron shields. Table 2: Attenuation profiles of neutron fluences through iron shields for NE-213 detector configuration x(cm) Experiment FLUKA C/E 20 6.544 5.44 0.83 40 2.084 2.214 1.06 60 0.738 0.845 1.14 80 0.359 0.333 0.93 100 0.122 0.132 1.08

Table 3: Attenuation length of the neutron fluence in iron λ (g cm-2) Configuration Experiment FLUKA Self-TOF NE-213 (A)

123.0 160.5

136.52 168.88

C/E 1.1 1.05

3.3.3

Neutron fluence length

The attenuation of the neutron fluence has been obtained by integrating the neutron flux over the given energy range of both experimental and calculated spectra. Figure 4 and Table 1 and 2 show the comparison of calculated results versus experiment results of the attenuation profiles of neutron fluence integrated from 100 to 600 MeV for Self-TOF detector arrangement and respectively from 20 to 800MeV for NE213 detector arrangement. The calculated neutron fluences are in reasonable agreement in absolute values with the experimental results the discrepancy increasing with the thickness of the shields up to a factor 1.5 for Self-TOF detector arrangement The obtained neutron fluence attenuation lengths of two different energy ranges are summarized in Table 3. Calculated attenuation lengths values have been derived by χ2– square method (MINUIT/PAW fitting subroutine). The comparison of attenuation lengths results between experiment and calculations is in good agreement within 10% although the absolute fluence values show larger discrepancies as can been previously discussed. 3.2

Results for concrete shield

3.2.1 Self-TOF detector arrangement For the NE213 detector arrangement the neutron energy spectra measured by the SelfTOF detector compared with the results of FLUKA calculations are shown in Figure 5. As can been discussed previously the experimental results given by this type of detector are available only for the energy range between 100 to 600 MeV. Similar to the iron penetrating neutron spectra the shape of the curves presented in Figure 5 has a broad peak around 200 to 300 MeV and very slight softening with increasing of the thickness of the shield. As can be seen in the Figure with increasing of the shielding thickness the FLUKA results tend to overestimate the experimental results for energy above 300MeV. At 200 m thickness the difference between experiment and calculation reach one order of magnitude in the energy range of 300 t0 400 MeV. 3.2.2 NE213 detector arrangement Figure 6 shows the comparison between experimental and calculation results for the case of NE213 detector position A. Over a large energy range (20-300) MeV the neutron spectra show a flat shape as in case of the iron shield attenuation. The calculations underestimate experimental spectra in the energy range above 400 MeV and overestimate the measurements in the energy range between 150 to 400MeV. A good agreement with the experiment is given for energy range less than 100 MeV, but large overestimation can bee seen at 200 cm thickness. One can conclude that in the entire energy range (20-800) MeV the obtained results are in a reasonable good agreement with experiment values.

Figure 5: Comparison of measured and calculated neutron energy spectra penetrating through concrete shields for self-TOF detector configuration.

Figure 6: Comparison of measured and calculated neutron energy spectra penetrating through concrete shields for NE213 detector configuration

3.2.3

Neutron fluence length

The same procedure as for iron shield calculations has been used to obtain the neutron fluence and corresponding attenuation lengths values. Figure 7 and Table 4 and 5 show the comparison of calculated results against experiment results of the attenuation profiles of neutron fluence integrated from 100 to 600 MeV for Self-TOF detector arrangement and respectively from 20 to 800MeV for NE213 detector arrangement. The calculated neutron fluences are in reasonable agreement in absolute values with the experimental results having the tendency to overestimate the experiment with increase of the shielding thickness. The big discrepancy of a factor 2.5 for Self-TOF detector arrangement can be explained by the fact that measurements for the 200 cm thickness have been done in the different machine time than for the first 150 cm thickness shield and by the large errors accompanying these results. The comparison of attenuation lengths results between experiment and calculations is in good agreement within 20% see Table 6.

Table 4: Attenuation profiles of neutron fluences through concrete shields for STOF detector configuration C/E X(cm) Experiment FLUKA 50 2.526 2.142 0.85 100 0.655 0.712 1.09 150 0.174 0.302 1.74 200 0.047 0.12 2.50 Table 5: Attenuation profiles of neutron fluences through concrete shields for NE-213 detector configuration X(cm) Experiment FLUKA C/E 50 4.002 5.02 1.26 100 1.84 2.05 1.12 150 0.68 0.83 1.22 200 0.155 0.3 1.97

Figure 7: Attenuation profiles of neutron fluences through concrete shields. Table 6: Attenuation length of the neutron fluence in concrete λ (g cm-2) Configuration Experiment FLUKA Self-TOF NE-213 (A)

86.9 124.4

104.71 121.40

C/E 1.2 0.98

3.2.4 C and Bi activation detector arrangement Neutron energy spectra measured by activation detectors have been obtained by an unfolding procedure based on a calculated spectrum behind 50 cm thick concrete shield. Due to this initial guess spectrum the shape of the experiment curves are similar with the calculation results. As can bee seen in the Figure 7 the measured spectra overestimate with up to 100% at the 50 cm depth in the concrete shield the FLUKA results. This large difference decreases with shield thickness reaching a factor of 20% at 250cm thickness of the shield. This discrepancy can be explained by the charged particles mainly protons, but also pions, deuterons and tritons generated from the graphite by fragmentation reactions which produce in detectors the same radionuclides as neutrons. Attenuation profiles for this experimental configuration are presented Figure 8 and Table 4. The large difference between experiment and calculation reflects the differences found the in energy distributions.

Figure 7: Comparison of measured and calculated neutron energy spectra penetrating through concrete shields for Activation detectors configuration Table 4: Attenuation profiles of neutron fluences through concrete shields for the activation detector configuration C/E x(cm) Exp FLUKA 50 93.908 8.216 0.087 100 23.127 2.772 0.12 150 7.794 1.09 0.14 200 250

2.336 0.63

0.389 0.138

0.166 0.218

Figure 8: Attenuation profiles of neutron fluences through concrete shields for the activation detector configuration

4. Conclusions From the comparison of the calculated results with the HIMAC shielding experiment data resulted that the neutron energy spectra typically agreed within 40 % over a broad energy range, whereas the attenuation profiles agreed with the experiment with the maximum differences reaching a factor of 1.5 at 100 cm iron depth and respectively a factor of 1.97 for 200 cm concrete thickness shield. These results can be considered reasonable for high energy neutron deep-penetration problems. The larger discrepancies found for activation detectors arrangement can be explained by the contribution of the charge particles to the activation products that can not be discriminated in the measurements or estimated theoretically due to the lack of data in this field. Based on these results one can conclude that FLUKA is a suitable code to be used in simulations necessary to design and optimize the biological shield of the high intensity neutron sources as those specific to the EURISOL facility. Acknowledgements We acknowledge the financial support of the EC under the FP6 "Research Infrastructure Action Structuring the European Research Area" EURISOL DS Project; Contract No. 515768 RIDS; www.eurisol.org. The EC is not liable for any use that may be made of the information contained herein.

5. Literature 1. “A Feasibility Study for a European Isotope-Separation-On-Line Radioactive Ion Beam Facility”, (2003), http://www.ganil.fr/eurisol/Final_Report.pdf 2. A. Herreara-Martinez, Y. Kadi, “ EURISOL Multi-MW Target: Preliminary Study of the Liquid Metal Proton to Neutron Convertoer”, EURISOL/DS/Task2/TN-05-01, 2005 3. Fasso, A, Ferrari, A, Ranft, J, Sala P.R. Status and Proscpective for Hadronic Applications, Proceeding ofthe Monte Carlo 2000 Conference, Lisbone October 2000, Springler -Verlag Berlin 4. SINBAD, International DataBase for Integral http://www.nea.fr/html/science/shielding/sinbad/sinbadis.htm

Shielding

Experiment,

5. S Leray et al., Phys. Rev. C 65 (2002) 6. I. Koprivinkar, E. Schachinger, NIM/A 487 571 (2002) 7. M. Sasaki et all., "Measurements of High-Energy Neutrons Penetrated Through Iron Shields Using Self-TOF detector and an NE213 organic liquid scintillator", NIM/B 196, 113-124 (2002) 8. M. Sasaki et all., "Measurements of High-Energy Neutrons Penetrated Through Concrete Shields Using Self-TOF, NE213, and Activation Detectors", NSE 141, 140-153 (2002)