March 8, 2006; CEA Saclay EURISOL DS/Task5/TN-06-04 www.eurisol.org

Benchmark calculations on particle production within the EURISOL DS project B. Rapp, J.C. David, V. Blideanu, D. Doré, D. Ridikas, N. Thiolliere CEA Saclay, DSM/DAPNIA, 91191 Gif-sur-Yvette, France

1. Introduction The validation of the physics models implemented in the MCNPX [1] and FLUKA [2] codes, which will be the main Monte-Carlo tools used for the EURISOL Design Study (DS) [3], is very important in order to ensure the reliability of the results obtained for target optimization, radioprotection and safety, and beam intensity calculations. The beam energy foreseen for the proton driver of EURISOL is about 1 GeV, consequently data around this energy have been selected for the benchmarks. Two major observables were examined, namely neutron and light charged particle production. The neutron production by spallation reaction in different materials has to be well reproduced by simulation codes for both radioprotection purposes and the EURISOL primary target optimization. It is also an important ingredient for beam intensity calculations in the double stage production targets [4]. Scattered primary protons and secondary proton production will contribute to the energy deposition and radiation level around the targets and accelerator structures. Equally, it is important to take into account with a good precision the production of other light charged particles as deuterons, tritons, and helium particles being the important contributors to the gas production and damage rates in the target window or other structure materials. In this paper we report the results obtained from the simulation of double differential cross sections of neutron and light charged particle production for various target materials as a function of variable incident proton energy. The MCNPX (with several physics models) and FLUKA were employed for this purpose.

2. Experimental data Neutron double differential cross sections were measured at the Los Alamos Meson Physics Facility [5] for thin targets of different materials at an incident proton kinetic energy of 800 MeV, for different angles (30°, 60°, 120°, and 150°). All measurements were made using a time of flight technique with neutron detectors composed of plastic scintillators. The neutron energy resolution was better than 5 % which is smaller than the data bin width in all energy range. The statistical errors on the measured double differential cross section are below 5 %, but the systematic errors can reach 20% due the uncertainty on the detector efficiency. The set of data files used was obtained from the EXFOR database [6]. Neutron spectra on thick targets have been measured at SATURNE at different incident proton energy from 800 MeV to 1.6 GeV [7][8]. Here we will use the experimental data for 1.6 GeV protons interacting with the iron and lead target. Only little data on light charged particle production exist above 100 MeV incident proton energy. Nevertheless, we were able to select a few sets of data. The first one is proton inclusive double differential production cross sections obtained at RCNP [9] with a 300 and 392 MeV proton beam on

carbon, aluminium and niobium thin targets. The particles emitted from nuclear reactions were detected by stacked scintillator spectrometers. The overall systematic uncertainty of absolute cross section values has been estimated to be less than ±10%. The bin width of data (10 MeV) is comparable to the energy resolution of the detectors. The second set is inclusive spectra of 4He particles measured at TRIUMF accelerator [10] with a proton beam of 300 MeV on a silver target. The measurements were made with telescopes composed of plastic and NaI scintillators.

3. Simulation codes Both MCNPX [1] (Version 2.5.f) and FLUKA [2] (Version 2005.6) have been used for benchmark. For energies below 20 MeV (or 150 MeV for some materials), corresponding cross-section data libraries are used during the simulation. Above this energy a reliable physics model has to be chosen to model these reactions. The most common physics approach is the intra-nuclear cascade followed by fissionevaporation model. The physics models used by FLUKA are fixed and cannot be changed by the user. In this case, a preequilibrium cascade model called PEANUT is coupled to an implementation of the RAL fission evaporation code, both with predefined input parameters. The code MCNPX allows the user to choose between different intra-nuclear cascade and fissionevaporation model combinations among ISABEL, BERTINI and INCL4 for cascade and DRESNER (associated with RAL or ORNL fission code) and ABLA for deexcitation. The last possibility is to use the package CEM (cascade and deexcitation). For microscopic cross section predictions the code MCNPX has been used without the particle transport. The probability of an inelastic interaction of an incident proton with a nucleus of the target material is small, so the simulation has been sped up and the statistic increased by allowing only the desired interactions. The spectra shown in this document have been normalized internally by MCNPX using its own cross section libraries. For the normalization of FLUKA distributions, the interaction cross section given by MCNPX has been used. Experimentally measured double differential cross sections are folded with the energy resolution of the detector used. In simulations only the angular acceptance of the detectors is taken account, and no neutron energy resolution function has been implemented.

4. Results 4.1 Neutron double differential cross sections (thin targets) Among a large number of combinations of incident proton energy, material and angles we could study, we have chosen to make calculations for 6 materials, namely Be, C, Fe, W, Pb, and U at the energy of 800 MeV for 5 angles (2.5°, 30°, 60°, 120°, and 150°). Figures 1-2 show the neutron double differential cross section for light nucleus as Be and an heavy nucleus as U. Both FLUKA and MCNPX reproduce well the shape and magnitude of the double differential cross section spectra. Two distinct contributions are visible in the spectra, the evaporation neutrons between 1 and ~20 MeV emitted isotropically, and cascade neutrons above ~20 MeV, which are more forward peaked. We note that the results obtained are more accurate for heavy nuclei than for light nuclei where some important differences appear at low neutron energy (say, below 10 MeV). For MCNPX, five models combinations have been used: INCL4-ABLA, ISABEL-ABLA, BERTINI-RAL, ISABEL-RAL, and CEM as explicitly shown in Figures 1-2. For a more quantitative comparison we have also plotted the ratio between simulation and data (an example is given in Figure 3 for an angle of 30°). Taking account the combined statistical and systematic error on data and simulation, the agreement is below a factor 2 until 600 MeV, except for the CEM model used within MCNPX. Above 600 MeV, and for forward angles in particular around the quasi-elastic and quasi-inelastic peaks, the agreement is not so good whatever the code and models used. We expect that

this disagreement is less important for realistic target simulations, where the neutron energy and angular distributions will be influenced by the multiple scattering with the increasing target thickness.

Figure 1 : Neutron double differential cross section in the case of the reaction Be(p,xn) at Ep=800 MeV. The experimental data are represented in black, and the results of simulation in red.

Figure 2 : Same as Fig. 1 but for the reaction U(p,xn) at Ep=800 MeV.

Figure 3 : Ratio between simulation and data for the Be(p,xn) reaction (on the left) and the U(p,xn) reaction (on the right) for the emission angle of 30°. The width of the lines is larger than the resulting error from data and simulation.

4.2 Neutron double differential cross sections (thick targets) Data for 1.6 GeV protons interacting with thick lead and iron targets are compared [11] using INCL4/ABLA and BERTINI/DRESNER (RAL) model combinations (see Figure 4). Measurements were performed with cylindrical targets that can be translated longitudinally with the help of collimators.

Figure 4 : Experimental and simulated neutron spectra for 1.6 GeV protons interacting with Fe (left) and Pb (right) targets. On the right of each figure, a diagram shows for each angle the full emission zone of the target seen by the detector.

In this way, neutrons coming from different emission zones can be selected, allowing to test the propagation of the cascade along the target. Both models show a good agreement with the data, and it is difficult to distinguish them. Indeed, with thick target the differences between models are difficult to quantify since the number of emitted neutrons is averaged due to the transport and multiple interactions in the thick target material. Theses results have been obtained with LAHET code, the same calculations have been done with MCNPX which includes LAHET, and are giving a similar behavior whatever the model combination is used.

4.3 Light charged particle production Figure 5 shows the proton double differential production cross section obtained for thin carbon and niobium targets. Results obtained by simulation are rather good for FLUKA and MCNPX used with ISABEL, INCL4 and CEM models. However, some important discrepancies are seen at forward angles and high proton energies with respect to the beam energy. In the case of production of helium (see Figure 8) also huge differences between models can be seen: except for CEM there is no 4He particles emitted above ~50 MeV, and the shape of the distribution is not correctly reproduced. It has to be stressed separately that only CEM is able to emit high energy alphas, while the other intranuclear cascade models (like ISABEL, INCL4, Bertini inside MCNPX and PEANUT inside FLUKA) are unable to emit energetic composites. The similarity of isotropic contribution at low energy is mainly given by evaporation (ABLA and DRESNER models). We can see that FLUKA and MCNPX with ISABEL-RAL give comparable results due to nearly the same evaporation fission models. We note that the evolution of the total production cross section as a function of the incident proton energy is really different between models (cross sections from simulation with MCNPX are listed in Tables 1-2). Some simulations also show that the maximum difference on proton production cross section

at Ep= 600 MeV is less than 10%, but at Ep= 2.5 GeV the difference can reach 60 %. Therefore we conclude that present intranuclear cascade models have difficulties to predict consistent values for production cross sections. This is in particular true for 4He, which is produced at much lower rate compared to protons (see Tables 1-2). It has to be mentioned that an attempt to emit light charged particles (d,t,3He,α) has been made [12] which allows to hope better results with INCL4/ABLA in the future. Table 1 : Proton production cross sections for different models used in MCNPX as a function of proton incident energy (in mbarn).

p + Pb → p + X Ep = 600 MeV Ep = 800 MeV Ep = 1.2 GeV Ep = 1.8 GeV Ep = 2.5 GeV

INCL4-ABLA 3712 4586 5966 7176 7874

CEM 3500 4478 6347 8968 12030

ISABEL-RAL 3966 4894 6130 6800 6950

ISABEL-ABLA 3468 4283 5442 6160 6370

Table 2 : Helium production cross sections for different models used in MCNPX as a function of proton incident energy (in mbarn)..

p + Pb → α + X Ep = 600 MeV Ep = 800 MeV Ep = 1.2 GeV Ep = 1.8 GeV Ep = 2.5 GeV

INCL4-ABLA 560 788 1157 1442 1556

CEM 306 502 926 1571 2346

ISABEL-RAL 701 925 1215 1340 1331

ISABEL-ABLA 796 1053 1353 1487 1476

Similarly as for neutron production, the ratio between simulation and data has been presented for more quantitative comparison. For proton production, all codes and models give results with differences generally less than a factor 2 in the case of the Nb target (Figure 7). For the C target (Figure 6), in the most of the cases FLUKA gives the worst results (between a factor 2 and 5) compared to other model predictions. As we already seen, results are not so good for alpha emission (Figure 9); here only CEM used with MCNPX gives results below a factor 2 for forward angles. Other models are showing a divergence of several decades with respect to data.

Figure 5 : Double differential cross section for proton production from 392 MeV protons interacting with carbon (left) and niobium (right) targets.

Figure 6 : Ratio between simulation and data data in the case of the reaction C(p,xp) at Ep=392 MeV.

Figure 7 : Ratio between simulation and data in the case of the reaction Nb(p,xp) at Ep=392 MeV.

Figure 8 : Helium double differential cross section production in the case of the reaction Ag(p,4He) at Ep=300 MeV.

Figure 9 : Ration between simulation and data in the case of the reaction Ag(p,4He) at Ep=300 MeV and as a function of emission angle.

5. Conclusion We have used both MCNPX and FLUKA codes to predict the production of neutrons, protons and alpha particles from incident protons with energies between 300 MeV and 1.6 GeV on thin and thick targets of different materials. Comparison of the simulation with experimental data shows a good agreement of codes and models for neutron production. For secondary proton production FLUKA have difficulties to reproduce the energy-angle distributions for light targets. As long as alpha production is concerned, actually only CEM employed within MCNPX gives reasonable results. However, due to the lack of data for incident protons above 400 MeV we are unable to give a definitive recommendation for calculations involving alpha particle emission.

References [1] MCNPX- Monte Carlo N-Particle Transport Code System for Multiparticle and High Energy Applications; http://mcnpx.lanl.gov/ (March 2006). [2] A. Ferrari, P.R. Sala, A. Fasso, and J. Ranft, “FLUKA: a multiparticle transport code”, CERN 200510 (2005), INFN/TC-05/11, SLAC-R-773; http://pcfluka.mi.infn.it/ (March 2006). [3] For more information the reader is asked to check the information available the EURISOL Design Study web page at http://www.eurisol.org (March 2006).

[4] A. Herrera-Martinez and Y. Kadi, “Eurisol Multi-MW Target: Preliminary Study of the Liquid Metal Proton-to-Neutron Converter”, EURISOL DS/TASK2/TN-05-01, July 2005. [5] W. B. Amian et al., Nucl. Sci. Eng., vol. 112, 1992, p.78 [6] EXFOR: Experimental Nuclear Reaction Data (EXFOR / CSISRS), http://www.nndc.bnl.gov/exfor3/exfor00.htm [7] S. Ménard, PhD Thesis, Université d'Orsay (1998) [8] C. Varignon, PhD Thesis, Université de Caen (1999) [9] Tadahiro Kin et al., Physical Review C 72, 014606 (2005) [10] R. E. L. Green et al., Physical Review C, vol. 35, Number 4 (1987), p. 1341 [11] J.-C. David et al., "Proceedings of the International Workshop on Nuclear Data for the Transmutation of Nuclear Waste" GSI-Darmstadt, Germany, September 1-5, 2003, ISBN 3-00-012276-1 Editors: Aleksandra Kelic and Karl-Heinz Schmidt. [12] A. Boudard et al., Nuclear Physics A 740 (2004) 195-210