Fission fragment distributions and their impact on the r-process nucleosynthesis in neutron star mergers

J.-F. Lemaˆıtre∗ Institut d’Astronomie et d’Astrophysique, Universit´eLibre de Bruxelles, Campus de la Plaine CP 226, BE-1050 Brussels, Belgium and CEA, DAM, DIF, F-91297 Arpajon, France

S. Goriely Institut d’Astronomie et d’Astrophysique, Universit´eLibre de Bruxelles, Campus de la Plaine CP 226, BE-1050 Brussels, Belgium

A. Bauswein GSI Helmholtzzentrum f¨urSchwerionenforschung, Planckstraße 1, 64291 Darmstadt, Germany and Helmholtz Forschungsakademie Hessen f¨urFAIR (HFHF), GSI Helmholtzzentrum f¨urSchwerionenforschung, Campus Darmstadt, Germany

H.-T. Janka Max-Planck-Institut f¨urAstrophysik, Postfach 1317, D-85741 Garching, Germany

Neutron star (NS) merger ejecta offer a viable site for the production of heavy r-process elements with nuclear mass numbers A & 140. The crucial role of fission recycling is responsible for the robustness of this site against many astrophysical uncertainties. Here, we introduce new improve- ments to our scission-point model, called SPY, to derive the fission fragment distribution for all neutron-rich fissioning nuclei of relevance in r-process calculations. These improvements include a phenomenological modification of the scission distance and a smoothing procedure of the distribu- tion. Such corrections lead to a much better agreement with experimental fission yields. Those yields are also used to estimate the number of neutrons emitted by the excited fragments on the basis of different neutron evaporation models. Our new fission yields are extensively compared to those predicted by the so-called GEF model. The impact of fission on the r-process nucleosynthesis in binary neutron mergers is also re- analyzed. Two scenarios are considered, the first one with low initial electron fraction is subject to intense fission recycling, in contrast to the second one which includes weak interactions on nucle- ons. The various regions of the nuclear chart responsible for fission recycling during the neutron irradiation as well as after freeze-out are discussed. The contribution fission processes may have to the final abundance distribution is also studied in detail in the light of newly defined quantitative indicators describing the fission recycling, the fission seeds and the fission progenitors. In particular, those allow us to estimate the contribution of fission to the final abundance distribution stemming from specific heavy nuclei. Calculations obtained with SPY and GEF fission fragment distributions are compared for both r-process scenarios.

I. INTRODUCTION γ-ray bursts has raised a lot of excitement about the so- called neutrino-driven wind environment. However, until The r-process, or the rapid neutron-capture process, now a successful r-process cannot be obtained ab initio without tuning the relevant parameters (neutron excess,

arXiv:2102.03686v1 [nucl-th] 6 Feb 2021 of stellar nucleosynthesis is invoked to explain the pro- duction of the stable (and some long-lived radioactive) entropy, expansion timescale) in a way that is not sup- neutron-rich nuclides heavier than iron that are observed ported by the most sophisticated existing models [4, 5]. in stars of various metallicities, as well as in the solar sys- Although these scenarios remain promising, especially in tem (for a review, see Ref. [1–3]). In recent years, nuclear view of their potential to contribute to the galactic en- astrophysicists have developed more and more sophisti- richment significantly, they remain affected by large un- cated r-process models, trying to explain the solar system certainties associated mainly with the still incompletely composition in a satisfactory way by adding new astro- understood mechanism responsible for the supernova ex- physical or nuclear physics ingredients. The r-process re- plosion and the persistent difficulties to obtain suitable r- mains the most complex nucleosynthetic process to model process conditions in self-consistent dynamical explosion from the astrophysics as well as nuclear-physics points of and neutron-star (NS) cooling models [5–8]. In particu- view. Progress in the modeling of type-II supernovae and lar, a subclass of core-collapse supernovae, the so-called collapsars corresponding to the fate of rapidly rotating and highly magnetized massive stars and generally con- sidered to be at the origin of observed long gamma-ray ∗ [email protected] bursts, could be a promising r-process site [9]. The pro- 2 duction of r-nuclides in these events may be associated be found for instance in Refs. [14, 32, 43–52]. with jets predicted to accompany the explosion, or with Despite the recent success of nucleosynthesis studies for the accretion disk forming around a newly born central NS mergers, the details of r-processing in these events is BH [10]. still affected by a variety of uncertainties, both from the Since early 2000, special attention has been paid to NS nuclear physics and astrophysics point of view. For this mergers as r-process sites following the confirmation by reason, we present here a new effort to further improve hydrodynamic simulations that a non-negligible amount the description of fission fragment distributions (FFD) of matter could be ejected from the system. Newtonian obtained on the basis of the so-called Scission Point Yield [11–15], conformally flat general relativistic [16–18], as (SPY) model [53]. Sec. II is devoted to the description well as fully relativistic [19–23] hydrodynamical simula- of such improvements. A phenomenological correction to tions of NS-NS and NS-black hole (BH) mergers with the scission distance and a new smoothing procedure for microphysical equations of state have demonstrated that deriving the fission yields are presented in Sec. II.1. Since −3 typically some 10 M up to more than 0.1 M can the FFD of relevance for the r-process application cor- become gravitationally unbound on roughly dynamical respond to the post-neutron-emission one deduced from timescales due to shock acceleration and tidal stripping. pre-neutron FFDs, our procedure for estimating the neu- Also the relic object (a hot, transiently stable hypermas- tron evaporation is detailed in Sec. II.2. As r-process sive NS [24] followed by a stable supermassive NS, or a is impacted by neutron captures, sensitivity studies are BH-torus system), can lose mass through outflows driven performed in Sec. II.3 to demonstrate that the neutron by a variety of mechanisms [18, 25–30]. emission depends mainly on the primary fission yields. SPY pre-neutron FFDs are compared with experimental Simulations of growing sophistication have confirmed data and the well-known GEF (“GEneral description of that the ejecta from NS mergers are viable strong r- Fission observables”) calculations [54] in Sec. II.4. process sites up to the third abundance peak and the In addition, it has been shown that weak interactions actinides [15, 17, 18, 22, 31]. The r-nuclide enrich- may strongly affect the composition of the dynamical ment is predicted to originate from both the dynamical ejecta and thus the efficiency of the r-process [22, 55– (prompt) material expelled during the NS-NS or NS-BH 63]. In this case, the initial neutron richness of the ejecta merger phase and from the outflows generated during the can be significantly reduced and consequently the role post-merger remnant evolution of the relic BH-torus sys- of fission, albeit in basically all calculations a significant tem. The resulting abundance distributions are found fraction of neutron-rich ejecta is present. In Sec. III, we to reproduce very well the solar system distribution, as re-evaluate the role of fission in NS merger in two distinct well as various elemental distributions observed in low- scenarios, namely the traditional one neglecting weak in- metallicity stars [3]. During the dynamical phase of the teractions on nucleons and an updated version of such a merging scenario, the number of free neutrons per seed simulation where neutrino interactions are included phe- nucleus can reach so high values (typically few hundreds) nomenologically [56, 62] to reproduce qualitatively the that heavy fissioning nuclei are produced. Fissioning re- results found in Ref. [63]. In Sec.III.1, we will present sults in a robust reproduction of the solar-system-like the final abundance distributions obtained within both abundance pattern of the rare-earth elements, as ob- scenarios as well as the role of the fission recycling and served in metal-poor stars [1, 3, 17, 32]. This supports the link them to fundamental initial properties of ejected ma- possible production of these elements by fission recycling terial, namely the initial neutron mass fraction, electron in NS merger ejecta. In addition, the ejected mass of r- fraction and entropy. The relevance of the fission re- process material, combined with the predicted astrophys- cycling and its contribution to the final r-abundances is ical event rate (around 10 My−1 in the Milky Way [33]) thoroughly detailed in Secs. III.2–III.4 for both scenarios. can account for the majority of r-material in our Galaxy, Finally, conclusions are drawn in Sec. IV. e.g. [17, 21, 34, 35]. A further piece of evidence that NS mergers are r-nuclide producers indeed comes from the very important 2017 gravitational-wave and electromag- netic observation of the kilonova GW170817 7,[36–42]. II. FISSION FRAGMENTS DISTRIBUTIONS In this specific NS merger scenario, the neutron rich- ness was found to be so high that heavy fissioning nuclei II.1. SPY model can be produced. For this reason, in this astrophysical site, fission plays a fundamental role, more particularly The SPY model is a static and statistical scission point by i) recycling the matter during the neutron irradia- model [53, 64] where a thermodynamic equilibrium at tion (or if not, by allowing the possible production of scission is assumed, hence the evolution between the sad- super-heavy long-lived nuclei, if any), ii) shaping the r- dle and the scission points is neglected. The model is abundance distribution in the 110 ≤ A ≤ 170 mass re- based on two pillars, namely the absolute available en- gion at the end of the neutron irradiation, iii) defining ergy balance at the scission configurations and the statis- the residual production of some specific heavy stable nu- tical description of the available phase space. The avail- clei, more specifically Pb and Bi, but also the long-lived able energy balance is performed for all energetically pos- cosmochronometers Th and U, and iv) heating the envi- sible fragmentations of a fissioning system at scission as ronment through the energy released. More details can a function of the deformation of both fragments. The 3 available energy is defined as the difference between the potential energy of the fissioning system at scission and the energy of the excited compound nucleus where both nascent fragments are supposed to be at rest. The po- tential energy of the fissioning system at scission is ob- tained as the sum of the individual binding energies of the two fragments and the interaction energy between the fragments composed of the Coulomb repulsion and the nuclear attraction. The system at scission is treated as a microcanonical ensemble where all available states are equiprobable. In this framework, the number of avail- able states of a given fragmentation is the product of the state densities of the two isolated fragments. The yield of a fragmentation is the number of available states as- sociated with this fragmentation regardless of the defor- mation of the fragments. All nuclear inputs to the SPY model, i.e. the individ- ual binding energies, proton distributions and proton and neutron single-particle level schemes of each fission frag- ment as a function of its axial deformation are estimated for about 7000 nuclei from Z = 20 to Z = 100 within the framework of the constrained self-consistent Hartree- Fock-Bogoliubov (HFB) formalism using the Skyrme FIG. 1: (Color online) (a) Distance correction BSk27 effective nucleon-nucleon interaction [65]. The nu- associated with the light fragment, d(N1) with clear state density is calculated in the framework of the N1 6 NCN/2 (orange curve), with the heavy fragment, statistical model of nuclear level densities [53, 66–69] on d(N2) with N2 > NCN/2 (green curve), and with the 236 the basis of the discrete single-particle level scheme ob- total one d(N1) + d(N2) (black dashed curve) for U tained self-consistently within the same HFB calculation. fission. The specific neutron numbers Ncorr (Eq. 2) are A scission criteria is needed to characterize the scission indicated by gray arrows. (b) Available energy of configurations, more precisely, to define the relative po- fragments from thermal neutron-induced fission of 235U sition of the fragments, hence to estimate the Coulomb calculated with (red curve) or without (green curve) the and nuclear energies. This criteria can be based on an phenomenological distance correction (Eq. 2). The blue inter-surface distance, as in the initial version of the SPY line corresponds to the the difference between corrected model [64] or through the proton density at the scission and non-corrected energies. neck which defines the scission distance as in our new ver- sion of the SPY model [53]. Presently, with respect to our last study [53], a phenomenological modification of the for example by around 3 MeV for the neutron-induced fis- scission distance is introduced in order to better repro- sion of 235U (Fig. 1b). Such an effect is of the same order duce experimental fission yields. The corrected scission of magnitude as pairing effects and can be regarded as distance (d ) for a fragmentation (Z ,N )+(Z ,N ) sc,corr 1 1 2 2 a way to correct the approximate pairing effects which is defined through the proton density at the scission neck assumes the pairing gap at zero temperature can be de- (d ), as in Ref. [53], though with two additional correc- sc scribed by a five-points mass difference [70] and remains tions, one per fragment constant with the nucleus deformation. For nucleosyn- thesis applications, the distance correction is kept con- dsc,corr = dsc + dcorr(N1) + dcorr(N2) . (1) stant irrespective of the fissioning system, even if a spe- Each correction term is composed by a positive term and cific adjustment could be performed. a negative one which are chosen to be dependent on the This distance correction help to better reproduce the difference between the fragment neutron number (N) and yields of the thermal neutron induced fission of 235U some specific neutron numbers (Ncorr), i.e. (Fig. 2a and d, green and blue curves), more particu- larly its symmetric component is reduced due to the dis- 2 X − (N−Ncorr) tance correction around N = 64, 66, 68, 70. As seen in dcorr(N) = 0.2 e 2 Fig. 2c and f, this correction has a negligible impact on Ncorr∈{47,51,81,85,89} 252 2 the FFD for the spontaneous fission of Cf. However, X − (N−Ncorr) 239 − 0.2 e 2 [fm] . yields of the thermal neutron-induced fission of Pu

Ncorr∈{64,66,68,70,80} around A = 130 becomes underestimated (Fig. 2b and e) (2) due to the same correction as in the U case. Both light and heavy fragments contribute to the dis- As described in Ref. [53], the SPY FFD or yields dis- tance correction (Fig. 1a) with a maximal correction of tribution present strong staggering patterns that are not ±0.4 fm. The available energy is impacted by a few MeV, found experimentally. Those can be smoothed, as tradi- 4 tionally done, by a double Gaussian function where T is the nuclear temperature before neutron evap- oration. T can be extracted from the statistical model Ysmooth(Z,N) = of nuclear level densities including BCS pairing interac-

4 15 2 2 tion, denoted as TBCS (see Eqs. 16 and 18 in Ref. [53] − i − j X X 2σ2 2σ2 Yraw(Z + i, N + j)Cze z Cne n (3) and Refs. [66–68]). i=−4 j=−15 Following this approximation, the distribution of evap- orated neutrons are calculated for the three well-known where Cz and Cn are normalization factors and σz = 0.55 fissioning systems using both raw and smooth pre- and σn = 3 are adopted. In the present paper the width neutron yields and shown in Fig. 3. In both cases, the of the Gaussian function is larger in the neutron direc- number of emitted neutrons is underestimated for the tion σn than in the proton direction in order to be si- heaviest fragments A = 150 − 160 due to the low avail- multaneously consistent with experimental width of the able energy, which, in turn, originates from an overes- isobaric and isotopic fission yields distributions (Fig. 2). timate of the kinetic energy of these related fragmenta- As there is not coupling between Z and N in the dou- tions, as mentioned in Sec. IV of Ref. [53]. However, for ble Gaussian function (Eq. 3), the isotopic yields are not heavy fragments with A = 130 − 140, the emitted neu- affected by the smoothing in the neutron direction. Simi- tron number is overestimated which, in this case, is due larly to yields, other fragments observables O, such as the to an underestimate of the total kinetic energy of the fis- available energy or the kinetic energy, are also smoothed sion fragments. The distributions of evaporated neutron following the same procedure deduced from smooth yields (Fig. 3, orange curves) fol- low roughly the same pattern as the raw ones except for Osmooth(Z,N) = the heaviest fragments region where the number of emit- 4 15 2 2 − i − j ted neutron is larger. This overestimation is due to the X X 2σ2 2σ2 Oraw(Z + i, N + j)Cze z Cne n . (4) smoothing procedure (Eq. 4) which leads to an increase i=−4 j=−15 of the total excitation energy by around 2 MeV, hence sometimes makes the emission of an additional neutron possible depending on its kinetic energy. More generally, II.2. Neutron evaporation model the number of neutrons emitted by the light fragment is in better agreement with data, though slightly underesti- mated, than for the heavy fragment where ν is systemati- The SPY model allows us to investigate the primary cally overestimated. This is linked to the available energy fragments, i.e. fragments formed at scission, hence their sorting within a fragmentation which is too favorable for de-excitation by neutron evaporation. The excitation en- the heavy fragment. ergy of fragments has two components. First, each frag- The overestimation the number of neutron emitted by ment carries a fraction of the available energy at scis- the heavy fragments leads an overestimation of the to- sion under an intrinsic excitation form. Second, since tal mean number of evaporated neutrons per fission. As each fragment can be deformed, a deformation energy displayed in Table I, it is overestimated by 0.77-1.1 neu- may contribute to the excitation energy. No assump- tron when smooth yields are used and by 0.6-1 neutron tion about available energy sorting at scission between for raw yields. The overestimation is higher with smooth primary fragments is needed within the SPY framework, yields due to a more important contribution of the heavy contrary to other neutron evaporation models. Indeed, fragments. some models require to introduce an energy sorting pa- rameter R (A) [80, 81] or a thermal equilibrium to de- T TABLE I: Total mean number of evaporated neutrons duce the energy sorting on the basis of level density pa- per fission,ν ¯ , from the evaluated data library rameters with phenomenological corrections in order to tot ENDF/B-VII.1 [83] and SPY using the raw or smooth increase the excitation energy of the light fragment [82]. yields (see Fig. 3) for the 3 benchmarked nuclei shown In the SPY model, the energy sorting, denoted by x in Fig. 3 and obtained either by neutron-induced fission (Eq. 2 in Ref. [53]), results from the competition between (nif) or spontaneous fission (sf). the state density of the light fragment (ρ1) and the heavy one (ρ2). The intrinsic excitation energy and deforma- tion energy of each fragment are obtained by averaging Nucleus reaction ENDF SPY(raw) SPY(smooth) over all possible energy sorting and deformations (Eq. 3 235U nif 2.44 3.08 3.20 in Ref. [53]). In the simplest model, the competition 239Pu nif 2.88 3.97 3.96 between neutron and gamma evaporation during the de- 252Cf sf 3.77 4.37 4.39 excitation cascade is not taken into account, so that the fission fragments are sequentially de-excited by neutron To study the sensitivity of the predicted number of evaporation as long as it is energetically possible. The ki- evaporated neutrons to the various model uncertainties, netic energy of evaporated neutrons En can be assumed the impact of the neutron kinetic energy distribution to follow a Maxwellian distribution φ(En), the nuclear temperature and the neutron-gamma 2 p −En/T competition are analyzed in the next sub-sections for the φM (En) = √ Ene , (5) 252 πT 3 specific case of Cf spontaneous fission. 5

FIG. 2: (Color online) Isotopic (top panels) and isobaric (bottom panels) fission yields of 236U, 240Pu and 252Cf. The black lines with dots represent experimental (pre-neutron-emission) fission yields for fission of 236U (Q = 6.5 MeV) (a) [71] and (d) [72], 240Pu (Q = 6.5 MeV) (b) [73] and (e) [74] and 252Cf (Q = 0 MeV) (c) [75] and (f) [76]. The green lines (labeled by “raw”) correspond to raw fission yields, the blue lines (labeled by “dcorr”) to raw yields including distance corrections and the orange lines (labeled by “smooth”) to the final smooth yields after distance corrections.

FIG. 3: (Color online) Evaporated neutron distributions as a function of the fragment mass number. The black dotted lines represent experimental evaporated neutron distribution of (a) 236U(Q = 6.5 MeV) [77], (b) 240Pu (Q = 6.5 MeV) [78] and (c) 252Cf (Q = 0 MeV) [79]. The orange/green lines (labeled by “smooth”/“raw”) are SPY evaporated neutron distribution using smooth/raw pre-neutron yields. 6

considered to estimate the nuclear temperature, p TFG = U/a (7)

where U is the excitation energy and a ∼ A/8 is the em- pirical high-energy limit of the level density parameter [88]. As seen in Fig. 4 (cyan and black curves, labeled respectively by “φM ,TFG” and “φW ,TFG”), both models for the nuclear temperature (TBCS or TFG) give rather similar predictions of the evaporated neutron distribu- tion. For the sake of coherence, TBCS is adopted in the subsequent study.

II.3.3. Impact of neutron-gamma competition

In the previous calculation of the number of emitted number of neutrons, it was assumed that neutron emis- sion dominates over the electromagnetic de-excitation. To test the impact of such an approximation, TALYS nu- FIG. 4: (Color online) (a) Evaporated neutron clear reaction code [89] is now used to describe the com- distribution of spontaneous fission of 252Cf using the petition between the strong and electromagnetic chan- SPY raw yields. (b) Difference with respect to the nels in the de-excitation of the the primary fragments evaporated neutron distribution computed with the (Fig. 4, green curve, labeled by “TALYS”). Note that Maxwell distribution (Eq. 5) and temperature from the all de-excitation channels are taken into account in the BCS state density. Hauser-Feshbach and pre-equilibrium framework. The difference between the evaporated neutron distribution from TALYS and the one obtained with the simplified II.3. Sensitivity studies of the evaporation model Maxwellian approach described above is shown in Fig. 4 (green curve, labeled by “TALYS”) and does not exceed II.3.1. Impact of the kinetic energy distribution of the 0.3 neutron. neutron These results indicate that the primary fragments de- excitation can be essentially approximated by a sequence There are various models for the kinetic energy dis- of neutron emissions, followed finally by gamma emission tribution of the neutron φ(En) [84]. The chosen one is when no more neutron can be evaporated. This simpli- related to the Maxwell-Boltzmann distribution describ- fied description based on energetic considerations seems ing the distribution of speeds of particles in an idealized to be sufficient within an accuracy of about 0.1–0.2 neu- gas which yields to the kinetic energy distribution of neu- tron. This conclusion is only valid for low excited pri- tron φM (Eq. 5). Another possible description relies on mary fragments up to 20–25 MeV. The crucial point to the evaporation model of Weisskopf and Ewing [85, 86] improve the description of the emitted neutron distribu- yielding to tion concerns the SPY model and particularly the kinetic energy of the fission fragments which directly affect the En −En/Tn available energy and mean deformation of the fragments. φW (En) = 2 e (6) Tn In the following, if not mentioned otherwise, the smooth yields (Eq. 3) are adopted and neutron evapo- for small En [84] and where Tn is the nuclear tempera- ture after the neutron emission. The impact of the neu- ration are deduced using the Maxwell distribution pre- tron kinetic energy distribution on the number of evap- scription (Eq. 5) with the BCS temperature. orated neutron of 252Cf spontaneous fission is negligible, as shown in Fig. 4, (orange and light green curves, labeled II.4. Systematics respectively by “φM ,TBCS” and “φW ,TBCS”) where the difference is seen to rarely exceed 0.1 neutron (Fig. 4b, light green curve, labeled by “φW ,TBCS”). II.4.1. Over all fissionable nuclei

Systematic SPY calculations were performed for iso- II.3.2. Impact of the nucleus temperature topic chains between Z = 70 and 124 from the proton to the neutron driplines with an initial excitation energy As an alternative to the BCS theory of nuclear level Q = 8 MeV (Fig. 5). With Q = 8 MeV, only few nu- densities, the phenomenological Fermi Gas approxima- clei with Z < 80 can fission. The smoothing procedure tion [87], excluding shell and pairing effects, can also be (Eq. 3) may impact locally the peak multiplicity (Fig. 5c) 7

(a) Peak multiplicity for the raw pre-neutron isobaric yields (b) Mean prompt neutron multiplicity per fission according to with corrected distance (Eq. 1). the prescription given in Sec. II.2, deduced from the smooth pre-neutron isobaric yields with corrected distance (Eq. 1).

(c) Peak multiplicity for the smooth pre-neutron isobaric (d) Mean available energy release per fission deduced from the yields with corrected distance (Eq. 1). smooth pre-neutron isobaric yields with corrected distance (Eq. 1).

(e) Peak multiplicity for the smooth pre-neutron isobaric (f) Isolines of mean available energy release per fission (orange yields without corrected distance. lines) from Fig. 5d and mean prompt neutron multiplicity per fission (green lines) from Fig. 5b.

FIG. 5: (Color online) Systematics in the (N,Z) plane of major fission observables for some 3000 nuclei with 70 6 Z 6 124 for an initial excitation energy of Q = 8 MeV.

when compared those obtained with raw yields (Fig. 5a). deficient nuclei from NCN ≈ 140 to NCN ≈ 133. The In particular, peaks can be merged if there are close each symmetric mode is disfavored by the negative distance other or if they are small enough. However, the location correction for Nfrag = 64, 66, 68, 70 leading to the ap- of transition between the various fission modes are seen pearance of the asymmetric mode while for Nfrag = not to be globally affected by the smoothing procedure 47, 51, 81, 85, 89 the positive distance correction favors and does not distort significantly the peaks location of the asymmetric mode. In the second zone, located at FFDs and their widths as shown in Fig. 2. 155 . NCN . 165, a second asymmetric mode emerges, less asymmetric than the one without distance correc- Four zones can be identified where the peak multiplic- tion. It is induced by a negative distance correction for ity is affected by the distance correction (Eq. 2), as seen the light fragment of the most asymmetric fission mode, when comparing Figs. 5c and 5e obtained with or with- i.e. Nfrag = 64, 66, 68, 70 and by a positive distance cor- out this correction. In general terms, a negative/positive rection for the heavy fragment of the new slightly asym- distance correction makes fragments closer/farther which metric fission mode (Nfrag = 85, 89). The peak multi- induced a decrease/increase of the available energy, hence plicity increases from 2 to 4 when the distance correc- of the fission probability. The first one is the shift of the tions are important. For more neutron-rich nuclei, in the symmetric/asymmetric fission transition toward neutron- 8 third zone 165 . NCN . 180, a symmetric mode emerges impacts moderately the shape of the FFD. for high distance corrections. It is induced by a nega- tive distance correction due to the light fragment of the In the Po to U region (Fig. 6a), the transition between asymmetric fission mode (Nfrag = 64, 66, 68, 70) which symmetric fission for neutron-deficient nuclei to asym- disfavors the asymmetric mode. The symmetric mode is metric fission for neutron-rich nuclei is rather different further accentuated by the positive distance correction between both SPY and GEF predictions. SPY FFDs for Nfrag = 81, 85, 89. Finally, for nuclei in the fourth are in good agreement with experimental data [90] for zone 180 . NCN . 200, the positive distance correction Th and Pa isotopes where the transition is well repro- for Nfrag = 85, 89 induces a slightly asymmetric fission duced, in contrast to what is obtained by the GEF model. mode on top of the symmetric one. Although some fis- The transition from symmetric to asymmetric fission pre- sion yields distributions are impacted by the phenomeno- dicted by GEF occurs for more neutron-deficient isotopes logical correction to the scission distance, globally, SPY compared to experimental data. predictions for neutron-rich nuclei remains rather robust with respect to such a correction, as seen in Figs. 5c – For Th isotopes, GEF transition occurs between 5e. 216 226 Th126 and Th136 for which no symmetric peak is The mean number of neutrons emitted per fissionν ¯tot found anymore. From an experimental point of view, ranges from 2 up to 37 neutrons depending on the fission- 222 the transition seems to start at Th132 where the sym- ing system (Fig. 5b). It depends mostly on the neutron 229 metric peak broadens while beyond Th139 the FFD is richness of the fissioning nucleus. Its evolution with pro- completely asymmetric. The symmetric to asymmetric ton number is not trivial despite the strong variation of transition of Th predicted by SPY is slightly more com- the mean available energy of fragmentations (Fig. 5d). plex (Fig. 5e and Fig. 6a); there are two peaks in addi- For a given fissioning nucleus, an increase of the exci- 222−226 tion to the symmetric component for Th132−136 tation energy (Q) enhances the intrinsic excitation en- which gives a peak multiplicity of two and even four ergy of fission fragments, hence more neutrons are emit- 226 for Th136. For more neutron-rich Th, this substruc- ted. However, this relation between available energy and ture disappears into one symmetric peak which become emitted neutrons is not sufficient to explain the relation smaller for more neutron-rich isotopes and completely between emitted neutrons and mean available energy of 235 vanishes for Th145 (Fig. 5e). This kind of transition fragmentations. Along a line of constant available energy from symmetric to asymmetric through a substructure (Fig. 5f) the number of evaporated neutron is multiplied at the top of the vanishing symmetric peak is also visible by at least five between proton-rich and neutron-rich fis- on the SPY Pa FFDs. sioning nuclei. Similarly, along a constant line of emitted neutron number (Fig. 5f), the available energy increases The FFD broadening in the Rn to Ac isotopic chains by a factor up to hundred between light and heavy nuclei. is well reproduced by SPY even if the FFDs of neutron- These features are due to the neutron binding energy of deficient isotopes tend to be narrower than the experi- fission fragments, i.e. the more neutron-rich the fission- mental ones. A strong odd-even staggering in GEF FFD ing nucleus, the more-neutron rich the fission fragments is observed for even isotopes of Po and Rn which makes and the lower their neutron binding energy, hence the FFD plots completely blue filled (Fig. 6a). more neutrons can be evaporated. Even if the fission of a heavy nucleus releases more available energy compared to a light one, the number of evaporated neutrons can be For actinides from U to Rf (Fig. 6b), FFDs from SPY similar because the neutron binding energies of fragments and GEF model are rather similar and in good agreement from heavy-nucleus fission is higher (less neutron-rich) with experimental data, particularly for U and Es iso- than the fragments from the lighter one. topes. Compared to experimental data for Am and Cm isotopes, the symmetric part of the yields distributions is underestimated by both models. The peaks are nar- rower with SPY compared to GEF model. However, for II.4.2. SPY vs GEF neutron-rich nuclei (like 258Fm), SPY predicts a purely asymmetric fission, contrary to GEF and evaluated data. SPY FFDs with corrected distance are compared in This asymmetric splitting predicted by SPY model is due Fig. 6 to experimental data or with the evaluation from to the doubly magic nuclei 132Sn which is disfavored. In- the ENDF/B-VII database [83], if no experimental data deed, even if 132Sn has more available energy than other is available. Evaluated data (Fig. 6, gray curves) cor- fragmentations, the available states for these fragmenta- respond to post-neutron FFDs while experimental data tions is significantly lower and consequently their yields is can be pre-neutron or post-neutron emission (Fig. 6, lower than other fragmentations leading to an asymmet- black curves). SPY FFDs (Fig. 6, green curves) are also ric fission. 132Sn affects fissioning systems in the region compared to GEF (version 2019/1.2) thermal neutron- around 258Fm (Fig. 5e) where a doubly asymmetric fis- induced fission yields [54, 103]. The excitation energies sion is found by SPY model whereas GEF model predicts of experimental or evaluated data may not be exactly the an symmetric mode. For the same reasons, the symmet- same as those considered in SPY or GEF calculations but ric fission of 259Lr and 262Rf are not reproduced with the a variation of the excitation energy by a few MeV only SPY model contrary to the GEF model. 9

(a) Evaluated isotopic yields (formed by Coulomb excitation) from Ref. [90] converted into isobaric yields using the UCD (Unchanged Charge Density) hypothesis. For 230Th post-neutron yields from ENDF/B-VII database [83].

(b) Yields of 232−233U are from [90] (converted into isobaric), 234U from [91], 235−237U (formed by 0.5 MeV neutron) from [83], 236U from [72], 238U and 239Np and 244Cm from [92], 239U [76], 238Np from[93], 240Pu from [94], 242−243Pu and 255Es (formed by thermal neutron) from [83], 243Am from [95], 244Am from [96], 246Cm and 253Es and 254−256Fm (spontaneous fission) from [83], 248Cm from [97], 250Cf from [98], 252Cf from [76], 256Cf and 258Fm from [99], 244−246−248Fm from [100], 259Lr from [101], 262Rf from [102].

FIG. 6: (Color online) Isobaric fission yields from SPY model (green curves) compared to those calculated by the GEF model (blue curves) [54] and experimental pre-neutron (black curves) or post-neutron (gray curves) yields. For each FFD, the x axis (mass number) ranges from A = 70 to A = 160 by step of 10 and the y axis (FFD) from 0 to 15% by step of 2%. The background color refers to the peak multiplicity. 10

FIG. 7: (Color online) Representation in the (N,Z) plan the decay mode with the highest decay rate (a) between β decay followed by 0, 1, 2 or 3 neutron emission (βkn), β-delayed fission (βdf), spontaneous fission (sf) or α decay; (b) the corresponding lowest lifetime τmin = 1/λmax among the various decay mode given in (a). (c) Temperature at which fission and radiative neutron capture are equally favorable, above this temperature the neutron induced fission dominates the neutron capture. (d) Nuclei for which neutron-induced fission is dominant over the (n, γ) channel for T < 0.2 GK or for which the main decay mode is spontaneous fission or β-delayed fission are depicted by the gray area (region of efficient fission. Blue hatched areas are the fission seeds areas : the “fission bottleneck” (A < 292 and N > 184) hatched by blue circles, the “fission roof” (A > 320) hatched by blue vertical+horizontal lines, the “decay roof 0” (A < 278) hatched by blue vertical lines, the “decay roof 1” (A > 278 and N 6 184) hatched by blue horizontal lines and the “decay roof 2” (292 6 A < 320 and N > 184) hatched by blue diagonal lines. The double black lines show the Z = 82, N = 126 and N = 184 shell closures. See text for more details. 11

III. IMPACT OF FISSION ON THE R-PROCESS in all details, the main reason of which is the not yet NUCLEOSYNTHESIS IN NS MERGERS manageable computational complexity associated with full neutrino transport in a generically three-dimensional (3D), highly asymmetric environment with nearly rela- The NS-NS merger simulation was performed with tivistic fluid velocities and rapid changes in time. It was a general relativistic Smoothed Particle Hydrodynam- long assumed that neutrino interactions could not, at ics scheme [16, 31, 104] representing the fluid by a set least not drastically, affect the initial neutron richness of particles with constant rest mass, the hydrodynamical of the ejecta. However, recent NS-NS merger simulations properties of which were evolved according to Lagrangian [22, 55, 57–61, 63] including the effect of weak interac- hydrodynamics. We only consider dynamical (prompt) tions on free nucleons demonstrate that neutrino reac- ejecta expelled during the binary NS-NS merger in the tions can significantly affect the neutron-to-proton ratio present paper since this ejecta component provides gen- in the merger ejecta, with direct consequences in partic- erally the most favorable conditions for fission recycling ular on the amount of synthesized low-mass (A < 140) due to the low initial electron fraction. A symmetric r-nuclides produced in the dynamical ejecta along with 1.365M – 1.365M binary system compatible with the the heavier species. To broadly study the impact of weak total mass detected in the GW170817 event [36] has been interactions of free nucleons on the r-process nucleosyn- modelled on the basis of the so-called SFHO equation thesis, a parametric approach in terms of prescribed neu- of state [105]. R-process calculations are performed on trino luminosities and mean energies, guided by hydrody- 500 ejected “particles” (i.e. mass elements) representa- namical simulations, was followed in Ref. [56]. A similar tive of the dynamical ejecta. The r-process nucleosyn- approach is also considered here to test the role of fis- thesis is calculated with a reaction network including all sion during the r-process nucleosynthesis in mainly two 5000 species from protons up to Z = 110 lying between scenarios. The first one (hereafter scenario I) neglects all the valley of β-stability and the neutron drip line. All weak interactions on free nucleons and assumes the initial charged-particle fusion reactions on light and medium- electron fraction of the ejected material is consequently mass elements that play a role when the nuclear statisti- not modified at the beginning of the expansion. In this cal equilibrium freezes out are included in addition to ra- case, the mean initial electron fraction at the time of diative neutron captures and photodisintegrations. The the network calculation (i.e. when the temperature has reaction rates on light species are taken from the NET- dropped below 10 GK) amounts to hY i = 0.03. This case GEN library, which includes all the latest compilations e is likely to be relevant for NS-BH mergers or highly asym- of experimentally determined reaction rates [106]. Ex- metric NS-NS mergers, where the lower mass component perimentally unknown reactions are estimated with the develops an extended tidal tail, from which considerable TALYS code [89, 107] on the basis of the Skyrme HFB amounts of cold, unshocked matter can be centrifugally nuclear mass model, HFB-21 [108]. On top of these re- ejected before neutrino exposure of these ejecta may play actions, β-decays as well as β-delayed neutron emission an important role [18, 21, 23]. In the second case (sce- probabilities are also included, the corresponding rates nario II), the equilibrium between electrons, positrons being taken from the relativistic mean-field model of and neutrinos is assumed to hold down to the fiducial Ref. [109]. Fission processes, including neutron-induced, density of ρ = 1012 g cm−3, and from that density on, spontaneous and β-delayed fission, are carefully intro- eq electron, positron and electron neutrino and antineutrino duced in the network together with the corresponding captures on nucleons FFD. Fission barriers obtained with the BSk14 Skyrme interaction are adopted here to estimate fission probabil- − νe + n p + e ities [110, 111]. The new smooth FFDs from SPY model, + (8) as described in the previous section, as well as GEF pre- ν¯e + p n + e dictions for comparison, are used in the present study. are systematically included assuming the (anti)neutrino Note that all fission products with a yield larger than −5 luminosities and mean energies remain constant in time, typically 10 % are linked to the parent fissioning nu- as detailed in Ref. [56], and compatible with the results cleus in the network calculation, i.e. about 500 fission obtained within the Improved Leakage-Equilibration- fragments are included for each fissioning nucleus. The Absorption Scheme (ILEAS) method [63] (see in partic- neutron-rich fission fragments located outside the net- ular their Fig. 14). To do so, we adopt the values of work, i.e. across the neutron drip line, are assumed to 53 53 Lν = 0.3×10 erg/s; Lν¯ = 10 erg/s; hEν i = 8 MeV; instantaneously emit neutrons down to the neutron drip e e e hEν¯ i = 12 MeV. The resulting mean electron fraction at line. The total mass fraction as well as number of nu- e the time of the network calculation reaches hYei = 0.23 cleons is conserved at all time and the emitted neutrons in relatively good agreement with the 0.26 value obtained are recaptured consistently by all the existing species. in Ref. [63]. Finally, α-decays are taken into account for all heavy Without weak interaction, the initial electron fraction, species, the rates being extracted from Ref. [112]. 0 Ye , of the various mass elements is low and characteristic Despite the recent success of nucleosynthesis studies of the NS inner crust, i.e. less than 0.06; consequently the for NS mergers, the details of r-processing in these events number of free neutrons per seed nuclei can reach a few is still affected by a variety of uncertainties. In particu- hundreds. With such a neutron richness, heavy fissioning lar, the exact impact of neutrino is not yet understood nuclei can be efficiently produced during the r-process. 12

Weak interactions on nucleons (Eq. 8) increase the initial electron fraction in the ejecta to values ranging in this scenario II from 0.06 to 0.38 and make the production of heavy fissioning nuclei and fission recycling during the r-process less favorable, though not impossible. On the basis of the BSk14 fission barriers adopted here, the production of super-heavy neutron-rich nuclei is pos- sible up to Z = 110 if the neutron irradiation is large enough to overcome the N = 184 shell closure. Beyond Z = 110, the production of elements is unlikely because these elements fission spontaneously (Fig. 7a) with a very short lifetime (Fig. 7b). This is the main reason why the reaction network has been limited to elements Z 6 110. Nuclei reaching this limit (Z = 110 and A > 320) define a “fission roof” area (A > 320) (Fig. 7d). To produce such super-heavy neutron-rich nuclei, the matter has to flow through the fissioning region FIG. 8: (Color online) Final composition (in terms of 286 formed by (i) β-delayed fission of 94 Pu192,(ii) spon- molar fraction) of matter ejected in scenarios I and II 286 taneous fission of 95 Am191,(iii) β-delayed fission of 288−289 and obtained either with SPY or GEF FFDs. For 97 Bk191−192, and (iv) spontaneous fission of ele- comparison, the solar r-abundance distribution [113] 286 ments with Z > 98 (Fig. 7a). 95 Am191 and Z > 98 (blue dotted circles) is shown and arbitrarily −4 elements have lifetime shorter than 10 s, which makes normalized. neutron capture by these nuclei unlikely. The β-delayed 286 288−289 fission lifetime of 94 Pu192 and 97 Bk191−192 ranges from 10−3 to 10−2 s. The neutron-induced fission of 287−291 285 studied in both scenarios in Sec. III.2. The contributions 98 Cf189−193 and 95 Am190 isotopes also happens to be efficient and contributes to fission recycling (Fig. 7c). of the various fissioning regions to the fission recycling The flow through Pu isotopes is negligible due to the fast is studied in Sec. III.3, while, in Sec. III.4, the impact 278 of the FFDs on the final abundances are studied with a β-delayed neutron emission of 94 Pu184 with a lifetime −4 279 −3 specific comparison made when adopting SPY or GEF τβkn = 6.6×10 s and 94 Pu185 with τβkn = 4.7×10 s (Fig. 7b). In addition, the neutron separation energy FFDs. of 279Pu equals to 0.28 MeV which makes the neutron capture by 278Pu unfavorable. Consequently, matter mainly flows through the Cm isotopic chain to produce III.1. Final abundances super-heavy neutron-rich elements. This region forms a “fission bottleneck” (A < 292 and N > 184) where 287 The final abundance distributions of the ejected mate- 96 Cm191 represents the “crossing point” (Fig. 7d). How- rial are shown in Fig. 8 for both scenarios I and II. Cal- ever, the production of super-heavy neutron-rich nuclei is culations are performed adopting either the SPY or GEF no more possible for temperatures higher than 2 GK (i.e. predictions of the FFDs. While in scenario I, only nuclei kBT = 0.17 MeV) because the neutron-induced fissions with A > 120 are essentially produced, weak interactions become more favorable than the radiative neutron cap- ∼ 285−288 285 on nucleons also allow the production of lighter species tures of Cm189−192 and Bk188 (Fig. 7c) which 96 97 with 80 <∼ A <∼ 120 in scenario II, but also decrease the close the “fission bottleneck”. production of the Pb peak nuclei as well as the long-lived At the end of the neutron irradiation, heavy neutron- actinides. In this case, the choice of the FFD model (SPY rich elements β-decay up to the region of efficient fis- or GEF) is seen to have a negligible impact on the final sion. Elements accumulated during the neutron irradia- abundance distribution. The same conclusions hold if tion along the N = 184 neutron shell closure β-decay up we consider our original version of the SPY FFD [53]. to reach a region of efficient fission. This can be split in Without weak interactions, with the low initial electron two zones : the “decay roof 0” (A < 278) and “decay roof fractions, the number of neutron per seed is rather high 1” (A > 278 and N 6 184) which correspond to the re- (a few hundreds), so that the production of the heavi- gion reached by nuclei Z < 94 and Z > 94, respectively. est nuclei is efficient. In this case, nuclear fission plays a Elements with 292 . A . 320 accumulated after the “fis- fundamental role during both the neutron irradiation and sion bottleneck” β-decay up to reach another region of at freeze-out [32] and the FFD model directly affects the efficient fission named “decay roof 2” (292 6 A < 320). abundance distribution, particularly around the second In the following subsection (Sec. III.1), we present the peak A = 130 where SPY FFDs give higher abundances final abundances of the ejected material obtained in both than GEF (see Sec. III.4). This nucleosynthesis has scenarios as well as their sensitivity to the initial condi- been described in detail in Refs. [17, 18, 31, 44, 56, 62]. tions of the trajectories. The fission contribution to the Note that the corrections introduced in the present study r-process and its impact on the final abundances will be (Sec. II.1) in comparison with our 2019 version [53] have 13

FIG. 9: (Color online) (a-e) Final abundances of the trajectories computed within scenario II classified into 5 groups : “1st peak”, “between 1st and 2ndpeak”, “2nd peak”, “rare earth & no fiss.” and “fiss.”. The green curves with dots are the final abundances averaged over all trajectories computed within scenario II (a-e). The gray curves with dots correspond to the mean final abundances within each group. (f) Final abundances of the trajectories computed within scenario I (black curves) and the final abundances averaged over all trajectories (gray curve with 0 0 0 dots). (g) Distribution of the trajectories in the (S ,Ye ) plane and iso-Xn curves (green curves). (h) Mean value of the initial neutron mass fraction X0 (green curve) and mean mass hAi (gray curve) of each group and the n A>50 associated ejecta mass fraction of groups for the scenario II (color bars). a small impact on the final abundance distribution. They curves) which is consequently a parameter of first rel- tend to slightly increase the second r-abundance peak at evance in the analysis of the efficiency of the r-process 0 A ' 130 and decrease its low-A tail. Such differences are nucleosynthesis. In scenario I, Xn ranges between 0.85 significantly smaller than the one observed between the and 0.98, as seen in Fig. 9g, while in scenario II it varies 0 0 SPY and GEF predictions. between 0.02 and 0.82. The lower Ye , the larger Xn It is well accepted [114] that the efficiency of the r- by definition, but also the higher the initial entropy, the 0 process nucleosynthesis can be related to three main larger Xn due to the release of free neutrons through 0 physical properties of the ejected material, namely the photodissociation. Therefore, Xn engulfs information on electron fraction Ye, the entropy per baryon S and the both the initial electron fraction and entropy. The sud- 0 0 expansion timescale τ [115]. In scenario I, trajectories den variation of Xn observed at S = 8−10 kB/nucleon, 0 independently of Y 0, is linked to the initial abundances of are characterized by a low initial electron fraction Ye e P 0 ranging between 0.02 and 0.06 at the beginning of the trans-alpha elements (A > 4) : A>4 Y (A). The mat- nucleosynthesis, i.e. when the temperature has dropped ter is not completely dissociated for trajectories with a 0 below 10 GK and the density below the drip density low initial entropy S < 10 kB/nucleon, which makes 11 3 0 (ρdrip ' 4 × 10 g/cm ) (Fig. 9g). Initial entropies S the initial abundance of trans-alpha elements (with a 0 reach values up to 170 kB/nucleon, though most of them mean mass hAiA>4 ranging from 50 to 85) non-negligible 0 remain below 50 kB/nucleon. As already mentioned, in and reduces the initial neutron mass fraction. For S > scenario II, the initial electron fraction covers a much 10 kB/nucleon, the initial abundances of trans-alpha el- larger range between 0.06 and 0.38 due to the weak in- ements is negligible, i.e. the ejected nuclear matter is teractions (Eq. 8). completely dissociated into protons, neutrons and alphas. Both the initial electron fraction and entropy affect the The low entropy effect is more important for trajecto- 0 initial mass fraction of free neutrons Xn (Fig. 9g, green ries in scenario II where for 18% of the ejected mass, 14

P 0 A>4 Y (A) > 0.001, whereas in scenario I, only 5% of P 0 the ejecta has A>4 Y (A) > 0.001. Trans-alpha ele- ments can boost the r-process depending on the neutron irradiation. Initial and final abundance distributions are more or less similar for trajectories with a low initial neu- 0 tron mass fraction (Xn < 0.3) and a low initial entropy 0 (S < 10 kB/nucleon); in both cases, the r-process is weak due to the low neutron irradiation. 0 The relevance of the initial neutron mass fraction Xn in the analysis of the efficiency of the r-process can be clearly seen in Figs. 9. The various trajectories can be classified according to their final abundance distributions (Figs. 9a-f). The trajectories computed within scenario I are classified in the group labeled “no ν” while those computed within scenario II are divided into five groups : the group labeled “fission” contains trajectories with the cum highest fission recycling (Xfiss,tot > 0.01, more details about the definition of this fission recycling indicator will be given in Sec. III.2). Among the trajectories with P Y f (A) > 0.01, if P Y f (A) < 0.001 FIG. 10: (Color online) Initial neutron mass fraction 1256A6135 1906A6200 X0 (black curve with dots) and the mass distribution of then they are classified in the group “2nd peak” else in n the ejecta (blue curve) depending on the fission the group “rare earth & no fission”. The trajectories recycling Xcum . The dashed line separates with P Y f (A) > 0.1 are classified in the group fiss,tot 756A685 abundances computed within scenario I (Xcum > 1) “1st peak”. Remaining trajectories correspond to the fiss,tot to the ones computed within scenario II (Xcum < 1). group “between 1st and 2ndpeak”. The two first groups fiss,tot “1st peak” and “between 1st and 2ndpeak” represent each 7% and 6%, respectively, of the ejected mass within sce- nd dXC nario II (see Fig. 9h). The other groups “2 peak”, where dt (Z, A, t, r) corresponds to the reaction flux, “rare earth & no fission” and “fission” represent 41%, expressed in terms of the mass fraction X for a given 30% and 16%, respectively. The mean initial neutron nucleus (Z,A), at a given time t and for a given fission cum mass fraction increases with the group (Fig. 9h, green reaction r. Xfiss,tot consequently corresponds, at a given curve), heavy elements production is possible only with time, to the cumulative mass fraction of fissioning nu- a high initial neutron mass fraction. Similarly, the mean clei that are destroyed by fission reactions and gives a mass of elements heavier than 50 (hAi ) also increases quantitative indicator of the amount of mass recycled by A>50 with the group (Fig. 9h, gray curve) but reaches a maxi- fission processes. Fission can take place by spontaneous mum around 170 when fission occurs in groups “fission” (at a rate λsf ) or β-delayed (at a rate λβdf ) fission or and “no ν” corresponding to the saturation regime in- by neutron-induced fission (at a rate Nnhσvin,f , where duced by the fission recycling (see Sec. III.2). In the hσvin,f is the corresponding astrophysical rate and Nn 0 0 the neutron density). Consequently, there are three fac- (S ,Ye ) plane (Fig. 9g), the trajectories of the various 0 tors contributing to the fission recycling : (i) the quan- groups are distributed according the iso-Xn lines which clearly demonstrate the relevance of the initial neutron tity of fissionable nuclei, (ii) the fission rates and (iii) the mass fraction X0 in the analysis of the r-process effi- amount of time over which fission reactions may occur. n cum ciency. The total fissioning cumulative mass fraction Xfiss,tot 0 increases with initial neutron mass fraction Xn (Fig. 10, black curve with dots). The fission recycling can there- fore be explained in term of initial neutron mass fraction III.2. Fission recycling 0 Xn, i.e. the more neutron-rich the initial ejecta, the larger the production of heavy elements, hence the more Fission recycling for a given trajectory can be quan- efficient the fission. tified by the so-called total cumulative fissioning mass fraction Z III.2.1. Scenario I X X dXC Xcum = (Z, A, t, r)dt (9) fiss,tot dt Z,A r=fiss cum Within scenario I, Xfiss,tot > 1 for all ejected tra- Z X h Z,A Z,A Z,Ai jectories due to the high initial neutron mass fraction = X(Z, A, t) Nnhσvin,f + λsf + λβdf dt 0 (Xn > 0.85). In Fig. 11a, the final abundances of Z,A the 500 trajectories have been sorted and binned as a X cum cum = Xfiss (Z,A) function of Xfiss,tot which is found to range between Z,A 1.19 and 1.82 with a mean value over all trajectories of 15

cum (a) For scenario I where the mean value of the total fissioning cumulative mass fraction hXfiss,totitraj = 1.45.

cum (b) For scenario II where the mean value of the total fissioning cumulative mass fraction hXfiss,totitraj = 0.03.

cum FIG. 11: (Color online) (α) Final abundances obtained with SPY FFDs, averaged for each bin of Xfiss,tot colored by cum their Xfiss,tot value and compared to the average final abundances (green solid curve with dots). (β) Same final cum cum abundances in the (A, Xfiss,tot) plan colored by their abundances. The mean mass hAiA>50 for each Xfiss,tot bins are represented by red crosses.

cum cum hXfiss,totitraj = 1.45. The final abundances distribution A > 50 mass, hAiA>50, does not increase with Xfiss,tot for a given trajectory can be directly linked to its total (Fig. 11aβ, red crosses), and stays rather constant be- cum fissioning cumulative mass fraction Xfiss,tot. An overall tween 166 and 172. This value remains constant be- shift of final abundances towards heavier elements can be cause the final abundances are dominated by the second expected with increasing fission recycling but the mean (A = 130) and third (A = 195) peaks. 16

The final abundance of light elements (A 6 4), mainly fissioning cumulative mass fraction, as shown in Fig. 11b. α-particles, is significant with 0.1 < P Y f (A) < 0.84 The bins, except for the first one, represent each about A64 depending on the trajectory. It increases with the initial 5% to 10% of the ejected mass (Fig. 10). 0 entropy S , inducing an overall downward shift of the The final abundance of the light A 6 4 elements in bins 0 final abundances of heavier elements (A > 4). As Xn are more or less similar ranging from 40% to 55% with cum 0 is linked to Xfiss,tot (Fig. 10), the initial entropy S of no global trend which induces a smaller shift of the final cum trajectories within a given bin of Xfiss,tot covers almost abundances due to trans-alpha elements in comparison 0 0 the whole range of possible S (since Xn isolines evolve in with scenario I. 0 0 0 cum the (S ,Ye ) plane, as seen in Fig. 9g). This S admixture Xfiss,tot is linked to the initial neutron mass fraction of trajectories within a bin of Xcum smooths the overall 0 fiss,tot Xn, as observed in Fig. 10 (black curve with dots) where shift of the final abundances of heavy elements (A > 4). 0  cum  Xn is seen to increase linearly with log10 Xfiss,tot from However, fission recycling has a significant impact : the cum final A < 195 abundances decrease with Xcum , whereas 0.5 to 0.75. In other words, Xfiss,tot increases exponen- fiss,tot 0 cum 0 for elements A > 195 they increase (Fig. 11aα). With tially with Xn. An increase with Xfiss,tot, hence with Xn, cum increasing values of X , the abundances in the mass of the mean mass hAiA>50 from 105 up to 168 is observed fiss,tot 0 region 140 < A < 170 drop by an order of magnitude, in Fig. 11bβ (red crosses), i.e. the higher Xn, the heavier but around A = 202 they increase from 4 × 10−4 up to the final nuclei produced. 2 × 10−3, while for A < 80 elements, they drop below For trajectories with a low initial neutron mass frac- −5 0 cum −7 10 (Fig. 11a). tion, Xn ≈ 0.5 and Xfiss,tot < 10 , A > 130 nuclei, Neglecting the effect related to the variation of light including lanthanides, are not significantly produced in elements abundance, a higher initial neutron mass frac- comparison with the other trajectories (Fig. 11b) and 0 production beyond A > 180 is negligible. Fission recy- tion Xn induces more fission recycling which implies a decrease of the final A < 80 abundances. A high neu- cling does not take place in this case because there are not tron irradiation leads to a significant production of super- enough neutrons to overcome significantly the N = 82 heavy elements (A > 300) whose fission fragments com- neutron shell closure. For higher initial neutron mass 0 −7 cum −5 bined to new neutron captures shift the third peak to fraction Xn ≈ 0.6 (10 < Xfiss,tot < 10 ), nuclei in the higher masses. rare earth and third peak regions start to be produced, Part of the nuclei that accumulated during neutron ir- but those with A > 200 are still highly underproduced radiation along the neutron drip line (including the neu- and fission recycling remains negligible. tron shell closure N = 184) with 208 < A 278 are Abundances of third peak elements increase in the 6 −5 cum −1 0 the progenitors of the trans-lead elements. Among these range 10 < Xfiss,tot < 10 (0.65 < Xn < 0.71) as accumulated nuclei, those with 260 . A 6 278 (79 . does the mean mass hAiA 50 which increases linearly   > Z 6 94) predominantly β-decay up to reach the “decay with log Xcum up to 168. In this Xcum range, roof 0” where they fission. Those with 208 < A < 232 10 fiss,tot fiss,tot abundances of rare earth nuclei (145 < A < 180) in- β- and α-decay up to the third abundance peak, while −4 cum −2 the remaining intermediate nuclei with 232 A 260 crease to reach a maximum at 10 < Xfiss,tot < 10 6 . cum −2 β- and α-decay up to the third abundance peak or β- before decreasing for Xfiss,tot > 10 . The N = 126 and decay and reach the “decay roof 0” where they fission or N = 184 neutron shell closures start to be overcome for 232 235−238 cum −4 decay into the long-lived Th and U nuclei. Fi- Xfiss,tot > 10 leading to the synthesis of A > 204 nu- nal A > 204 abundances decrease with Xcum because clei in a non-negligible amount. Fission recycling starts fiss,tot cum −4 their progenitors are the nuclei accumulated along the to play a role for trajectories with Xfiss,tot > 10 (or 0 neutron shell closure N = 184 during the neutron irra- equivalently with Xn > 0.65), which represent 29% of diation. This accumulation is reduced for high neutron the ejected mass in scenario II. irradiations which enable to overcome the neutron shell In the last bin of total fissioning cumulative mass cum closure N = 184, hence to increase the fission recycling. fraction, i.e. 0.1 < Xfiss,tot < 0.61 and an average 0 initial neutron mass fraction Xn = 0.76, the third r- abundance peak becomes higher than the second one and III.2.2. Scenario II A > 204 nuclei dominate the composition. The mean

mass hAiA>50 does not increase significantly, a satura- 0 tion regime is reached with hAiA 50 ≈ 170 as observed With the low initial neutron mass fraction Xn found in > scenario II, about 55% of the ejected mass is not subject for scenario I (see Sec. III.2.1); the r-process dynamics cum −7 for such trajectories is similar to the one taking place in to fission recycling (Xfiss,tot < 10 ) (Fig. 10, blue curve) cum scenario I. and the maximum value of Xfiss,tot = 0.6. The fission recycling is reduced by a factor 50 with a mean value of To complete the discussion of Sec. III.2.1, fission re- cum cycling plays a dominant role when Xcum > 0.1 cor- the total fissioning cumulative mass fraction hXfiss,totitraj fiss,tot 0 that now amounts to 0.03 to be compared with 1.45 for responding to Xn > 0.75 which concerns 9.5% of the scenario I. The present scenario is well suited to study low ejected mass within scenario II. A saturation regime is

fission recycling effects on the final abundances. For this reached hAiA>50 ≈ 170, when a high initial neutron mass 0 reason, trajectories are sorted in log10 bins of their total fraction Xn induces more fission processes feeding back 17

recycling. The contribution of a given fission seed area (see Fig. 7d defining the 5 areas of relevance for the nu- clear physics adopted in the present study) to the fission cum recycling Xfiss,tot is the sum of xFS(Z,A) for all nuclei located in such a specific fission seed area. Such con- tributions evolve with the fission recycling, as shown in Fig. 12. The contribution of the “decay roof 0”, corre- sponding to the fission of nuclei originally accumulated along the neutron shell closure N = 184 and Z < 94 and reaching the fission efficient region by β-decay after cum the freeze-out, decreases with increasing Xfiss,tot since the fission recycling increases with the initial neutron mass 0 cum 0 fraction Xn. With increasing Xfiss,tot and Xn, matter accumulated along N = 184 isotone spreads up to Fm thanks to the successive β-decays and neutron captures along this shell closure. More nuclei above Z = 94 are produced and increase the contribution of the “decay roof 1” and consequently decrease the one of the “decay cum −6 roof 0”. For Xfiss,tot > 10 , the neutron shell closure FIG. 12: (Color online) Repartition of the fission N = 184 starts to be overcome and nuclei in the “fis- cum recycling Xfiss,tot between the 5 fission seed areas sion bottleneck” area contribute to the fission recycling. −6 cum −1 defined in Fig. 7d. The dashed line separates scenario I In the range 10 < Xfiss,tot < 10 , the contribution cum cum (Xfiss,tot > 1) from scenario II (Xfiss,tot < 1). of the “fission bottleneck” is roughly constant represent- ing 20% of the total cumulative fissioning mass fraction. The matter starts to flow through the “fission bottle- the 80 < A < 200 mass region. neck”, via Cm isotopes, and reach the “fission roof” for cum −4 cum Xfiss,tot > 10 . For high fission recycling (Xfiss,tot > 1 as in scenario I), the total cumulative fissioning mass frac- III.3. Fission seeds tion is dominated by the “fission bottleneck” and “fission roof” areas, representing around 60% and 30%, respec- We now turn to the mean contribution of fissioning tively. The contribution of the “decay roof 2” is low nuclei, or fission seeds, to the fission recycling and to because matter does not pile up in this region due to the the final abundance distribution. The contribution of a absence of shell closure. given fissioning nucleus (Z,A) to the fission recycling, xFS(Z,A), can be defined, for a given set of trajectories, as the fraction of the total fissioning cumulative mass cum III.3.2. Scenario I fraction hXfiss,totitraj due to this specific fissioning nu- cleus (Z,A), i.e. Sixteen fission seeds are found to contribute for more cum hXfiss. (Z,A)itraj than 1% to the fission recycling (Fig. 13a) within sce- xFS(Z,A) = cum . (10) hXfiss,totitraj nario I. They are distributed mainly in the “fission bot- tleneck” and the “fission roof” areas, representing respec- cum In the specific NS merger cases studied here, hXfiss,totitraj tively 57% and 31% of the total cumulative fissioning reaches rather similar values of 1.45 and 1.41 in scenario I mass fraction (Fig. 13a). when using SPY or GEF FFDs, respectively and 0.03 for A major part of the flow crossing the N = 184 both FFDs within scenario II. This similitude is due to shell closure is concentrated along the Am chain, fed by the fact that SPY and GEF FFDs are not fundamentally 278 285 the favorable β-decay of 94 Pu184, where 95 Am190 with different (see Fig. 15) and the properties of the fission xFS = 45% is the major contributor to the fission re- seeds are not related to their FFDs (see Fig. 7). Conse- cycling. This favorable fission is due to the low fission quently, only results with SPY FFDs are presented here. barriers of a few nuclei dropping below 4 MeV accord- ing to the BSk14 Skyrme HFB calculations [111]. They are also main contributors in the “fission roof” area, the III.3.1. Location in the (N,Z) plane of the contributing main one in this area and the second major one of all fission seeds 344 the contributors is 109Mt233 with xFS = 12%. In this case, the primary fission barrier drop below 2 MeV. As Due to the specific β-decay and neutron capture prop- the main fission seeds are in the 275 < A < 290 and erties of nuclei above Th, in particular the “fission bottle- 330 < A < 346 ranges, the FFDs of these fission seeds neck”, the “fission roof” and the N = 184 neutron shell affect the final production of nuclei in the 80 < A < 210 closure, not all the trans-Th nuclei play a role in fission range. 18

(a) For scenario I.

(b) For scenario II.

FIG. 13: (Color online) Fission seed contribution (xFS) depending on the mass of fission seed (the total height). Each bins contains the isotopic distribution of xFS for a given A.

III.3.3. Scenario II occurs mostly after freeze-out in scenario II. The post- freeze-out fission seeds, located in the “decay roof 0 and 1”, represent about 54% of Xcum . As the initial neutron mass fraction is low in this case, fiss,tot the fission recycling remains low too. The contribution of the “fission bottleneck” and the “fission roof” amount to 37% and 6%, respectively, to be compared with 57% III.4. Fission progenitors and 31% in scenario I. The main fission seeds area is the “decay roof 1” (Fig. 13b) because the low neutron irradi- We now turn in the analysis of the so-called fission pro- ation makes it difficult to overcome the N = 184 neutron cum genitors, i.e. the nuclei for which fission contributes in shell closure. It represents 45% of Xfiss,tot and the main a non-negligible way to their final abundance. Through contributors are Md and No isotopes. The contribution a detailed tracking of the fission processes taking place of the “decay roof 0” increases from 2% in scenario I to during the r-process, the impact of fission on the fi- the non-negligible value of 9% in scenario II. nal abundances can be quantified by the analysis of the 285 The main fission seed is still 95 Am190 with xFS = 25% fission fragments progenitors and their related fission with additional ones located in the 276 < ACN < 288 seeds. For each trajectory, the progenitors of a given mass region with xFS = 2 − 10%. The fission recycling fission fragment can be identified and the fission contri- 19

FIG. 14: (Color online) Contribution of the fission fragments to the final production of nuclei with mass number A f cum f cum xFFP(A, Xfiss,tot) obtained with SPY FFDs. Gray curve delimits xFFP(A, Xfiss,tot) = 0.1 and white one f cum xFFP(A, Xfiss,tot) = 0.5.

bution to the production of this final nucleus can be esti- the fission fragments xFFP(Z, A, ti) (FFP stands for fis- mated. More precisely, the contribution stemming from sion fragments progenitors) to a nucleus (Z,A) with the molality Ym(Z, A, ti) (in mol/g) at time ti is defined as

∆Ym,P P xFFP(Z, A, ti−1)Ym(Z, A, ti−1) + ∆t (Z, A, ti−1)∆t r xFFP(Zr,Ar, ti−1)xRP(Z, A, r, ti−1) xFFP(Z, A, ti) = (11) ∆Ym,P Ym(Z, A, ti−1) + ∆t (Z, A, ti−1)∆t

where ∆t = ti − ti−1 is an arbitrary time step taken here production of light nuclei. ∆Ym,P to be of the order of ≈ ti/10 and (Z, A, ti−1) is the The contribution of fission fragments to the final pro- ∆t cum total production rate (in mol/g/s) of nucleus (Z,A) be- duction depends on the fission recycling indicator Xfiss,tot tween ti−1 and ti, expressed in terms of the molality Ym (Fig. 14, gray curve) and starts to be significant, i.e. (if normalized to 1, it corresponds to the molar fraction  f cum  cum max xFFP(A, Xfiss,tot) (Xfiss,tot) > 0.1, Y ). The reaction fraction xRP(Z, A, r, ti−1) is the frac- A tion of the total production rate due to the reaction (r). cum −4 0 for Xfiss,tot > 10 (Xn > 0.65) which corresponds to the The reactions considered here are (n, γ), (γ, n), (β, γ), cum −2 (β, n), (β, 2n), (β, 3n), (γ, α), (sf) (i.e. spontaneous fis- value given in Sec. III.2.2. For Xfiss,tot > 10 , fission sion), (β, f) and (n, f). In other words, the contribution contributes at least to 50% of the production of the final A nuclei with 108 < A < 120 and 139 < A < 155. For of fission fragments to the production of a nucleus Z X a high fission recycling (Xcum ≈ 1), this contribution at time ti is the weighted average between the previ- fiss,tot ous contribution xFFP(Z, A, ti−1) and the contributions can reach nearly 100% for the elements with A ≈ 110. xFFP(Zr,Ar, ti−1) due to the production of this element In addition to the strong sensitivity to the fission recy- f Ar A cling, x (A) is obviously affected by the FFDs model by a given reaction r : Y → Z X + ..., between ti−1 FFP Zr adopted, even if the fission recycling does not depend and t , at a rate of ∆Ym,P (Z, A, t )x (Z, A, r, t ). If i ∆t i−1 RP i−1 significantly on the FFDs model. This sensitivity to the a nucleus is exclusively produced by fission, x = 1. FFP FFDs model is studied for both scenario in the two sub- This quantity is well suited to quantify the impact of sections below. In order to clarify the link between xFS fission fragments on final abundances because it is not f and xFFP, we also introduce here two new quantities, affected by the decay of post-neutron fission fragments namely the relative contribution of each fission seed to thanks to the time tracking. To obtain the overall fi- the production of a final nucleus A by fission f nal xFFP, hereafter xFFP, (Figs. 16, red dotted curves), f the tracking (Eq. 11) is performed up to t = 1 yr when f xFFP(A, AFS) xFS/FFP(A, AFS) = f (12) the residual fission does not contribute anymore to the xFFP(A) 20

f where xFFP(A, AFS) is obtained by tracking only fis- (see Fig. 15a). Those nuclei are characterized by an sion fragments produced by the specific fission seed AFS, admixture of slightly asymmetric fission and a triple- xRP(Z, A, ti−1) being replaced by xRP(Z, A, AFS, ti−1) in peak fission where the maximum yields are found around Eq. 11. and the contribution of a compound nucleus A = 130 for SPY FFDs. GEF FFD is more asymmet- (ZCN,ACN) to the fission recycling, which can be ex- ric leading to post-freeze-out mean fission yields roughly pressed from Eq. 10 as constant in the A = 110 − 160 mass range. During the neutron irradiation, fission fragments cap- X xCN(ZCN,ACN) = xFS(Zr,Ar, r) tures neutrons. The A < 130 fission fragments accumu-

(Z,A)CN,r={sf,βdf,nif} late along the neutron shell closure N = 82 and the rel- f (13) ative contribution from the “fission bottleneck” xFS/FFP where xFS(Zr,Ar, r) is the contribution to the fission re- increases at A = 120 − 130. The fission fragments pro- cum cycling hXfiss,totitraj of the fission seed (Zr,Ar) fission- duced above N = 82 (A & 130) spread out up to N = 126 ing by a given fission mode r (spontaneous, β-delayed (A . 195). For A > 160, the fission contribution to the fi- f or neutron induced). In Eq. 13, the summation is per- nal production xFFP and the relative contribution of the f formed in such a way that the mass of the compound “fission bottleneck” xFS/FFP is higher with GEF than nucleus formed from the fission seed is equal to ACN and with SPY since the pre-freeze-out distribution is more 246 the proton number to ZCN. asymmetric due to the more asymmetric FFD of Am. At freeze-out, there is no more fission fragments produced by the “fission bottleneck” and the “fission roof”. The III.4.1. Scenario I final abundance peak A = 130 predicted by SPY model (Fig. 8) is due to the high post-freeze-out mean fission yields around A = 130 (Fig. 16a, orange curves). In scenario I, fission processes contribute significantly f to the production of nuclei with 80 < A < 200 for both xFFP is locally reduced around A = 130 due to the FFDs models (Fig. 16a, red dotted curves) due to the accumulation of non-recycled material along the neutron f shell closure N = 82. This reduction is limited with SPY high fission recycling. The shape of xFFP(A) is affected by the FFDs model (Figs. 16aα and γ, red dotted curves). model since the pre- and post-freeze-out mean fission f yields are higher than in the GEF case around A = 130. However the link between xFFP and FFDs is no straight- f forward. A low xFFP for 80 < A < 100 could be expected since the mean fission yields are negligible for both FFD models. Two mean fission yields distributions hYFFiFS can be defined, the first one is associated with the “fission bot- However, in this case, there is no matter accumulation in this mass region, so that the low 80 < A < 100 produc- tleneck” and the “fission roof”, denoted as pre-freeze-out f −4 mean fission yields, corresponds to the mean fission yields tion (Y < 10 ) occurring after freeze-out is essentially due to the asymmetric fission of seeds located in the “de- distribution during the neutron irradiation, before freeze- f cay roof 0”. This leads to a non-negligible value of xFFP out (Figs. 16, green curves). The second one, denoted as f post-freeze-out mean fission yields, is associated with the for 80 < A < 100. The drop of xFFP for A > 180 is due “decay roofs” (Figs. 16, orange curves), after freeze-out. to the accumulation of matter along the neutron shell closure N = 126. For both FFDs model, the pre-freeze-out mean fission 286 yields is dominated by the FFDs of Am (xCN = 0.45) 345 and Mt (xCN = 0.12) formed by the neutron capture of 285Am and 344Mt. With SPY FFDs (see Figs. 15, green III.4.2. Scenario II curves), these two main contributors fission asymmetri- cally. The three peaks pattern seen in the pre-freeze- In scenario II, the reduced fission recycling leads to f out hYFFiFS (Fig. 16aβ) is due to (i) the light fragments a non-negligible xFFP that does not exceed 30% only of 286Am feeding the peak around A = 120 − 130, (ii) around 130 < A < 185, irrespective of the FFDs model the heavy fragments of 286Am plus the light fragments adopted (Fig. 16b, red dotted curves). of 345Mt feeding the peak around A = 145 − 160, and For both FFDs models, the post-freeze-out mean fis- (iii) the heavy fragments of 346Mt responsible for the sion yields is roughly similar than the one within sce- peak A = 180 − 190. In contrast, the GEF model (see nario I since the main contributors from the “decay roof” Figs. 15, blue curves) predicts a more asymmetric fission regions are the same in both cases (see Fig. 15a) where 286 346 282 of Am but a more symmetric one for Mt. As a xCN reaches values up to 0.08 for No. The fission result, a double hump pattern is obtained for the pre- seed contribution from the “fission roof” is small, so that freeze-out hYFFiFS (Fig. 16aδ) where the peak around the A = 180 − 190 pre-freeze-out mean fission yields are A = 100 − 120 is due to to the light fragments of 286Am negligible with the SPY model. The GEF pre-freeze-out and the one around A = 150−180 to the heavy fragments mean fission yields is roughly similar to the one found in of 286Am plus the symmetric fragments of 346Mt. scenario I, though the A = 100 − 120 peak is now found The main contributors to the post-freeze-out mean fis- higher than the A = 160 − 180 one. sion yields, it i.e. to the fission recycling in the “decay The low neutron irradiation makes the contribution roofs” areas, are isotopes from Fm to Lr with ACN = 278 of the “fission bottleneck” and the “fission roof” rel- 283 to 284 where xCN reaches values up to 0.016 for No atively small within this scenario, leading to a total 21

(a) Around the neutron shell closure N = 184, these nuclei are responsible for 64/86% of the fission recycling in scenario I/II.

(b) Around the “fission roof”, these nuclei are responsible for 30/5% of the fission recycling in scenario I/II.

FIG. 15: (Color online) Post-neutron FFDs from SPY model computed at a constant excitation energy Q = 8 MeV (green curves) and from GEF model for thermal neutron-induced fission (blue curves). For each box, the x axis ranges from A = 80 to 210 by step of 10 and the y axis from 0 to 15% by step of 2%. Colored background correspond to the contribution of the compound nucleus xCN in scenario I/II. Only contributions xCN > 0.001 are displayed. FFDs in thick orange boxes correspond to the NCN = 184 neutron shell closure. 22

(a) For scenario I. (b) For scenario II.

f FIG. 16: (Color online) (α and γ) Contribution of the fission fragments to the final production (xFFP) (red dotted f curves) and fission seeds contributions to the production of fission fragment progenitors (xFS/FFP) (color scale) computed with SPY(α) or GEF (γ) FFDs . (β and δ) Mean fission yields hYFFiFS for SPY (β) or GEF model (δ) averaged over fission seeds from the “fission bottleneck” and the “fission roof” (green curves, denoted as pre-freeze-out) or from “decay roofs” (orange curves, denoted as post-freeze-out).

cum xFS < 0.25 for 90% the ejected mass (i.e. Xfiss,tot < 0.1) cation of the scission distance and an updated smoothing (Fig. 12). The role played by the pre-freeze-out mean procedure of the distribution to take into account effects fission yields is consequently reduced. It remains more that are neglected in the original model. Such correc- difficult to overcome the neutron shell closure N = 82 tions lead to a better agreement with experimental fission and more non-recycled material accumulates along this yields. In particular, SPY FFDs are now in overall good shell closure. For this reason, the final abundance around agreement with the experimental symmetric/asymmetric A = 130 is higher than in scenario I with a negligible transition in the Th and Pa isotopic chains. Those yields fission contribution to the A 6 130 abundances. As the at scission are also used to estimate the number of neu- SPY post-freeze-out mean fission yields is peaked around trons emitted by the excited fragments on the basis of f A = 120−160, xFFP is maximal for A = 135−160. Com- different neutron evaporation models. Our new fission pared to SPY, GEF pre- and post-freeze-out mean fission yields significantly differ from to those predicted by the f widely used GEF model. yields are higher for A & 160 which leads to a higher xFFP in this mass region. The impact of fission on the r-process nucleosynthesis in neutron mergers is also re-analyzed. Two scenarios are considered, the first one with low initial electron fraction IV. CONCLUSION is subject to intense fission recycling, in contrast to the second one which includes weak interactions on free nu- In the present study, new improvements have been cleons resulting in less pronounced fission recycling. The brought to our scission-point model, called SPY, to de- efficiency of the r-process nucleosynthesis and the corre- rive the fission fragment distribution for all neutron-rich sponding abundance distribution in a given mass element fissioning nuclei of relevance in r-process calculations. has been discussed essentially in terms of the initial neu- 0 These improvements include a phenomenological modifi- tron mass fraction Xn which engulfs information on both 23 the initial electron fraction and entropy. The contribu- jor nuclear as well as astrophysical uncertainties still need tion fission processes may have to the final abundance dis- to be further addressed before concluding on the definite tribution has been studied in detail in the light of newly role of fission processes. These concern in particular the defined quantitative indicators, namely indicators for the determination of the fission probabilities for all the nu- cum fission recycling (Xfiss,tot), fission seeds (xFS) and fission clei potentially involved during the r-process as well as f a detailed description of neutrino absorption and hydro- fragment progenitors (xFFP). The various regions of the nuclear chart responsible for fission recycling during the dynamics in simulations. Much progress remains to be neutron irradiation as well as after freeze-out have been done along these two directions. identified for the nuclear physics adopted in the present calculations, what we defined as the “decay roofs”, the “fission bottleneck” and the “fission roof”. Within the ACKNOWLEDGMENTS “fission bottleneck” region, Am isotopes around A = 285 play a key role by limiting the nuclear flow to higher J.-F.L. and S.G. acknowledge financial support from elements and strongly contributing to the fission recy- FNRS (Belgium). This work was supported by the Fonds cling, mainly due to the low fission barriers predicted by de la Recherche Scientifique (F.R.S.-FNRS) and the the BSk14 interaction in this region. Above Z = 110, the Fonds Wetenschappelijk Onderzoek - Vlaanderen (FWO) so-called “fission roof” prohibits the production of super- under the EOS Project nr O022818F. The present re- heavy elements. “Decay roofs” essentially recycle heavy search benefited from computational resources made material after the neutron irradiation freezes out when available on the Tier-1 supercomputer of the F´ed´eration the material accumulated at the N = 184 shell closure Wallonie-Bruxelles, infrastructure funded by the Wal- decays back to the valley of β-stability. loon Region under the grant agreement nr 1117545. H.-T.J. acknowledges financial support from Deutsche When considering an efficient r-process (scenario I), Forschungsgemeinschaft (DFG, German Research Foun- fission fragments are found to contribute almost entirely dation) through Sonderforschungsbereich (Collaborative to the final abundances of nuclei with 100 <∼ A <∼ 180, Research Center) SFB-1258 “Neutrinos and Dark Mat- while in scenario II where weak interactions on nucleons ter in Astro- and Particle Physics (NDM)” and under initially limit the number of free neutrons, such a contri- Germany’s Excellence Strategy through Cluster of Excel- bution does not reach more than 30% and is concentrated lence ORIGINS (EXC-2094)—390783311, and from the in the A = 140−180 region depending on the FFD model European Research Council through Grant ERC-AdG adopted. Calculations based on the SPY and GEF FFDs No. 341157-COCO2CASA. A.B. acknowledges support have been compared for both r-process scenarios. by the European Research Council (ERC) under the Eu- ropean Union’s Horizon 2020 research and innovation The present study allows us to have a deep insight on programme under grant agreement No. 759253, and the role of fission processes in hydrodynamical r-process by Deutsche Forschungsgemeinschaft (DFG, German Re- simulations on the basis of sound well-defined quantita- search Foundation) - Project-ID 279384907 - SFB 1245 tive indicators. Fission remains an important nuclear and - Project-ID 138713538 - SFB 881 (“The Milky Way process taking place in binary NS mergers, though ma- System”, subproject A10).

[1] M. Arnould, S. Goriely, and K. Takahashi, Phys. Repts. [9] D. Siegel, J. Barnes, and B. Metzger, Nature 569, 241 450, 97 (2007). (2019). [2] M. Arnould and S. Goriely, Prog. Part. Nucl. Phys. 112, [10] C. Winteler, R. K¨appeli, A. Perego, A. Arcones, N. Vas- 103766 (2020). set, N. Nishimura, M. Liebend¨orfer, and F.-K. Thiele- [3] J. J. Cowan, C. Sneden, J. E. Lawler, A. Aprahamian, mann, The Astrophysical Journal 750, L22 (2012). M. Wiescher, K. Langanke, G. Mart´ınez-Pinedo, and [11] C. Freiburghaus, S. Rosswog, and F. K. Thielemann, F.-K. Thielemann, Rev. Mod. Phys. accepted (2020). Astrophys. J. Lett. 525, L121 (1999). [4] S. Wanajo, H.-T. Janka, and B. M¨uller, Astrophys. J. [12] M. Ruffert and H.-T. Janka, Astronomy and Astro- 726, L15 (2011). physics 380, 544 (2001). [5] H.-T. Janka, Ann. Rev. of Nucl. Part. Sci. 62, 407 [13] H.-T. Janka, T. Eberl, M. Ruffert, and C. Fryer, As- (2012). trophys. J. Lett. 527, L39 (1999). [6] L. H¨udepohl, B. M¨uller,H.-T. Janka, A. Marek, and [14] O. Korobkin, S. Rosswog, A. Arcones, and C. Winteler, G. G. Raffelt, Phys. Rev. Lett. 104, 251101 (2010). MNRAS 26, 1940 (2012). [7] L. F. Roberts, S. E. Woosley, and R. D. Hoffman, The [15] L. Roberts, D. Kasen, W. Lee, and E. Ramirez-Ruiz, Astrophysical Journal 722, 954 (2010). Astrophys. J. Lett. 736, L21 (2011). [8] T. Fischer, S. C. Whitehouse, A. Mezzacappa, F. K. [16] R. Oechslin, H.-T. Janka, and A. Marek, Astronomy Thielemann, and M. Liebend¨orfer, Astronomy and and Astrophysics 467, 395 (2007). Astrophysics 517, A80 (2010), arXiv:0908.1871 [astro- [17] S. Goriely, A. Bauswein, and H.-T. Janka, Astrophys. ph.HE]. J. Lett. 738, L32 (2011). 24

[18] O. Just, A. Bauswein, R. Ardevol Pulpillo, S. Goriely, [45] M. Eichler, A. Arcones, A. Kelic, O. Korobkin, K. Lan- and H.-T. Janka, MNRAS 448, 541 (2015). ganke, T. Marketin, G. Martinez-Pinedo, I. Panov, [19] K. Kyutoku, K. Ioka, and M. Shibata, Phys. Rev. D T. Rauscher, S. Rosswog, C. Winteler, N. T. Zinner, 88, 041503(R) (2013). and F.-K. Thielemann, The Astrophysical Journal 808, [20] K. Hotokezaka, K. Kyutoku, M. Tanaka, K. Kiuchi, 30 (2015). Y. Sekiguchi, M. Shibata, and S. Wanajo, Astrophys. [46] J. d. J. Mendoza-Temis, M.-R. Wu, K. Langanke, J. Lett. 778, L16 (2013). G. Mart´ınez-Pinedo, A. Bauswein, and H.-T. Janka, [21] A. Bauswein, R. Ardevol Pulpillo, H.-T. Janka, and Phys. Rev. C 92, 055805 (2015), arXiv:1409.6135 [astro- S. Goriely, Astrophysical J. Lett 773, L9 (2014). ph.HE]. [22] S. Wanajo, Y. Sekiguchi, N. Nishimura, K. Kiuchi, [47] S. A. Giuliani, G. Mart´ınez-Pinedo, and L. M. Robledo, K. Kyutoku, and M. Shibata, Astrophys. J. Lett 789, Phys. Rev. C 97, 034323 (2018), arXiv:1704.00554 [nucl- L39 (2014). th]. [23] F. Foucart, M. B. Deaton, M. D. Duez, E. O’Connor, [48] M. R. Mumpower, T. Kawano, T. M. Sprouse, N. Vassh, C. D. Ott, R. Haas, L. E. Kidder, H. P. Pfeiffer, M. A. E. M. Holmbeck, R. Surman, and P. M¨oller,The As- Scheel, and B. Szilagyi, Phys. Rev. D 90, 024026 (2014). trophysical Journal 869, 14 (2018). [24] T. Baumgarte, S. Shapiro, and M. Shibata, Astrophys. [49] Y. Zhu, R. T. Wollaeger, N. Vassh, R. Surman, T. M. J. 528, L29 (2000). Sprouse, M. R. Mumpower, P. M¨oller,G. C. McLaugh- [25] M. Ruffert and H.-T. Janka, Astronomy and Astro- lin, O. Korobkin, T. Kawano, P. J. Jaffke, E. M. Holm- physics 344, 573 (1999). beck, C. L. Fryer, W. P. Even, A. J. Couture, and [26] L. Dessart, C. Ott, A. Burrows, S. Rosswog, and J. Barnes, The Astrophysical Journal 863, L23 (2018). E. Livne, Astrophys. J. 690, 1681 (2009). [50] E. M. Holmbeck, T. M. Sprouse, M. R. Mumpower, [27] R. Fernandez and B. Metzger, MNRAS 435, 502 (2013). N. Vassh, R. Surman, T. C. Beers, and T. Kawano, [28] D. Siegel, R. Ciolfi, and L. Rezzolla, Astrophys. J. Lett. The Astrophysical Journal 870, 23 (2018). 785, L6 (2014). [51] M.-R. Wu, J. Barnes, G. Mart´ınez-Pinedo, and [29] B. Metzger and R. Fernandez, MNRAS 441, 3444 B. D. Metzger, Phys. Rev. Lett. 122, 062701 (2019), (2014). arXiv:1808.10459 [astro-ph.HE]. [30] A. Perego, S. Rosswog, R. Cabezon, O. Korobkin, [52] N. Vassh, R. Vogt, R. Surman, J. Randrup, T. M. R. Kappeli, A. Arcones, and M. Liebend¨orfer,MNRAS Sprouse, M. R. Mumpower, P. Jaffke, D. Shaw, E. M. 443, 3134 (2014). Holmbeck, Y. Zhu, and G. C. McLaughlin, Journal [31] A. Bauswein, S. Goriely, and H.-T. Janka, Astrophysi- of Physics G: Nuclear and Particle Physics 46, 065202 cal J. 773, 78 (2013). (2019). [32] S. Goriely, The European Physical Journal A 51, 22 [53] J.-F. Lemaˆıtre,S. Goriely, S. Hilaire, and J.-L. Sida, (2015). Phys. Rev. C 99, 034612 (2019). [33] M. Dominik, K. Belczynski, C. Fryer, D. Holz, E. Berti, [54] K.-H. Schmidt, B. Jurado, C. Amouroux, and T. Bulik, I. Mandel, and R. O’Shaughnessy, Astrophys. C. Schmitt, Nuclear Data Sheets 131, 107 (2016), spe- J. 759, 52 (2012). cial Issue on Nuclear Reaction Data. [34] Y. Z. Qian, Astrophys. J. Lett. 534, L67 (2000), [55] Y. Sekiguchi, K. Kiuchi, K. Kyutoku, and M. Shi- arXiv:astro-ph/0003242 [astro-ph]. bata, Phys. Rev. D 91, 064059 (2015), arXiv:1502.06660 [35] B. D. Metzger, G. Mart´ınez-Pinedo, S. Darbha, [astro-ph.HE]. E. Quataert, A. Arcones, D. Kasen, R. Thomas, P. Nu- [56] S. Goriely, A. Bauswein, O. Just, E. Pllumbi, and H.-T. gent, I. V. Panov, and N. T. Zinner, MNRAS 406, 2650 Janka, MNRAS 452, 3894 (2015). (2010), arXiv:1001.5029 [astro-ph.HE]. [57] F. Foucart, R. Haas, M. D. Duez, E. O’Connor, C. D. [36] B. Abbott, R. Abbott, T. Abbott, et al., Phys. Rev. Ott, L. Roberts, L. E. Kidder, J. Lippuner, H. P. Pfeif- Lett. 119, 161101 (2017). fer, and M. A. Scheel, Phys. Rev. D 93, 044019 (2016), [37] B. Abbott, R. Abbott, T. Abbott, et al., Astrophys. J. arXiv:1510.06398 [astro-ph.HE]. Lett. 848, L12 (2017). [58] L. Lehner, S. L. Liebling, C. Palenzuela, O. L. Ca- [38] P. S. Cowperthwaite, E. Berger, V. A. Villar, B. D. ballero, E. O’Connor, M. Anderson, and D. Neilsen, Metzger, M. Nicholl, et al., Astrophys. J. Lett. 848, Classical and Quantum Gravity 33, 184002 (2016), L17 (2017). arXiv:1603.00501 [gr-qc]. [39] M. Nicholl, E. Berger, D. Kasen, B. D. Metzger, J. Elias, [59] D. Radice, F. Galeazzi, J. Lippuner, L. F. Roberts, et al., Astrophys. J. Lett. 848, L18 (2017). C. D. Ott, and L. Rezzolla, MNRAS 460, 3255 (2016), [40] R. Chornock, E. Berger, D. Kasen, P. S. Cowperthwaite, arXiv:1601.02426 [astro-ph.HE]. M. Nicholl, et al., Astrophys. J. Lett. 848, L19 (2017). [60] L. Bovard, D. Martin, F. Guercilena, A. Arcones, [41] E. Waxman, E. O. Ofek, D. Kushnir, and A. Gal-Yam, L. Rezzolla, and O. Korobkin, Phys. Rev. D 96, 124005 MNRAS 481, 3423 (2018). (2017), arXiv:1709.09630 [gr-qc]. [42] D. Watson, C. J. Hansen, J. Selsing, A. Koch, D. B. [61] D. Martin, A. Perego, W. Kastaun, and A. Arcones, Malesani, A. C. Andersen, J. P. U. Fynbo, A. Ar- Classical and Quantum Gravity 35, 034001 (2018), cones, A. Bauswein, S. Covino, A. Grado, K. E. Heintz, arXiv:1710.04900 [astro-ph.HE]. L. Hunt, C. Kouveliotou, G. Leloudas, A. J. Levan, [62] S. Goriely, Il Nuovo Cimento C 42, 115 (2019). P. Mazzali, and E. Pian, Nature 574, 497 (2019). [63] R. Ardevol-Pulpillo, H.-T. Janka, O. Just, and [43] I. Y. Korneev and I. V. Panov, Astronomy Letters 37, A. Bauswein, MNRAS 485, 4754 (2019). 864 (2011). [64] J.-F. Lemaˆıtre, S. Panebianco, J.-L. Sida, S. Hilaire, [44] S. Goriely, J.-L. Sida, J.-F. Lemaˆıtre, S. Panebianco, and S. Heinrich, Phys. Rev. C 92, 034617 (2015). N. Dubray, S. Hilaire, A. Bauswein, and H.-T. Janka, [65] S. Goriely, N. Chamel, and J. M. Pearson, Phys. Rev. Phys. Rev. Lett. 111, 242502 (2013). C 88, 061302(R) (2013). 25

[66] P. Decowski, W. Grochulski, A. Marcinkowski, K. Si- [89] A. Koning and D. Rochman, Nuclear Data Sheets 113, wek, and Z. Wilhelmi, Nucl. Phys. A 110, 129 (1968). 2841 (2012), special Issue on Nuclear Reaction Data. [67] L. Moretto, Physics Letters B 40, 1 (1972). [90] K.-H. Schmidt, S. Steinh¨auser, C. B¨ockstiegel, [68] L. Moretto, Nucl. Phys. A 185, 145 (1972). A. Grewe, A. Heinz, A. Junghans, J. Benlliure, H.-G. [69] P. Demetriou and S. Goriely, Nucl. Phys. A 695, 95 Clerc, M. de Jong, J. M¨uller,M. Pf¨utzner,and B. Voss, (2001). Nuclear Physics A 665, 221 (2000). [70] D. Madland and J. Nix, Nucl. Phys. A 476, 1 (1988). [91] H. Baba, T. Saito, N. Takahashi, A. Yokoyama, [71] C. Schmitt, A. Guessous, J. Bocquet, H.-G. Clerc, T. Miyauchi, S. Mori, D. Yano, T. Hakoda, R. Brissot, D. Engelhardt, H. Faust, F. G¨onnenwein, K. Takamiya, K. Nakanishi, and Y. Nakagome, Journal M. Mutterer, H. Nifenecker, J. Pannicke, C. Ristori, of Nuclear Science and Technology 34, 871 (1997). and J. Theobald, Nuclear Physics A 430, 21 (1984). [92] D. Ramos, M. Caama˜no,F. Farget, C. Rodr´ıguez-Tajes, [72] C. Romano, Y. Danon, R. Block, J. Thompson, L. Audouin, J. Benlliure, E. Casarejos, E. Clement, E. Blain, and E. Bond, Phys. Rev. C 81, 014607 (2010). D. Cortina, O. Delaune, X. Derkx, A. Dijon, D. Dor´e, [73] G. Mariolopoulos, C. Hamelin, J. Blachot, J. Bocquet, B. Fern´andez-Dom´ınguez, G. de France, A. Heinz, R. Brissot, J. Cran¸con,H. Nifenecker, and C. Ristori, B. Jacquot, A. Navin, C. Paradela, M. Rejmund, Nuclear Physics A 361, 213 (1981). T. Roger, M.-D. Salsac, and C. Schmitt, Phys. Rev. [74] C. Tsuchiya, Y. Nakagome, H. Yamana, H. Moroyama, C 97, 054612 (2018). K. Nishio, I. Kanno, K. Shin, and I. Kimura, Journal [93] F.-J. Hambsch, F. Viv`es,P. Siegler, and S. Oberstedt, of Nuclear Science and Technology 37, 941 (2000). Nuclear Physics A 679, 3 (2000). [75] J. Bocquet and R. Brissot, Nuclear Physics A 502, 213 [94] H. Thierens, A. De Clercq, E. Jacobs, D. De Frenne, (1989). P. D’hondt, P. De Gelder, and A. J. Deruytter, Phys. [76] S. Zeynalov, F.-J. Hambsch, and S. Obertstedt, J. Ko- Rev. C 23, 2104 (1981). rean Phy. Soc. 59, 1396 (2011). [95] V. Rubchenya, W. Trzaska, I. Itkis, M. Itkis, J. Kli- [77] Vorobyev, A.S., Shcherbakov, O.A., Gagarski, A.M., man, G. Kniajeva, N. Kondratiev, E. Kozulin, L. Krupa, Val’ski, G.V., and Petrov, G.A., EPJ Web of Confer- I. Pokrovski, V. Voskressenski, F. Hanappe, T. Materna, ences 8, 03004 (2010). O. Dorvaux, L. Stuttge, G. Chubarian, S. Khlebnikov, [78] K. Nishio, Y. Nakagome, I. Kanno, and I. Kimura, D. Vakhtin, and V. Lyapin, Nuclear Physics A 734, Journal of Nuclear Science and Technology 32, 404 253 (2004). (1995). [96] A. Goverdovskij, V. Mitrofanov, and V. Khryachkov, [79] V. Dushin, F.-J. Hambsch, V. Jakovlev, V. Kalinin, Yadernaya Fizika 58, 1948 (1995). I. Kraev, A. Laptev, D. Nikolaev, B. Petrov, G. Petrov, [97] Y. L. Zhao, I. Nishinaka, Y. Nagame, K. Tsukada, V. Petrova, Y. Pleva, O. Shcherbakov, V. Shpakov, K. Sueki, S. Goto, M. Tanikawa, and H. Nakahara, V. Sokolov, A. Vorobyev, and T. Zavarukhina, Nuclear Journal of Radioanalytical and Nuclear Chemistry 255, Instruments and Methods in Physics Research Section 67 (2003). A: Accelerators, Spectrometers, Detectors and Associ- [98] J. Unik, J. Gindler, L. Glendenin, K. Flyn, A. Gorski, ated Equipment 516, 539 (2004). and R. Sjoblom, Proc. Symp. on physics and chemistry [80] O. Litaize, O. Serot, and L. Berge, The European Phys- of fission, Rochester 2, 19 (1974). ical Journal A 51, 177 (2015). [99] D. C. Hoffman, J. B. Wilhelmy, J. Weber, W. R. [81] P. Talou, B. Becker, T. Kawano, M. B. Chadwick, and Daniels, E. K. Hulet, R. W. Lougheed, J. H. Landrum, Y. Danon, Phys. Rev. C 83, 064612 (2011). J. F. Wild, and R. J. Dupzyk, Phys. Rev. C 21, 972 [82] R. Vogt and J. Randrup, Physics Procedia 47, 82 (1980). (2013), scientific Workshop on Nuclear Fission Dynam- [100] D. Hoffman, D. Lee, A. Ghiorso, M. Nurmia, and ics and the Emission of Prompt Neutrons and Gamma K. Aleklett, Phys. Rev. C 22, 1581 (1980). Rays, Biarritz, France, 28-30 November 2012. [101] T. M. Hamilton, K. E. Gregorich, D. M. Lee, K. R. [83] M. Chadwick, P. Obloˇzinsk´y,M. Herman, N. Greene, Czerwinski, N. J. Hannink, C. D. Kacher, B. Kadkho- R. McKnight, D. Smith, P. Young, R. MacFarlane, dayan, S. A. Kreek, M. J. Nurmia, M. R. Lane, M. P. G. Hale, S. Frankle, A. Kahler, T. Kawano, R. Little, Neu, A. T¨urler, and D. C. Hoffman, Phys. Rev. C 46, D. Madland, P. Moller, R. Mosteller, P. Page, P. Talou, 1873 (1992). H. Trellue, M. White, W. Wilson, R. Arcilla, C. Dun- [102] M. R. Lane, K. E. Gregorich, D. M. Lee, M. F. Mohar, ford, S. Mughabghab, B. Pritychenko, D. Rochman, M. Hsu, C. D. Kacher, B. Kadkhodayan, M. P. Neu, A. Sonzogni, C. Lubitz, T. Trumbull, J. Weinman, N. J. Stoyer, E. R. Sylwester, J. C. Yang, and D. C. D. Brown, D. Cullen, D. Heinrichs, D. McNabb, H. Der- Hoffman, Phys. Rev. C 53, 2893 (1996). rien, M. Dunn, N. Larson, L. Leal, A. Carlson, R. Block, [103] http://www.khs-erzhausen.de/GEF.html. J. Briggs, E. Cheng, H. Huria, M. Zerkle, K. Kozier, [104] A. Bauswein, H.-T. Janka, and R. Oechslin, Phys. Rev. A. Courcelle, V. Pronyaev, and S. van der Marck, Nu- D 82, 084043 (2010). clear Data Sheets 107, 2931 (2006), evaluated Nuclear [105] A. W. Steiner, M. Hempel, and T. Fischer, The Astro- Data File ENDF/B-VII.0. physical Journal 774, 17 (2013). [84] T. Kawano, P. Talou, I. Stetcu, and M. Chadwick, Nu- [106] Y. Xu, S. Goriely, A. Jorissen, G. Chen, and clear Physics A 913, 51 (2013). M. Arnould, A&A 549, 10 (2013), A106. [85] V. Weisskopf, Phys. Rev. 52, 295 (1937). [107] S. Goriely, S. Hilaire, and A. J. Koning, Astronomy [86] V. F. Weisskopf and D. H. Ewing, Phys. Rev. 57, 472 and Astrophysics 487, 767 (2008). (1940). [108] S. Goriely, N. Chamel, and J. M. Pearson, Phys. Rev. [87] H. A. Bethe, Rev. Mod. Phys. 9, 69 (1937). C 82, 035804 (2010). [88] J. T¯oke and W. Swiatecki,´ Nuclear Physics A 372, 141 [109] T. Marketin, L. Huther, and G. Martinez-Pinedo, Phys. (1981). Rev. C 93, 025805 (2016). 26

[110] S. Goriely, M. Samyn, and J. M. Pearson, Phys. Rev. [113] S. Goriely, Astronomy and Astrophysics 342, 881 C 75, 064312 (2007). (1999). [111] S. Goriely, S. Hilaire, A. J. Koning, M. Sin, and [114] R. Hoffman, S. Woosley, and Y.-Z. Qian, Astrophys. J R. Capote, Phys. Rev. C 79, 024612 (2009). 482, 951 (1997). [112] H. Koura, T. Tachibana, and T. Yoshida, J. Nucl. Sci. [115] Y.-Z. Qian and S. Woosley, Astrophys. J 471, 331 Tech. Suppl. 2, 774 (2002). (1996).