TALYS: Comprehensive Nuclear Reaction Modeling

A.J. Koning , S. Hilaire† and M.C. Duijvestijn


Nuclear Research and Consultancy Group NRG, P.O. Box 25, NL-1755 ZG Petten, The Netherlands
†
Commissariat à l’Energie Atomique, DAM/DIF/DPTA, Boîte Postale 12, 91680 Bruyères-le-Châtel, France

Abstract. TALYS is a nuclear-reaction program which simulates nuclear reactions that involve neutrons, gamma-rays,
protons, deuterons, tritons, helions, and alpha-particles, in the 1 keV - 200 MeV energy range. A suite of nuclear-reaction
models has been implemented into a single code system, enabling us to evaluate basically all nuclear reactions beyond the
resonance range. An overview is given of the main nuclear models used, such as newly developed optical models, various
compound nucleus, fission, gamma-ray strength, level density, and pre-equilibrium models, all driven by a comprehensive
database of nuclear-structure parameters. The predictive power of the code is demonstrated by comparing calculated results
with a very diverse set of experimental observables. Our aim is to show that TALYS represents a robust computational
approach that covers the whole path from fundamental nuclear-reaction models to the creation of complete data libraries
for nuclear applications.

INTRODUCTION existing experimental database.

With the advent of fast computers, software that sim-
ulates nuclear reactions is able to play an increasingly
important role in nuclear data. It is possible to pro-
vide an exact computational scheme for sophisticated
nuclear models, not only new ones but also those that
have been lying on the shelf for decades, and only now
become amenable for numerical implementation. Large-
scale comparisons with measurements are within reach.
TALYS is a nuclear reaction program created at NRG
Petten, the Netherlands, and CEA Bruyères-le-Châtel,
France. The idea to make TALYS was born in 1998,
when we decided to implement our combined knowledge
of nuclear reactions into one single software package.
The objective is to provide a complete and accurate
simulation of nuclear reactions in the 1 keV - 200 MeV
energy range, through an optimal combination of reliable
nuclear models, flexibility, and user-friendliness.
There are two main purposes of TALYS, which are
strongly connected. First, it is a nuclear physics tool that
can be used for the analysis of nuclear-reaction exper-
iments. The interplay between experiment and theory
gives us insight in the fundamental interaction between
particles and nuclei, and precise measurements enable
us to constrain our models and their parameters. In re-
turn, when the resulting nuclear models are believed to
have sufficient predictive power, the prediction can even
give an indication of the reliability of measurements. The
many examples of TALYS presented in this paper and
in various other contributions to this conference confirm
that our software project would be nowhere without the
After the nuclear physics stage comes the second func-
tion of TALYS, namely as a nuclear data tool: After
fine-tuning the adjustable parameters of the various re-
action models using available experimental data, TALYS
can generate nuclear data for all open reaction channels,
interpolating between and extrapolating beyond exper-
imental data, on a user-defined energy and angle grid
beyond the resonance region. The associated nuclear-
data libraries that can be constructed provide essen-
tial information for existing and new nuclear technolo-
gies. Important applications that rely directly on the out-
put of nuclear-reaction simulation software like TALYS
are conventional and innovative power reactors (GEN-
IV), accelerator-driven systems, and transmutation of
radioactive waste, fusion reactors, homeland security,
medical-isotope production, and radiotherapy, oil-well
logging, geophysics, and astrophysics.
NUCLEAR MODELS
Figure 1 shows the nuclear models implemented in
TALYS.
Nuclear Structure and Model Parameters
All nuclear models make use of structure and model
parameters. There is an automatic reference to parame-
ters as masses, resonances, etc.; see the lower left box of
Fig. 1. With a few exceptions, our database is based on
the Reference Input Parameter Library [1].

CP769, International Conference on Nuclear Data for Science and Technology,

edited by R. C. Haight, M. B. Chadwick, T. Kawano, and P. Talou
© 2005 American Institute of Physics 0-7354-0254-X/05/$22.50
1154
 Loops possible over: Direct reaction: Preequilibrium: Output:
* Spherical OM * Exciton model *File ’output’
* Incident energies − 2−component
* DWBA defined by
* Natural isotopes * p−h LD phenom.
* Rotational CC
− surface effects keywords
* Vibrational CC
Optical Model: * Kalbach systematics *Dedicated
* Giant resonances − angular distribution
* Phenomenology * Weak−coupling files with
Input: local / global − cluster emission
spectra, ...
* Keywords, eg: * γ −ray emission
projectile n
element fe
mass 56
energy 14. Nucl. Structure: Compound: Multiple emission:
* Abundancies
ENDF:
* Width fluctuations * Exciton (any order)
* Discrete levels * transport libs
− Moldauer
* Hauser−Feshbach * activation libs
* Deformations − GOE triple integr.
* Masses − HRTW * Fission competition
* Level density par. − isotopic yields
* Hauser−Feshbach
* Resonance par. * γ −ray cascade
* Fission competition
* Fission barrier par. − isotopic yields * All flux depleted
* Thermal XS * γ −ray emission * Exclusive channels
* Microscopic LD * GC+ Ignatyuk * Recoils
* Prescission shapes

FIGURE 1. Flowchart of TALYS.

Optical Model and Direct Reaction Model Compound Nucleus Model

We use the coupled-channel code ECIS-97 [2] as a
subroutine for all optical model and direct reaction cal-
culations. It is the only module in TALYS that is not
written from scratch but adopted from outside. ECIS-
97 delivers the basic observables such as the elastic an-
gular distribution, the reaction, and the total cross sec-
tions. Moreover it yields the transmission coefficients
for compound-nucleus calculations and all cross sec-
tions and angular distributions for discrete states. The
default optical-model potentials (OMP) used in TALYS
are the local and global parameterisations for neutrons
and protons of [3]. For nuclides outside the scope of this
OMP, i.e., strongly deformed nuclides, we allow input
of potentials on an individual basis. With this, coupled-
channel calculations for various types of deformation
(symmetric-rotational, harmonic-vibrational, vibration-
rotational, and asymmetric-rotational) can be automati-
cally performed. For near-spherical nuclides, direct re-
actions are calculated with DWBA, and inelastic scatter-
ing off odd-A nuclei is described by the weak-coupling
model. For deuteron, triton, Helium-3, and alpha OMPs,
we use a folding approach applied on the aforementioned
OMPs.
The term compound nucleus reaction is commonly
used for two different processes: (i) capture of the pro-
jectile in the target nucleus to form a compound nucleus,
which subsequently emits a particle or gamma; (ii) mul-
tiple emission from the chain of excited residual nu-
clides following the binary reaction. Both are included
in TALYS. At low incident energy (i) plays an important
role. It differs from (ii) at two important points: (a) the
presence of width-fluctuation corrections and (b) non-
isotropic, though still symmetric, angular distributions.
It is calculated with the Hauser-Feshbach formalism in-
cluding width fluctuation corrections (WFC). The WFC
factor accounts for the correlations that exist between the
incident and outgoing waves. From a qualitative point of
view, these correlations enhance the elastic channel and
accordingly decrease the other open channels. In gen-
eral, the WFC factor may be calculated using three dif-
ferent expressions, which have all been implemented in
TALYS: The HRTW model, the Moldauer model, and the
model using the Gaussian Orthogonal Ensemble (GOE).
A comparison between the three models is given in [4],
where the Moldauer model is confirmed as the best de-
fault choice. All WFC models are generalized to include
continuum particle emission, gamma-ray competition,
and fission. Gamma-ray coefficients are modeled with

1155
Kopecky-Uhl’s generalized Lorentzian and the appropri-
ate giant-dipole resonance parameters. Besides cross sec-
tions, compound angular distributions are calculated us-
ing Blatt-Biedenharn coupling factors, again within a full
Hauser-Feshbach expression with WFC.
For multiple emission, the whole reaction chain is
followed by depleting each [nucleus-excitation energy-
spin-parity] bin with particle, gamma, or fission decay
until all channels are closed. In the process, all particle
and residual production cross sections are accumulated
to their final values. Non-equidistant energy grids in this
decay scheme ensure enough precision in the compound-
evaporation peaks.
Level Densities
We use several models for the level density in TALYS,
which range from phenomenological analytical expres-
sions to tabulated level densities derived from micro-
scopic calculations. So far, the most robust approach
seems to be a Fermi gas model at high energies, with
shell- and energy-dependent level-density parameter a,
and a constant-temperature model, fitted to the known
discrete states, at low energy. For non-fissile nuclides,
we generally use an effective level-density model, i.e., all
collective enhancements are included in the level-density
parameter a. For fissile nuclides, we account for an ex-
plicit rotational and vibrational enhancement as well as
their appropriate damping at high energies. More precise
details on the various forms for the level density can be
found in the manual and in [5].
Fission
For fission, the default model used in TALYS is based
on the Hill-Wheeler expression for the transmission co-
efficient for one, two, or three barriers. If the excitation
energy of the compound nucleus is lower than the bar-
rier heights, fission-transmission coefficients display a
resonant structure that is due to the presence of nuclear
excited levels in the second well of the potential-energy
surface. These so-called class II states modify the fission
transmission coefficients. The total fission transmission
coefficient for a compound nucleus is then obtained by
summing the individual Hill-Wheeler terms over all head
band transition states and, for the continuum, integrating
it using the aforementioned fission level densities. Multi-
chance fission for all residual nuclides is included.
A novel feature for any general nuclear model code is
the ability to predict fission yields. This is done with the
multi-modal random neck rupture model and the scission
point model, as described in more detail in [5].
1156
Pre-Equilibrium Model
For energies above a few MeV, pre-equilibrium reac-
tions play an important role. For nucleon reactions, we
have implemented a two-component exciton model with
a new form for the internal transition rates based on the
OMP of [3], which yields an improved description of
pre-equilibrium processes over the whole energy range
[6]. Another feature necessary to cover a large energy
range is the generalization of multiple pre-equilibrium
processes up to any order of particle emission. This is
accomplished by keeping track of all successive particle-
hole excitations of either proton or neutron type; see [6]
for the mathematical outline.
On top of the contribution of the single-particle exci-
ton model, which yields essentially structureless emis-
sion spectra, we add a contribution from giant reso-
nances, computed with a macroscopic, phenomenologi-
cal model, accounting for the energy-weighted sum rule.
Pre-equilibrium photon emission is taken into account
with the model of Akkermans and Gruppelaar [7]. For
pre-equilibrium reactions involving deuterons up to al-
pha particles, a (too-low) contribution is automatically
calculated within the exciton-model reaction equations.
However, for nuclear reactions involving projectiles and
ejectiles with different particle numbers, mechanisms
like stripping, pick-up, and knock-out play an important
role and these direct-like reactions need to be added in-
coherently. Kalbach [8] developed a phenomenological
contribution for these mechanisms, which we have in-
cluded in TALYS, resulting in a much better prediction
of complex-particle cross sections as compared to many
older reaction codes. However, the (very) phenomeno-
logical nature of the model still provides a challenge to
construct a more physical approach for these reactions in
the future.
Finally, pre-equilibrium angular distributions are pre-
dicted by Kalbach’s systematics. From a physical point
of view, the quantum-mechanical multi-step approach
is preferable, although it is difficult to find justifica-
tion (from applications) for angular precision that goes
beyond that of the systematics. Including our existing
multi-step direct software is nevertheless left as future
work.
THE CODE
The present version of TALYS is written in Fortran77,
and so far has been successfully compiled with various
f77 and f90/f95 compilers. We have aimed at a setup
that is as modular as Fortran77 allows it to be, using
programming procedures that are consistent throughout
the whole code. In total, there are about 250 subroutines
adding up to a total of more than 40000 lines, plus the
20000 lines of the ECIS-97 subroutine.
The code is rather flexible in its use. Indeed, a com-
plete set of cross sections can already be obtained with
minimal effort, through a four-line input file of the type
given in Fig. 1, which produces the best “blind” answers
TALYS can currently give. Generally, a user wants to be
more specific on the choice of nuclear models, their pa-
rameters, and degree of output. For this, more than 150
different keywords are available that can be specified
to e.g., fit experimental data. The code has been thor-
oughly tested on a formal level, through random input
files (probing every corner of the code). Full dripline-to-
dripline calculations (including the production of ENDF-
6 data files) for all types of projectiles up to 200 MeV,
have been performed to validate the code computation-
ally and to test the continuity (smoothness) of the results.
WHAT CAN WE DO WITH TALYS?
The main aim of TALYS is to provide a complete set
of answers for a nuclear reaction, for all open channels
and associated cross sections, spectra, and/or angular dis-
tributions. It depends on the current status of nuclear-
reaction theory, and our ability to model that theory,
whether these answers are generated by more or less so-
phisticated physical methods or by simpler phenomeno-
logical approaches. The following data can be calculated:
• Total, elastic, and reaction cross sections,
• Non-elastic cross sections per discrete state,
• Elastic and non-elastic angular distributions,
• Exclusive reaction channels ((n,2n), (n,np), etc.),
• Exclusive double-differential spectra,
• Exclusive isomeric-production cross sections,
• Discrete and continuum gamma-ray production
cross sections,
• Extrapolation of non-threshold cross sections down
to the thermal energy range [9],
• Total particle-production cross sections, e.g., (n,xn),
• Single- and double-differential particle spectra,
• Residual production cross sections ( isomers),
• Recoils,
• Fission cross sections and fission yields.
Figures 2 and 3 give an impression of the variety of nu-
clear reactions that can be simulated. The shown TALYS
curves are obtained from blind calculations (e.g., resid-
ual production, energy spectra), after minor fitting (e.g.,
capture, n α , inelastic), and after significant parameter
adjustments (e.g., fission cross section).
A complete ENDF-6 data file, above the resonance
range, up to 20 MeV can be computed in about 4 minutes
1157
(20-MeV file) up to 4 hours (200-MeV file) on a 1-
GHz PC. Various new evaluations for the JEFF-3 library
have been produced [10] through fitting, using model
parameters, of the experimental data. With default input
files, TALYS is also suitable for mass production of
nuclear data and has been used to generate data for large-
scale activation libraries such as EAF [11], but also for
more fundamental purposes such as astrophysics [12].
CONCLUSIONS
The development of TALYS has followed the “first com-
pleteness, then quality” principle. This merely means
that, in our quest for completeness, we try to divide
our effort equally among all nuclear-reaction types. We
think that, with the exception of a few fission items (neu-
tron spectrum, evaporating fragments), the code is indeed
“complete.” “Quality” is obviously a different and sub-
jective issue, and it is certain that future enhancements in
various theoretical models are needed to bring our com-
puted results even closer to measurements.
REFERENCES
1. Reference Input Parameter Library, http://www-
nds.iaea.or.at/RIPL-2/.
2. J. Raynal, Notes on ECIS94, CEA Saclay Report No.
CEA-N-2772, (1994).
3. A.J. Koning and J.P. Delaroche, Nucl. Phys. A713, 231
(2003).
4. S. Hilaire, Ch. Lagrange and A.J. Koning, Ann. Phys.
306, 209 (2003).
5. M.C. Duijvestijn and A.J. Koning, “Fission yield
predictions with TALYS,” this conference.
6. A.J. Koning and M.C. Duijvestijn, “A global pre-
equilibrium analysis from 7 to 200 MeV based on the
optical model potential,” Nucl. Phys. A, in press.
7. J.M. Akkermans and H. Gruppelaar, Phys. Lett. 157B,
95 (1985).
8. C. Kalbach, “Pre-equilibrium reactions with complex
particle channels,” this conference.
9. Collaboration with JUKO Research (J. Kopecky).
10. A.J. Koning, M.C. Duijvestijn, S.C. van der Marck, R.
Klein Meulekamp, and A. Hogenbirk, “New nuclear
data evaluations for Ca, Sc, Fe, Ge, Pb, and Bi isotopes,”
this conference.
11. R.A. Forrest, “European Activation File EAF-2005,”
this conference.
12. E. Khan, S. Goriely, D. Allard, E. Parizot, T. Suomijarvi,
A.J. Koning, S. Hilaire, and M.C. Duijvestijn, “Impact of
the giant dipole resonance on the photodisintegration of
ultrahigh energy cosmic rays,” submitted to AstroParticle
Journal.
 4 93
10
(a) n + Nb: Total cross section (b)
63 63
3
Cu(n,n) Cu
10 11000
1.6
2
10 10000

1 2.0 9000
10

10
0 2.2 8000
dσ/dΩ (mb/sr)

Cross section (mb)

−1 7000
10 2.6

−2
6000
10
3.0
5000
−3
10
3.4
4000
−4
10
3.8 3000
−5
10
2000
5.5
−6
10
1000 −2 −1 0 1 2
10 10 10 10 10
10
−7
Energy (MeV)
0 30 60 90 120 150 180
Θc.m. (deg)
93
n + Nb at 10 keV Pu−239(n,f)
Elastic angular distribution
670 2500
(c) (d)
Hauser−Feshbach 2400
660 Moldauer
HRTW 2300
GOE
2200
650
2100
Cross section (mb/sr)

640 2000

1900
630 1800

1700
ENDF/B−VI
620
1600 JEFF−3.0
JENDL−3.3
1500 TALYS
610
1400

600 1300
0 30 60 90 120 150 180
Angle (deg) 1200
0 2 4 6 8 10 12 14 16 18 20
74 208
Ge(n,γ) Pb(n,n1’)
Π −
J = 3 , Ex = 2.61 MeV
100
Exp (e) 1200 (f)
TALYS Towle and Gilboy (1963)
ENDF/B−VI.8 Almen−Ramstroem (1975)
JENDL−3.3 1000 TALYS
ENDF/B−VI.8
JENDL−3.3
10 ENDF/B−VI.8 mod
800
Cross section (mb)

Cross section (mb)

600

1 400

200

0 0
1e−02 1e−01 1e+00 1e+01 0 2 4 6 8 10 12 14 16 18 20
E (MeV) Energy (MeV)

FIGURE 2. TALYS vs. experimental data: (a) n + 63 Cu elastic angular distributions, (b) n + 93 Nb total cross section, (c) different
width fluctuation correction models compared for 10 keV n + 93 Nb, (d) n + 239 Pu total fission cross section, (e) 74 Ge n γ  cross
section, (f) 208 Pb inelastic cross section for first discrete level.

1158
 207 238
Pb(n,2nγ) 5.5 MeV n+ U
Level 1 −> Level 0: Eγ=0.80 MeV Pre−neutron emission mass yield curve
1
2000 10
Vonach et al. (1994)
(a) (b)
TALYS

0
1500 10
cross section (mb)

Yield [%]
−1
1000 10

−2
500 10
Vives et al (2000)
TALYS

−3
10
0 70 80 90 100 110 120 130 140 150 160 170
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Energy (MeV) A
40 28
Ca(n,α) 20 MeV n + Si: recoil spectrum

400 150

TALYS
(c) Exact approach
(d)
ENDF/B−VI.8 Average energy approximation
JENDL−3.3

300
Cross section (mb/Mev)

100

200

50

100

0
0 1 2 3 4 5
24
0 Recoil energy of Mg (MeV)
0 2 4 6 8 10 12 14 16 18 20
11 nat
10 p + Fe
10
10
(p,xp): 62 MeV 3
10
10
9
27
Al (e) (f)
54 2
Fe 10
8
10 56
56
Fe 1 Co
7 10
10
dσ/dE [mb/MeV]

60
6 Ni 0
10 10
Cross section (mb)

89
5
Y 55
Co
10 10
−1
120
4
Sn
10 197 −2
Au 10
3 54
10 nat Mn
Pb 10
−3
2
10 209
Bi −4
10
1 10

0 52
10 10
−5
Mn
−1
10 −6
0 10 20 30 40 50 60 70 10
0 10 20 30 40 50 60 70 80 90 100
Eout [MeV] Incident energy (MeV)

FIGURE 3. TALYS vs. experimental data: (a) exclusive gamma-ray production for 207 Pb(n,2n), (b) fission mass yield curve for
5.5 MeV n + 238 U, (c) 40 Can α  cross section, (d) two different recoil models for 20 MeV n + 28 Si, (e) 62 MeV (p,xp) spectra for
various targets, (f) p + nat Fe residual production cross sections.

1159