Status of ENDF/B-VII.1 library
Report to WPEC, May 12-13, 2011, NEA Headquarters, Paris, France
M. Herman for CSEWG
National Nuclear Data Center, Brookhaven National Laboratory, Upton, NY 11973-5000, USA
The US nuclear data community, coordinated by CSEWG, is preparing release of the ENDF/BVII.1 library. This new version will address deficiencies identified in ENDF/B-VII.0, include improved neutron evaluations for several materials, provide covariances for 185 materials, revise fission
product yields and contain essential update of the decay data sub-library. This report summarizes
current work, describes status of the new files and outlines further developments.
I.
INTRODUCTION
The ENDF/B-VII.0 library, made available in December 2006, has been the first major release of the US nuclear reaction data library in 16 years. It was extensively
documented in the special issue of Nuclear Data Sheets,
which since then, has ben cited more than 330 times. Intensive validation proved generally good performance of
the library but a number of deficiencies were discerned.
• The sodium void worths are under-predicted and
exhibit changes with core loading.
• Deficiencies in 2 H cross sections impact upon calculated D2 O moderated HEU benchmark eigenvalues and ZED-2 coolant void reactivity coefficients.
Mosteller noted significant (about 1%) decrease in
the calculated kef f for HEU-D2 O benchmarks attributed to elastic scattering angular distribution
changes below 3 MeV.
• Several minor actinides, carried over from the previous versions of the library, are often crude since
role of these materials in applications used to be
marginal. Minor actinides gained importance only
with the advent of new generation of nuclear reactors.
• It was recognized that that nuclear data structural
materials: Ti, V, Mn, Cr, Ni, W in ENDF/B-VII.0
need upgrading, due to poor performance in some
critical assemblies, such as ANL ZPR.
• Dean noted that thermal capture for certain fission
products are deficient and need to be revisited.
• A study of fission product yields for plutonium
identified that the England and Rider (ENDF/BVI) data for the important fission product 99Mo
are too low by 4%.
• It turned out that theory based delayed neutrons,
which replaced ENDF/B-VI data, were problematic and we decided to revert to previous estimates.
In 2008 CSEWG decided to undertake multilaboratory effort leading to VII.1 version of the library
to be released in December 2011. Highlights of the new
library include (i) a consistent set of covariance data targeting explicit needs of the Advanced Fuel Cycle Initiative, (ii) new R-matrix based evaluations for several
light nuclei, (iii) evaluations for reactions on structural
materials in both the fast neutron region and the resonance region, (iv) improvements of resonances regions
and thermal cross sections for certain fission products
and neutron absorber materials (Cd, Gd), (v) improvements in minor actinide evaluations for isotopes of U,
Np, Pu, and Am , (vi) adoption of JENDL4.0 evaluations for the Cm, Bk, Cf, Es, Fm, and some other
minor actinides, (vii) fission product yield advances for
fission-spectrum neutrons and 14 MeV neutrons incident
on 239 Pu, and (viii) a new Decay Data sub-library. We
note that no significant changes are expected for the major actinides 235,8 U and 239 Pu, except reverting delayed
neutron data to ENDF/B-VI.8 and adding covariances.
Current update concerns only the most important neutron sub-library plus fission yields, and decay data leaving the remaining 11 sub-libraries unchanged.
II.
STATUS OF ENDF/B-VII.1
Three preliminary versions of the library have been
released for testing. The most recent beta2 was made
available on February 22, 2011. At the time this report
is being written the beta3 version is being assembled and
verified at BNL. Its release is expected by the middle of
May 2011. While most of the new cross section evaluations were already available in beta2 the forthcoming
version will contain a full set of covariances expected in
ENDF/B-VII.1.
Since January 2010 the ENDF/B library is
maintained under the NNDC GForge server
(https://ndclx4.bnl.gov/gf/). The previous ENDF/A library containing newly submitted files has been replaced
by the Subversion system, which keeps track of all new
submittals and modifications. The GForge system stores
also all beta releases of the future library, evaluation and
release reports, CSEWG minutes, and other auxiliary
files. It manages also the ENDF mailing list.
EMPIRE: Nuclear Reaction Code...
A.
NUCLEAR DATA SHEETS
is relativistic, so that even zero-mass particles, such as
photons, are treated correctly. Experimental data can
be modified by the use of adjustable normalizations and
energy shifts, and the calculations can fold in the effects
of beam energy resolution/spread.
In ENDF/B-VII.1 new analyses have been performed
for neutrons plus 4 He, 6 Li , 9 Be, and 16 O. The 9 Be evaluation includes new data from RPI. In the following we focus on 16 O because of it’s paramount importance for various applications. The ENDF/B-VII.0 evaluation, carried over from ENDF/B- VI.8, combined analyses at Los
Alamos above 3.4 MeV and work at the KAPL naval reactor Lab below this energy. It was, however, performing
well in integral validation. For the VII.1 we have developed a new, more consistent, R-matrix evaluation that
better represents the available cross section data. The
new evaluation of the n+16 O cross sections is based on
an R-matrix analysis of reactions in the 17 O system at
energies up to 7 MeV. Several new measurements were
added to the data set for α +13 C reactions. This resulted
in a change that essentially undid the previous revision
that had lowered the evaluated (n,α) cross section by 32%
at energies below about 9 MeV. Validation performed so
far indicates that the new file performs at least as good
as the previous one, while the cross sections are in better
agreement with the differential measurements.
Covariances
The ENDF/B-VII.1beta3 release will contain 185 materials with covariances in the neutron sublibrary. These
constitute the vast majority of covariances in the future
VII.1 and represent one of the major improvements. Covariances for 110 materials explicitly address needs of the
AFCI community. Specifically, they are intended to facilitate data adjustment for fast reactors. The current
library of covariances builds on three previous projects,
with the third one (COMMARA-2.0) being actual source
of the 110 core covariance files.
BOLNA A pilot library created by five laboratories
(BOLNA = Brookhaven-Oak Ridge-Los AlamosNRG Petten-Argonne) within WPEC Subgroup 26.
The library contains covariances for 52 materials,
put together on a short timetable.
Low-Fidelity A library created in 2007-2008 by four
US National Laboratories under a DOE project for
testing nuclear criticality safety methods. A full set
of materials included in ENDF/B-VII.0 using simple, yet not unreasonable, methods. The emphasize
was on completeness rather than quality, hence the
name low-fidelity.
COMMARA-2.0 A library produced through a BNLLANL collaboration during 2008-2011 for 110 materials of importance for the Advanced Fuel Cycle
Initiative.
C.
Halogens and alkali metals
These new evaluations include resonance ranges in 19 F,
Cl, 39,41 K evaluated by ORNL and full new evaluation of 23 Na, which is still under development.
The ORNL evaluations of neutron cross sections in the
resolved resonance region were done with the multilevel
Reich-Moore R-matrix formalism of the code SAMMY
and incorporate recent high-resolution capture and transmission measurements. The resonance parameters for
19
F, 35 C were the first to use extended Reich-Moore
formalism coded in the ENDF representation with the
LRF=7 option. This novelty allows to describe open inelastic channels with the resonance parameters. For example, in the energy range up to 1 MeV 19 F has two
inelastic channels starting at the energies 109.9 keV with
spin 1/2- and 197.2 keV with spin 5/2+. These two inelastic channels were included in the SAMMY evaluation.
Comparison of the thermal elastic, capture and total
cross sections at room temperature with the data listed in
Atlas of Neutron Resonances is generally very favorable.
Only the case of thermal scattering cross-section for 41 K
the Atlas value is almost 3 times smaller than that of the
present evaluation. The reasons for the discrepancies are
not known.
The BNL evaluation of 23 Na is still in progress but it
is worth mentioning because of the novelty in the evaluation approach. The methodology, applied in collaboration with INL, is called ’consistent adjustment’ or ’assimilation’ and involves use of both differential and in35,37
In addition, ENDF/B-VII.1 will contain covariances
already present in VII.0, new covariances produced
by ORNL and LANL which do not make part of
COMMARA-2.0 and a consistent set of covariances in
the adopted JENDL-4.0 files for minor actinides. Last
but not least, standards’ covariances have been incorporated in the respective VII.1 files, although in a few
limited energy regions in 235,238 U and 239 Pu central values of cross sections differ from those recommended for
standards. In these cases standards’ covariances will be
properly scaled to account for the difference.
The detailed list of materials with covariances
is given in the file ENDF-B-VII.1beta3-covar.txt at
https://ndclx4.bnl.gov/gf/project/endf/frs/.
B.
M. Herman et al.
Light nuclei
As is true for all the light-element evaluations done
at Los Alamos, these are based on R-matrix analyses
of experimental data for the neutron plus target compound system, using the LANL Energy Dependent Analysis code, EDA. The multichannel R-matrix formalism
has been implemented in its full generality, and without
any approximations, in the EDA code. It accommodates
any number of two-body channels having particles with
arbitrary spins, masses, and charges. The formulation
2
EMPIRE: Nuclear Reaction Code...
NUCLEAR DATA SHEETS
Such process involves a tight interaction of the modelers
and benchmark evaluators and requires a truly interdisciplinary evaluation team.
tegral data to constrain nuclear reaction model parameters. Thus the adjustment to the integral data occurs
on the level of physical model parameters rather than on
the level of multi-group cross sections, removing application dependance of the classical adjustment procedure.
The consistent adjustment requires that the evaluation
be fully determined by a set of model parameters and
related nuclear reaction models and needs clean, well defined integral experiments, which are predominantly sensitive to the considered material.
D.
M. Herman et al.
E.
Fission products and absorbers
Modified were: 95 Mo, 99 Tc, 103 Rh, 109 Ag, 133 Cs,
Nd, 147,149,152 Sm, 153 Eu, 62 Ni, 90,91 Zr, 113 Cd,
157
and
Gd. Improvements to the fission products and
absorbers regarded mostly adjustment of the thermal
cross sections and in some cases of the full resonance
region. These modifications were performed by Said
Mughabghab (BNL) partly in response to the criticism
by Dean who reported poor performance of VII.0 in his
reactivity worths measurements. New analysis, also taking into account new experimental data when available,
brought VII.1 significantly closer to the Dean’s results.
In addition, there were dramatic changes in thermal cross
sections for 90 Zr and 91 Zr, for 113 Cd and 157 Gd suggested
by the new experiments. In summary:
143,145
Structural materials
New evaluations have been prepared for the following structural materials: Ti, and V (LANL), 55 Mn and
W (IAEA), 50,52,53,54 Cr and 58,60 Ni (ORNL), 181 Ta,
185,187
Re (LLNL).
It is interesting to note that important modification
to the 48 Ti was elastic scattering angular distributions
at low energies. In ENDF/B- VII.0 the elastic scattering angular distributions are calculated with the optical
model and the Hauser-Feshbach statistical model. We
found that the calculated P1 (the L = 1 component of
the Legendre expansion coefficients) for the differential
elastic scattering tends to overestimate reflection of neutrons in the critical assemblies with a Ti reflector, and
adjustment of the optical potential parameters does not
solve this problem. This was finally by replacing the elastic scattering angular distributions of 48 Ti up to 4 MeV
by those in ENDF/B-VI in which the angular distributions were evaluated based experimental data. The larger
P1 values in ENDF/B- VII.1 in the fast energy range give
more neutron scattering in the forward angles, which results in less reflection of neutrons inside the reflector.
New evaluations for tungsten isotopes are the high
points of the VII.1 library. Notoriously poor performance
of tungstens in VII.0 has been dramatically improved
thanks to the consistent modeling in the fast neutron
region with EMPIRE code . Here, the essential factors
were the optical model potentials developed specifically
for the current evaluations. This new class of dispersive
optical model potentials serve as backbone of undertaken
nuclear reaction modeling. Selected benchmarks covering both criticality and fusion domains were calculated
for the assembled evaluated data file; benchmark results
were analyzed to further constrain the theoretical modeling and model parameters. Then, the whole process
was repeated starting from a new theoretical calculation
and concluding with a new least- squares fitting that produces a new evaluated data file. Iterations were carried
on until benchmarks performance does not improve anymore. Described feedback to the theoretical modeling
based on benchmarks results is considered to be an extremely important step in improving evaluated data file.
A big advantage of going from benchmarks results back
to modeling stage is that internal consistency and general applicability of derived evaluated files are preserved.
• Except for 95 Mo, the major significant discrepancies of the reactivity-worth results of Dean are removed in the ENDF/B-VII.1 evaluations.
• The 62 Ni capture cross section was reevaluated
so that the computed ENDF/B-VII.1 30-keV
Maxwellian capture cross section agreed with direct measurements.
• The thermal capture cross sections of 90 Zr and 91 Zr
were reevaluated to account for a new measurement
and those of 113 Cd and 157 Gd modified to reflect
recent differential and integral measurements.
F.
Actinides
While major actinides were not modified, except
reverting to ENDF/B- VI.8 delayed neutron parameters, minor actinides 237 Np, 233, 236, 237, 238, 239U,
238,239,240
Pu, and 241 Am were either reevaluated or improved.
The 237 Np evaluation has been updated using a recent
evaluation of Maslov et al. Their evaluation concentrated
on the production of the short-lived 236 Np isomer and
replaces the (n,2n) and (n,3n) cross sections.
Considered for adoption is Phil Youngs latest evaluation for 233 U that reevaluates the inelastic scattering
cross sections following feedback from Trkov.
We have made two modifications to the 236 U evaluation: in the energy region near 100 keV the fission cross
section has been modified slightly to provide a smoother
match to the unresolved resonance region; and radiative capture has been increased by about 10% for the
energy region above 100 keV. The capture modification
was motivated mainly by MCNP simulations of 236 U(n,γ)
3
EMPIRE: Nuclear Reaction Code...
NUCLEAR DATA SHEETS
M. Herman et al.
a new evaluated fission cross section down to 150 eV. The
sub-threshold fission cross sections are given in File 3.
in fast critical assemblies that previously underpredicted
the measured LANL data by about 10%.
The ENDF/B-VII.0 evaluation for 237 U had various
limitations, originating largely from the fact that very
little measured data exist for this unstable nucleus and
nuclear reaction modeling has a limited predictive power
for reactions that involve fission. Presently, model calculations have an improved predictive power once they involve some calibration to measured data on nearby uranium isotopes. These results were employed to derive
new recommendations for 237 U in fast neutron region.
These data were smoothly matched onto the JENDL3.3
resolved and unresolved resonance evaluation below 100
keV (we modified the unresolved resonance representation slightly to facilitate this).
In performing our new evaluation for 239U, we changed
the resonance region and all of the cross section data.
Our new (n,f) cross section evaluation combines the reanalyses of Younes and Britt with the new surrogate reaction experiment performed by Burke et al. The resonance
range was generated with the EMPIRE code within a
picket fence model with realistic resonance spacings.
A new evaluation was performed for neutron-induced
reactions on Pu-238 in the fast neutron region. The evaluation is based on model calculations, as well as analysis of experimental data. The ECIS94 code was used to
perform coupled-channels optical model calculations, and
obtain total, shape and reaction cross sections, as well as
all discrete elastic and inelastic cross sections and angular distributions. Neutron transmission coefficients used
for statistical Hauser-Feshbach calculations were also inferred from the coupled-channels results. The optical
model potential developed recently by Soukhovitskii et
al. for even-even plutonium isotopes was used in this
work. The GNASH and COMNUC codes, which implement the Hauser-Feshbach equations, width fluctuation
corrections as well as pre-equilibrium components, were
used to compute (n, xn) reaction cross sections. The
COH code was used for computing the neutron radiative capture cross section. The GLUCS statistical analysis code was used to analyze experimental data sets,
and in particular, infer the fission cross section as well
as prompt fission neutron multiplicity. The JENDL-4.0
evaluation [160] was also used to complement the present
work in certain areas. The average prompt fission neutron spectrum (PFNS) and multiplicity (PFNM) were
evaluated using Los Alamos (Madland-Nix) model calculations. The systematics developed by Tudora and Vladuca for the model input parameters were used as prior
parameters in our analysis.
The evaluation of neutron-induced reactions on Pu240 follows the exact same methodology used for Pu-238.
The ECIS, COH, GNASH, COMNUC, GLUCS, PFNS
codes were also used here.
Our americium 241 Am evaluation for ENDF/B-VII.1
builds on the VII.0 work, and makes some modest
changes for fission and capture. We performed a new
SOK code statistical analysis of measured data to obtain
G.
Fission yields
For ENDF/B-VII.1 a new fission product yield (FPY)
evaluation for n+239Pu was developed by Chadwick et
al., updating the previous evaluation for ENDF/B- VI
by England and Rider. The evaluation was not changed
for incident neutrons at thermal energies as we find that
the original evaluation is reliable here; but significant
changes were made for incident neutrons with energies
corresponding to fission spectrum, and 14 MeV neutrons.
The fission spectrum evaluation work is described in detail in a previous issue of Nuclear Data Sheets.
H.
Decay data
The decay data sub-library is based on the Evaluated
Nuclear Structure Data File (ENSDF), following a translation into the ENDF-6 format. On average, data in
ENSDF are updated every 7 years, which basically means
that an important amount of the data in VII.0 was reevaluated. The VII.1 version of the sub-library includes fixes
to errors in VII.0, such as the positron intensity, which
was left as zero in the VII.0 version. Additionally, the
atomic radiation part of the library was modified. Atomic
radiation, X-rays and Auger electrons, are produced from
the filling of atomic vacancies produced in electron capture and electron conversion.
In the VII.1 version, the atomic data from the Evaluated Atomic Data Library (EADL) developed by LLNL
was used . All the K-L, K-M and K-N as well as
the L?, L? and L? X-rays are included. In addition,
the KLL, KLX, KXY, LLX, LMM, LMX, LXY, MMX,
and MXY average Auger electrons are also listed. The
electron conversion to atomic sub-shells was calculated
with the code BRICC. Finally, the average electromagnetic energy (EEM) and average light particle energy
(ELP) values from the recently published TAGS data
for 102,104,105,106,107 Tc, 101 Nb and 102 Mo, were also included in the VII.1 release. The inclusion of these data
has improved the calculation of decay heat for n+239 Pu
at cooling times 10 - 10000 seconds.
III.
RELEASE SCHEDULE
The next preliminary release of the library (beta3) is
currently being assembled and should be made available
by middle of May 2011. This release will differ from the
beta2 mostly in the AFCI and standards’ covariances
that will be included. Subsequent beta4 version might
be released between mini-CSEWG meeting in June and
full CSEWG meeting in November 2011. This one will
mostly contain formal fixes and eventual modifications
4
EMPIRE: Nuclear Reaction Code...
NUCLEAR DATA SHEETS
resulting from the validation exercise. A very few, previously agreed upon and well validated, new evaluations
might be included (e.g., 23 Na) at this stage.
Final release of the library is scheduled for December 2011. It will be accompanied by the dedicated issue
of Nuclear Data Sheets with extended documentation of
M. Herman et al.
the library. Contained will be ”Big Paper II”, about
100 pages, describing cross section part of the neutron
sub-library and several papers dedicated to the covariances prepared by the participating Labs (BNL, LANL,
ORNL). In addition, general overview of the covariances
in ENDF/B-VII.1 will be given in a paper by D. Smith.
5