
ENDF/B-VIII.0 Errata
On February 2, 2018, ENDF/B-VIII.0 was released. This library included hundreds of new and improved evaluations. Inevitably, it also contains a few errors. On this page we list the most serious of those issues.
Known Issues with ENDF/B-VIII.0:
- n + 10B: When the (n,t) cross section section got switched from MT=113 to MT=700, someone later added back in a (3,113) section from ENDF/B VII.1. This essentially doubled (but not quite) the contribution of the (n,t) cross section to the total. To make matters worse, the modified elastic cross section meant for VIII.0 never made it into the file, and the elastic cross section was still the VII.1 cross section at energies above 1 MeV. Both of these issues are now fixed and now the sum of MT=2 (elastic) and MT=3 (non-elastic) gives the total (MT=1). This is corrected in revision 1541. Corrected file: n-005_B_010.endf
- p + 6Li: The Q value for the p+Li6 -> He3+He4 exit channel should be 4.0201 MeV, not 4.0201 eV. Corrected in revision 1540. Corrected file: p-003_Li_006.endf
- d + 7Li: Low-energy behavior for n+2He4 channel does not follow charged-particle form (is too large) in White et al evaluation. Fixed in revision 1543. Corrected file: d-003_Li_007.endf
- t + 7Li: Cross-section to 2n+2He4 should be halved, as Brune measured inclusive n production. Fixed in revision 1544. Corrected file: t-003_Li_007.endf
- Decay data for 92Rb: beta and gamma intensities and mean energies updated in revision 1542. Corrected file: dec-037_Rb_092.endf
- All elements in the atomic relaxation library: binding energies were not updated from their ENDF/B-VII.1 values (they were updated in the ENDL-formatted files available from the EPICS web area). Fixed in revision 1545. Corrected file: atomic_relax.tar.gz
Long-Standing Issues:
- Energy balance This is by far the most commonly reported issue in the history of the ENDF project. In ENDF/B-VII.1 alone, energy balance tracker items were added here, here, here and here. The ADVANCE continuous integration system for ENDF found that nearly 80% of the data files in the neutron sublibrary have energy balance problems of one form or another. In fact, this class of errors have been reported as early as the late 1970's (R.E. MacFarlane, "Energy Balance of ENDF/B-V", Trans. Am. Nucl. Soc. 33, 681 (1979)). One would think that such a well known issue would have been resolved by now. However, one must realize that the smallest mistake in a file (say an outgoing energy table extending 1 eV too far in E') can result in an energy or momentum imbalance. The most common cause of large imbalances is the lack of outgoing gamma data in many older evaluations. Energy imbalances are much less in newer evaluations but correcting the energy balance in older evaluations is notoriously difficult.
- (Barely) Negative angular distributions at high energy The must commonly used way to represent angular distributions in the ENDF-6 format is using a Legendre series expansion. Such expansions are very effective at representing the quantum mechanical interference effects that result in "holes" in the angular distributions. As one moves up in incident energy, angular distributions get more forward peaked simply due to kinematics. Therefore, more and more Legendre moments are needed to accurately represent high energy angular distributions. These higher order moments are much smaller in absolute magnitude and are crucial for getting the ever-increasing number of "holes" in the distributions correct. As these moments are small in magnitude, they are susceptable to floating point round-off error resulting in the distributions dipping below zero in the distribution "holes". This on-going problem is straightforward to fix with codes such as Red Cullen's LEGEND and may be automatically fixed in processing codes such as NJOY2016, AMPX or FUDGE.
- RRR-URR-Fast region disconnects/discontinuities The physical representation of nuclear data changes abruptly between the Resolved Resonance Region (RRR) and the Unresolved Resonance Region (URR) and again between the URR and fast neutron region. Therefore it is very difficult to achieve an exact match of the reconstructed cross section at the boundaries of the different regions. This is one of the first things a newcomer to ENDF notices and is such a commonly reported issue that we have added a discussion of this issue in the ENDF-6 formats manual in Section 2.4.4.
- Non-positive definite covariance matrices The ENDF-6 format provides several formats for storing covariance data. Ultimately however, all of these formats require the storage of the actual componants of a covariance matrix. Covariance matrices are required to be symmetric positive semi-definite matrices. If one or more subspaces of the matrix correspond to very small or zero eigenvalues, then standard floating point round off error can result in spurious negative eigenvalues. There are no simple solutions to this problem (see the 2011 CSEWG talk by D.Brown for a discussion). Optionally on can seek to store the eigenvalues directly using the MT=30 format (which has actually never been tried in practice and is not supported by current processing codes). We are working toward a systematic solution to this problem using the GNDS format.
- Missing gamma data Prior to the 1990's, the impact of gammas on nuclear applications was mostly ignored. In fact, the ENDF format may require complete outgoing neutron data but does not require outgoing gamma data. We find that many older ENDF evaluations have incomplete gamma data or lack outgoing gamma data altogether. As we revisit these older evaluations, we aim to add the gamma data.