C.T. Chantler, School of Physics, University of Melbourne
Parkville, Victoria 3052, Australia
1. Absorption edge precision
Some comment has been received regarding the accuracy of energies of absorption edges. Since this is where the errors have been introduced it is useful to address this. Accuracies of edge positions are limited by chemical shifts and the detailed structure of the experimental material observed. Usually an accuracy of absolute energies below 1 - 3 eV is unattainable for this reason. At low energies (less than 200 - 500 eV) the occurrence of collective valence effects and dipole resonances can lead to much larger deviations (up to e.g., 50 eV or 10%).
LII to LIII energy separations are well understood theoretically, and yet the tabulation claims that these energy separations are 0 eV for Z=7 (N) to Z=14 (Si). Better estimates of these separations for Z=12 to Z=14 are 0.4 eV, 0.4 eV and 0.6 eV. For Z=11 and below, the separations are 0.1 eV or less; and in this range the separation is much less than the absolute energy precision. Discrepancies of the absolute energies with respect to other sources reveal uncertainties of 3 eV for Ne LII, 0.7 eV for Na LII and 1.8 eV for Mg LII, for example. These are commensurate with uncertainties enumerated above. Similarly, the MII and MIII edges for Cl through to Zn (Z=17-30) will be separated by up to 2.8 eV, and the MIV and MV edges for Zn (Z=30) to Kr (Z=36) are separated by up to 1.2 eV, as opposed to the common values given in Table 1, Ref. 1 and Ref. 2. These separations are similarly below 1-3 eV and hence are consistent with the uncertainties listed.
The edge energies used are not derived from theory but are based on earlier experimental data . Hence they should be consistent with the estimated experimental uncertainties - as they are. The reader was commended to allow for a shift of particular edge energies where the chemical shift in the structure of interest was known to a greater precision than those in the tabulation. However, this should be done with care, if at all, since the data only separates the K shell attenuation from the other shells - for convenience and length constraints it does not give separate orbital dependencies.
The author would like to acknowledge comments and responses from several colleagues, but in particular would like to thank Dr. J.-L. Staudenmann at NIST for a detailed critique which has accelerated the presentation of this erratum.