2 Method

In order to find the line broadening cross-section for a given line from the tabulated theoretical data, one must correctly determine the effective principal quantum numbers n^* and orbital angular momentum quantum numbers l for each state of the transition. To determine n^* one must know the binding energy of the optical electron. Often this is just the difference between the ionisation energy for the species (which is defined as that energy required to ionise the ground state) and the energy of the state considered. However if the atom has two excited electrons then this series limit is no longer appropriate, and one must account for the excited parent configuration.

These types of states are common in iron group elements, and a particular example was explained in detail by Barklem et al. ([1998]). Failure to account for this can lead to quite substantial overestimates of the broadening. An extreme example is the 6439 Å line of neutral calcium, where if one does not account for the excited parent configuration one finds a cross-section of 920 atomic units, while if one accounts for the parent configuration correctly one obtains 366 atomic units. This latter value is completely consistent with the observed broadening in the solar spectrum as shown by Fig. 1.

Figure 1: Comparison of synthetic flux spectra with solar observations (double line - NSO/Kitt Peak data) for the Ca 6439 Å line. Synthetic spectra are shown for cross-sections of 366 (full) and 920 (dashed) atomic units using LTE and the Holweger & Müller ([1974]) model atmosphere, meteoritic abundances (Grevesse et al. [1996]) and other atomic data from VALD

Hence to correctly compute n^* requires that one knows both the energy of the level and the correct series limit for the parent configuration of the state. If the series limit is not available, the next best thing to have then is detailed information about the electron configuration of the state from which one could determine the parent configuration and series limit.

The Kurucz ([1993]) CDROM lists of spectral line data, which appear to be the most comprehensive currently available, provide the energy levels but unfortunately do not provide the detail in the electron configuration required. It does provide the term designations which can be converted to electron configurations but requires significant extra work which we may consider doing in the future.

However, the NIST Atomic Spectra Database (http://physics.nist.gov/cgi-bin/AtData/main_asd) does provide complete electron configuration information, and although the database is not as comprehensive as the Kurucz lists it still covers a large number of spectral lines. Most importantly for our purposes, the list covers a large fraction of the lines which attain substantial strength in cool stars where this broadening mechanism will be important. For the purpose of identification and integrating the data into VALD, it is also necessary to have species, wavelengths and J quantum numbers. All of this is also provided by the NIST Atomic Spectra Database.

Data were extracted from the NIST database for lines of all neutral species in the wavelength region between 2300 and 13000 Å, where all of the required information was available. This included all elements with atomic number up to and including 28 (Ni). Data for hydrogen lines have not been included as the broadening is very complex and is currently under investigation. Lines of noble gases have not been included as the model of the interaction is inappropriate. Lines of noble gases are never strong in cool stars in any case. Lines with $\log gf < -5$ were also excluded, as such lines will always be weak in stellar spectra and hence their equivalent widths will be unaffected by this broadening mechanism.

In a previous paper (Barklem et al. [1998]) a fortran program was announced which was able to interpolate data from the three sets of tables for lines of neutral atoms. The code was placed on the World-Wide-Web. This code formed the core of the program used to compute the line broadening data in this list. Line data were then computed for all lines from the NIST list where the transition type and the n^* values fell on the available tables, which was certainly the majority of the lines. The most difficult part of the processing was the determination of the correct parent configuration.

A list of approximately 300 lines, which had been previously compiled manually, was merged with the list computed from the NIST data. The manually compiled list contained the strongest and most important lines of neutral species in the solar spectrum, and also included the data for lines of ionised atoms that has been computed to date. This served to fill any important gaps in the NIST line list.