Astrophysical gf Values
Astronomers require precise and accurate atomic line parameters such as oscillator strengths to study and model a wide range of astrophysical problems. The determination of stellar magnetic field strengths, the measurement of atmospheric hydrodynamics (such as diffusion, turbulences, and convection), and the study of stellar physical parameters (such as Teff and log g) are a few examples. Ideally one would prefer to use precise and accurate data obtained in the laboratory by atomic physicists. Unfortunately the large number of lines contributing to stellar spectra, of order 10,000 in the optical region of an A type star, is far too many to be measured by them. Theoretical values often have good relative strengths within multiplets (when this concept makes physical sense) but overall the values may have to be renormalized by a factor as large as an order of magnitude. Moreover for intercombination lines the situation may be worse.
Thus there arose the astrophysical tradition of obtaining excellent quality spectra of stars (e.g. Arcturus by Peterson et al. 1993), and particularly the Sun (Thevenin 1989, Thevenin 1990), to be used as sources for determining the needed atomic data, especially the oscillator strengths. As stellar photospheres are often hotter than the laboratory sources, many lines observed in stars are difficult to analyze with terrestrial sources. By combining available laboratory data and theoretical calculations with new astrophysical oscillator strengths it is possible to obtain an internally consistent set of values with good absolute accuracy. We have already shown how improved oscillator strengths derived from the study of a limited wavelength region in o Peg improved the analysis of Vega. New reliable gf values will enable us to reduce the errors in the analyses of normal and peculiar B, A, and F stars and allow us to better interpret the abundances in terms of galactic evolution and photospheric processes. More importantly, it will allow us to use spectrum synthesis techniques with greater confidence.
The importance of being able to obtain reliable results from the spectra of broad-lined stars cannot be overstated. It is crucial to our understanding of galactic evolution and the hydrodynamics of stellar photospheres to be able to systematically and confidently obtain abundances for cluster stars in general, not just the few that happen to have sharp lines. Analyses of the F-stars are supposed to yield the abundances out of which the cluster stars formed since convection is thought to have prevented diffusion from altering the surficial abundances. With these abundances and ages, one can study galactic chemical evolution. Many questions about how abundance anomalies arise cannot be answered by studying field stars where the uncertainties in age and initial composition make it almost impossible to study the dependences of the derived abundances on parameters such as age, rotation, temperature, surface gravity, microturbulence, and the like.
To improve the gf values of atomic lines the use of spectra of several stars is very desirable. Each individual star can be expected to have its peculiarities including undetected binarity, magnetic fields, and rotational effects. Although non-LTE effects are expected to be small, they must be taken into consideration also. Computing such effects for the light and iron peak elements has become relatively routine with the data from the Opacity and Iron Projects. This is an area where Hubeny is an expert.
To obtain astrophysical gf values we must establish the Teff, log g, microturbulence, and initial abundance estimates for each star. With these parameters, we will calculate synthetic spectra for comparison with the observations. We will ascertain which features are the most blend-free and refine our initial abundances by using a small subset of lines with the highest quality laboratory gf values and minimal blending. Fuhr will assist us using the NIST database of gf values which contains mainly light and iron peak elements. Cowley and Ryabchikova will do the same with lines of the rare earth elements. Then we will adjust the gf values for the other lines. Intercomparison among the standards will be necessary. We will check our values using standard atomic physics tests.
After we have completed our analyses we will make our data available to the community through the Astronomical Data Centers and other means. The internet will allow us to post the spectra in an interactive form as for Deneb on my webpage. Vega and o Peg will follow shortly. With the new oscillator strengths and spectral atlases, observers will have the ability to pursue a ‘new generation’ of abundance analyses of stars, both in the field and in clusters while theoreticans should be able to improve our understanding of galactic chemical evolution and stellar hydrodynamics. By comparing our atlas of Procyon with that of Griffin using spectra from the 2.5-m Mt. Wilson coude, we will be able to assess instrumental differences. This type of analysis needs to be
done between different pairs of high dispersion instruments. Our data have a considerably greater S/N ratio and they show weaker lines. The existing stellar atlases for the optical region are based on photographic plates. Using CCD (and Reticon) spectra, ours will be at a much higher S/N ratio and should have a much more reliable intensity scale.
The new astrophysical gf values will reduce the systematic errors in abundance analyses, direct the atomic physicists to those species needing improved laboratory observations, and result in synthetic spectra which match more closely the observed spectra.
An important component of our strategy has been recently added; Adelman has been awarded an NSF grant of US $180,000 to build and operate a stellar spectrophotometer. This device is expected to be in place on an automatic (spectro)photometric telescope in either Arizona or California within about one year. The instrument design is complete and negotiations are underway with two observatories. The telescope will have 50% of its time dedicated to our spectrophotometric observations. Data will be reduced using CCDSPEC. Using STELLAR, the spectrophotometric observations will enable us to determine more accurate stellar Teff and log g values.