The goal of our research is to advance on a broad front our understanding of stars, their atmospheres and synthetic spectra. The three principals in this research are Graham Hill (The University of Auckland), Saul Adelman (The Citadel) and myself. We have entitled our project Astrophysical GF Values but this more correctly is one of the goals, albeit the most ambitious. While we pursue that goal many other milestones have been and will be reached.
This ambitious project involves multiple stages. Two of the more important stages have been the development and testing of the programs CCDSPEC and STELLAR. CCDSPEC reduces the raw CCD frames efficiently and effectively, producing the very high quality spectra that we require for our goals. The development of this software has consumed much more of our time than expected but that process is now essentially complete. A brief report on its capabilities has been presented at the Astronomical Data Analysis Software Systems XI meeting in October 2001 (Gulliver and Hill 2002).
CCDSPEC has demonstrated that the scattered light in various spectrographs is variable in a systematic manner along the dispersion. In the high S/N regime in which we are operating, removing such systematic effects are critical for high quality analyses. This variation could explain the small differential adjustments that had to be applied to supposedly high confidence gf values in Vega.
Recent enhancements to CCDSPEC have prompted us to redo the reductions of all spectra acquired with CCDs. To date all files dating back to the beginning of 1998 have been reprocessed with signifiant improvements in the quality of the spectra. This reprocessing will continue back to 1996.
In its present form, STELLAR is also essentially complete. It can determine Teff, log g, V sin i, microturbulence, macroturbulence and scaled solar abundance in a new and more consistent manner. It does so by fitting, in any combination, the observed metallic line, hydrogen line, continuous flux spectra and any desired photometric indices. To perform the fits STELLAR uses a four-dimensional grid of SYNTHE synthetic spectra and ATLAS9 continuous energy distributions. These input files include as variables, Teff, log g, microturbulence and scaled solar abundance. The calculated model atmospheres range from Teff = 5000 K to 37000 K in steps of 500 and 1000 K, from log g = 2.0 to 5.0 in steps of 0.25, at microturbulences = 0, 1, 2 and 4 km/s and from [M/H] = -2.0 to +1.0, respectively. These models form the basis from which the huge synthetic spectra input files for STELLAR are generated for a given project. To our knowledge they form the largest collection of converged ATLAS9 model atmospheres in existence.
The successful use of astrophysical gf values was demonstrated in the modeling of Vega. These gf values were derived from o Peg for 4519 to 4540 A in a laborious manual process. Because they were derived only for this short stretch, analyses of other stars are still restricted to this interval. However a version of STELLAR has already been run to find the best gf value for a single test line. The remaining step is to automate this process for 1000’s of lines.
STELLAR has also been used to determine the fundamental stellar parameters for 95 stars of the UMa moving group (Gulliver et al. 2002) as part of the Hipparcos Guest Investigator Program. These stars range from A0 to K5 in spectral type. Using only fits to the metallic line spectrum from 4519 to 4540 A, most of the results compare well with those derived from photometric observations or other published values. Tests on Kurucz’s solar FTS atlas and on the stars of Hill (1995) and Hill and Landstreet (1993) have also shown good agreement.
The notable exception was in the mid F stars which give Teff too high by 500 K. F. Kupka has championed the use of the convection theory of Canuto & Mazzitelli rather than mixing length theory. We have calculated ATLAS9 model atmospheres with both theories. The models for CM convection indicate a 500 K cooler model for the early F star sigma Boo than MLT models. This may also resolve why ATLAS9 models give too high a temperature for Procyon.
Large blocks of time have been assigned to the Astrophysical GF Values Project on the DAO 1.2-m telescope. Over the last four years we have averaged some 50 clear nights per year. We have observed at high S/N (in excess of 750) spectra of several bright sharp-lined stars which will serve as primary stars: omicron Peg (A1 IV), a slightly enriched metallic line star; gamma Gem (A0 IV), a star which has nearly solar abundances; and, Procyon A (F5 IV-V), the canonical mid-F type dwarf with which we can eventually relate our studies to those based on the Sun. Given a satisfactory model even Vega (A0V) can be used. We have also added two sharp-lined early B stars to our program to sample other atomic species, gamma Peg (B2 IV) and iota Herculis (B3 IV). We also plan to add HD 161817, a sharp-lined prototype Field Horizontal Branch A star with a metallicity of 1/150 solar so that we can check more easily the oscillator strengths of the strongest lines, especially those of iron, by minimizing problems related to line broadening parameters.
The collection of spectral sections of these atlas stars continues with the present status as follows.
|gamma Gem||3860 – 4740||4905|
|Procyon||3860 – 4740||4905|
|iota Her||4630 – 4685|
|gamma Peg||4025||4135 – 4190||4355||4575 – 4685||4905||5015||5840||6600|
|o Peg||3860 – 4905||5125|
|Vega||3695 – 6200||6600||Wavelengths are central values.|
|Deneb||3830 – 5100||6600|
Abundance analyses based upon the measurement of the equivalent widths of o Peg, gamma Gem, and Procyon are underway. Deneb has been completed (Albayrak 2000). Again, we are finding many cases of lines which differ significantly and systematically in strength compared to those calculated from synthetic spectra using the mean abundances of the best determined lines of a given element. Thus, it has become clear that there are important errors in the tabulated gf values of even the strong lines of interest in astrophysical spectra.