The need for accurate spectroscopic data in the VUV has become particularly acute with the launch of the Hubble Space Telescope (Leckrone et al. 1990; Leckrone et al. 1993). Elemental abundance studies on stars and on the Interstellar Medium require improved transition probability or oscillator strength (f-value) data. The absorption method is one of the most accurate methods for relative oscillator strength measurements on neutral atoms, especially when applied to gas phase samples in an oven (Blackwell et al. 1979). Although it will be difficult to match the 1% accuracy of the best experiments on neutral atoms in ovens, experiments using the absorption method on ions in plasmas are quite promising. Recently Bergeson et al. (1996b) described a VUV absorption experiment which has unique capabilities. This VUV absorption experiment uses the Aladdin storage ring at the Synchrotron Radiation Center as a continuum source, a hollow cathode discharge (HCD) as an absorbing sample, and a large (3 m) echelle spectrograph equipped with a state of the art charge coupled device (CCD) detector array. This experiment is sensitive, i.e. it can be used to measure small equivalent widths. Its sensitivity is due to: the high spectral radiance provided by the storage ring, the high () resolving power provided by the echelle spectrograph, and the powerful advantages of a detector array in absorption spectroscopy (Wamsley et al. 1993). The experiment is broadly applicable to many neutral atoms and atomic ions due to the use of a HCD as an absorbing sample.
This work is an extension and improvement of the first results from the high sensitivity absorption experiment at the University of Wisconsin-Madison (Bergeson et al. 1996b). Although the use of a HCD as an absorbing sample provides great flexibility for studying lines from many different metastable lower levels in many atoms and ions, it does have two minor disadvantages. The lower level populations in the HCD are not in local thermodynamic equilibrium (LTE), as are lower level populations in an oven. This minor disadvantage is overcome by using a larger set of reference lines with accurately known f-values (Bergeson et al. 1996a) to normalize relative f-value measurements on pairs of lines from a common lower level. The other minor disadvantage is due to the fact that ions in the negative glow region of a HCD often do not have a perfectly Maxwellian velocity distribution. Diffusive cooling tends to deplete the high energy tail of the Maxwellian distribution and thus affect the curve of growth. Extensive measurements in this work of the effects of diffusive cooling of ions in the negative glow of the HCD allow us to construct a satisfactory curve of growth for use in data analysis. A satisfactory curve of growth is generated by assuming a full Voigt profile with an artificially low ion temperature in some cases. We present the first laboratory measurement of 6 absorption f-values for VUV lines of the multiplet (#8) of Fe II. These lines have been identified in solar spectra and have been observed in the ISM using the Goddard High Resolution Spectrograph on board the Hubble Space Telescope (Fawcett 1988; Cardelli & Savage 1995). The absolute scale for the measurements is determined by work on branching fractions done previously at Wisconsin & Lund (Bergeson et al. 1994, 1996a), and by work on radiative lifetimes using the laser-fast beam method (Biemont et al. 1991; Guo et al. 1992). The combination of these highly accurate measurements of branching fractions and radiative lifetimes yields 67 f-values of UV transitions in Fe II; 33 of the f-values are accurate to 5% or better, and 29 more are accurate to within 10% (Bergeson et al. 1996a). The UV transitions with small () uncertainty serve as the primary reference lines in this study.