5 Applications and conclusion

With judicious application, the data presented in this paper may be used to determine for the first time the abundances from Ndiii in CP stars. This is important in corroborating the abundances of this element derived from Nd⁺, typically the dominant species in the atmospheres of cooler CP stars. Such calculations may also provide evidence of unusual atmospheric conditions related to element stratification and/or non-LTE effects in cases of substantial disagreements between abundances derived from different stages of ionization (Proffitt et al. [1999]). Ndiii has been observed in the spectrum of several CP stars having strong lines of Ndii, for example HR465, HD51418, and HD200311 (Cowley [1976]; Aikman et al. [1979]); Ndii and Ndiii have also been found in the Si star HD192913 (Cowley & Crosswhite [1978]; Ryabchikova et al. [1990]). We have previously reported on Ndiii in HR6870 and $\gamma$ Equ (Cowley & Bord [1998]), and recent studies by our group have also revealed the strong presence of both Ndii and Ndiii in the spectrum of HD101065 (Cowley et al. [2000]), a rapidly oscillating Ap (roAp) star and arguably one of the most chemically bizarre objects in the Galaxy.

We have applied the oscillator strength data presented herein in the analysis of this star using the Michigan spectrum synthesis programs (Cowley [1996]). Adopting a model stellar atmosphere with $T_{{\rm eff}}$ = 6600 K and log(g) = 3.5 which provided acceptable fits to the Balmer line profiles in high resolution ( $R = \lambda/\Delta \lambda = 80\,000$), high signal-to-noise (S/N $\geq$ 200) echelle spectra, we find a logarithmic abundance of Ndiii of $5.2\pm0.4$ on the usual scale where log(H) = 12.0; this result is based on 5 lines longward of 5000 Å that were free of significant blends and had wavelengths in reasonable agreement with predicted values ( $\Delta \lambda \leq 0.04$ Å). This value is about 5000 times the solar abundance of Nd, but it comports well with the abundance of this element derived from our studies of the Ndii spectrum in this star, viz. $4.9\pm0.2$dex based on 14 unblended lines with wavelength differences with respect to the laboratory values of $\leq 0.02$ Å. These calculations assumed a microturbulent velocity $\xi_{\rm t} = 1.0$ kms^-1. These results differ from those published by Cowley & Mathys ([1998]) in part due to differences in the model stellar atmosphere adopted, but also because of the use of a corrupted data file in the earlier analysis (Cowley [1999]); for additional details, the reader is referred to Cowley et al. ([2000]).

HD101065 possesses a weak surface magnetic field of $\approx 2500$ G(Wolff & Hagen [1976]; Cowley & Mathys [1998]) which has not been explicitly taken into account in our analysis; attempts to introduce magnetic intensification effects by raising the the microturbulence to 2.0 kms^-1 reduce the abundance of Nd in HD101065 by about 0.7 - 0.8 dex.

Ryabchikova et al. ([1999a]) have also identified Ndii and Ndiii in the spectra of several other roAp stars; an analysis of one such object, HD122970 (Ryabchikova et al. [2000]), using our g-factors and gf-values, has demonstrated that this cool ( $T_{{\rm eff}} = 6930$ K), weakly magnetic star is over abundant in Nd by a factor of about 275 relative to the standard abundance distribution (SAD, Grevesse & Sauval [1996]). Her calculations give log( $N_{{\rm Nd}}/N_{{\rm Total}}$) = $-8.0\pm0.3$ from Ndiii assuming $\xi_{\rm t} = 0.85$ km s^-1.

In summary, we have presented gf-values and Landé g-factors for the RE ion Ndiii for a number of lines in the optical region with accuracies sufficient to facilitate abundance work (and certain other types of stellar atmosphere calculations) in CP stars. These results have been applied in the analysis of HD101065 and yield abundances with uncertainties consistent with the errors in the log(gf) values, plus an additional 0.1-0.2 dex attributable to the choice of parameters used in constructing the stellar model atmosphere and in treating line broadening (e.g., rotation, instrumental and Zeeman effects, hfs, etc.) in this very unusual star.

The gf-values reported here are not generally comparable in precision to the best ones available from highly sophisticated multi-configuration Dirac-Fock approaches (cf. Brage et al. [1999]), or from enhanced Cowan-code calculations conditioned by new, extensive energy level data (cf. Palmeri et al. [2000] on Priii), where uncertainties as low as 15% ( $\approx \pm0.06$ dex) are attainable. The former type of calculation, because of its complexity and the number of active orbitals required for ions like Ndiii, is not likely to be carried out except perhaps for a small number of lines of special importance. The latter method, which depends on a significant expansion of the energy level structure, requires an extended analysis of the spectrum (which for Ndiii is now underway [Wyart [2000]]) for its successful implementation. For global abundance studies, data of the type and quality given here will thus remain useful until additional laboratory measures are made and/or more extensive theoretical work is performed. We encourage both avenues of research to be pursued.

Acknowledgements

The author expresses his appreciation to Dr. Tanya Ryabchikova for her continuing encouragement to make these data available to broader community of workers in this field, and for her generous provision of the results of her studies of the abundances of Ndiii in roAp stars in advance of their publication. Additional thanks are due to Dr. Charles R. Cowley for his assistance with the REE abundance analysis of HD101065. The author also acknowledges helpful correspondance with Dr. E. Biémont, Dr. J.-F. Wyart, and Dr. R.D. Cowan during the preparation of this manuscript for publication.