The abundance ratios of heavy elements are sensitive to the initial mass function (IMF), the star formation history, and variations in stellar nucleosynthesis with, e.g., metallicity. In particular, comparison of abundances of elements produced in stars with relatively long lifetimes (such as C, N, Fe, and the s-process elements) with those produced in short-lived stars (such as O) probe the star formation history. Below, I review the accumulated data on C, N, S, and Ar abundances (relative to O) in spiral and irregular galaxies, covering two orders of magnitude in metallicity (as measured by O/H). The data are taken from a variety of sources on abundances for H II regions in the literature.
Helium, the second most abundant element, has significance for cosmology and stellar structure. Most 4He was produced in the Big Bang, and the primordial mass fraction Yp is a constraint on the photon/baryon ratio and thus on the cosmological model. The He mass fraction also affects stellar structure, but He is difficult to measure in stars and so must be inferred from other measurements. On the other hand, He I recombination lines are relatively easy to measure in H II regions, and so a large amount of data is available on He/H in ionized nebulae.
A great deal of effort has been spent in determining Yp, and is covered in Gary Steigman's contribution, so I will be brief on this aspect. Peimbert & Torres-Peimbert (1974) initiated the current modern study of Yp by making the simple assumption that the He mass fraction varies linearly with metallicity (or O/H), and thus used measurements of abundances in H II regions with a range of O/H to extrapolate to the pre-galactic He abundance at O/H = 0. Today there is very high signal/noise data on He abundances in approximately 40 metal-poor dwarf irregulars with O/H ranging from 2% to 10% solar, so the extrapolation to O/H = 0 can be estimated to high statistical precision.
The good news is that the best current estimates of Yp agree to within 5%. This is amazing agreement for measurements derived from spectroscopy of distant galaxies, so we should all feel proud. Nevertheless, the differences in Yp estimates are a source of consternation for theory of cosmological nucleosynthesis, as the two largest studies obtain values for Yp which disagree at the 4-5 significance level: Olive, Skillman & Steigman (1997) derived Yp = 0.234±0.003, while Izotov & Thuan (1998) derived Yp = 0.244±0.002 (statistical uncertainties only for both studies), from similar-sized H II region samples. Depending on which estimate is considered most reliable, Yp either agrees with the best current estimate of D/H under standard Big Bang nucleosynthesis (for Yp = 0.244, or it does not.
At present the battleground for Yp is focused on sources of systematic error, and these are likely to yield the greatest improvements in He measurements, rather than measuring more data points. The areas that need work are:
Each of these error sources contribute perhaps 1-2% to the uncertainty in derived He abundances, but it is how they sum that determines the systematic error, which is not fully understood.
Helium abundances in spiral galaxies are less well-determined, because of more uncertain electron temperatures. The He abundance does have an effect on ionization structure, so it is of interest to know how He/H varies with metallicity in the inner disks of spirals. Does He/H continue to rise linearly with metallicity as in the metal-poor galaxies, or does it level off? This may be difficult to determine from H II region spectroscopy, as there appears to be a drop in the He I line strengths in the inner disks of spirals (Bresolin, Kennicutt, & Garnett 1999), contrary to what one would expect from ionization models with rising or even constant He/H (Figure 15). The low He I 5876 line strengths in the metal-rich regions observed so far are lower than expected even for primordial He/H. The trend of decreasing He I line strength is best explained if ionizing clusters in H II regions with metallicity above solar have radiation fields with characteristic temperatures of about 35,000 K. This is the strongest evidence available for a possible variation in the upper limit of the massive star IMF.
Figure 15. I(He I 5876) / I(H ) vs. R23 from spectroscopy of H II regions in spiral galaxies (Bresolin et al. 1999), showing how the He I line strengths decrease for low R23 (high O/H). Overplotted are trends obtained from photoionization models with various values of Teff for the ionizing stars, assuming that He varies with metallicity as Y / = 2.5.