Annu. Rev. Astron. Astrophys. 1994. 32: 115-52 Copyright © 1994 by Annual Reviews. All rights reserved

Chemical Abundances

Chemical abundance determinations are now of a quality and number to enable relationships with other galaxian parameters to be described. This has been done for such quantities as luminosity (Skillman et al. 1989), mass (Garnett & Shields 1987; Vila-Costas & Edmunds 1992), and type. Figure 7 displays values of O/H as derived from emission line strengths measured in H II regions, as a function of type (upper panel) and of luminosity (lower panel). The trend is in the sense that the fainter, less massive and later-type galaxies are lower in specifically O/H.

Figure 7. O/H abundance versus morphological type (upper) and absolute magnitude (lower). Different symbols represent different sources: filled circles: Oey & Kennicutt (1993), plus signs: Garnett & Shields (1987), open circles: Skillman et al. (1989). The absolute magnitudes are as given in these references.

Using absorption lines and colors, a trend for Fe/H as a function of luminosity is well described for E's and S0's, in the sense of a lower Fe value in fainter galaxies (Pagel & Edmunds 1981; Da Costa 1992). With a reasonable assumption regarding the O/H and Fe/H abundances, Skillman et al. (1989) show that the spheroidals and dwarf elliptical galaxies form a smooth extension of the faint end of the abundance-luminosity relation found for spirals and irregular type galaxies.

Intercomparison between abundances based on emission lines from H II regions and on absorption lines and colors from a stellar population is uncertain. The approaches and assumptions differ significantly (e.g., Pagel & Edmunds 1981). Atomic processes guide the emission line analysis while the absorption lines must be evaluated in terms of both the composite stellar population, and metallicity using one of two methods: evolutionary synthesis or stellar population synthesis (e.g., by Rocca-Volmerange 1992, Chiosi et al. 1992, and Bruzual 1992).

Even within the spiral class where there is a systematic trend of H II region size and luminosity, the emission-line analysis can be type dependent. For early spirals, the widely-used temperature indicator [OIII]4363 line is often too weak to use. Lacking such data, an ``empirical'' method is applied in which the abundance is derived from ([OII]3727 + [OIII]5007)/H and its calibration versus [O/H], a procedure which may be uncertain in the high abundance regime. These problems and the derivation of abundances for early-type galaxies are discussed by Diaz (1992) and by Oey & Kennicutt (1993).

Although different techniques and often large uncertainties are encountered, the following consistent picture emerges:

1.	The least luminous and least massive of galaxies, Im and dE, have the lowest abundances (O/H or Fe/H), 10-1 to 10-2 solar. The most massive and most luminous of galaxies, both ellipticals and (the central regions of) spirals are overabundant by factors of up to a few (Strom & Strom 1978; Aaronson & Mould 1985; Worthey et al. 1992). Gorgas et al. (1990) suggest that, in the mean for their sample of E's and S0's, the factor is one.
2.	When measurable, most galaxies including our own show an abundance gradient in the sense of a lower metallicity at the outskirts, i.e., at larger galactocentric radii (Diaz 1992; Vila-Costas & Edmunds 1992). There appears to be a type dependence for these gradients. Early-type spirals have shallower [O/H] gradients than do late-type systems (Vila-Costas & Edmunds 1992; Oey & Kennicutt 1993). Gradients within E's and S0's show a wide range and appear not to be correlated with any galaxian parameter (Gorgas et al. 1990).
3.	An abundance-type effect is present among spirals and irregulars in the sense that earlier galaxies have a higher abundance. This trend reaches an approximately constant value among the early spirals. The entire effect may only reflect a luminosity or mass distribution in the highly selected sample of galaxies thus far studied for abundances (Garnett & Shields 1987; Skillman et al. 1989; Oey & Kennicutt 1993).

There are exceptions to these various findings. Bertola et al. (1993) describe a metal-poor luminous early-type galaxy NGC 5018. They find that the inner region has a relatively blue color together with a weak Mg₂ index, both measures of low metallicity. Shields et al. (1991) describe the first sample of cluster galaxies studied for abundances. Here they find higher interstellar abundances in Virgo cluster Sc's than for field galaxies of similar type.

A type-abundance dependence is present only in the broadest of terms. More fundamental may be the abundance trend with luminosity and with mass over the range from dwarf galaxy to giant (classical) galaxy.