Annu. Rev. Astron. Astrophys. 1984. 22: 37-74 Copyright © 1984 by Annual Reviews. All rights reserved

2.3 Abundances

The metallicities of the ionized gas (i.e. abundances of O, Ne, and N; for a review, see 268) have been determined from the emission-line ratios under the assumption of photoionization (4, 30) for a large number of high surface brightness Irrs (120, 191, 209, 216a, 226, 340, 381) and low surface brightness, dwarf Irrs (cf. 189, 226, 267, 270, 329, 340, 361). The abundances generally range from SMC-like to LMC-like, with no distinction between the high and low surface brightness systems. Only some of the dwarf blue compacts or ``intergalactic HII regions'' seem to be much more metal poor than the SMC (4, 120, 209, 216a, 226, 377), and no systems are known with emission spectra consistent with extreme metal-poor galactic Pop II abundances. The lower half of Figure 3 shows the number distribution of Irrs as a function of O/H for galaxies in the above references (including blue compact systems and intergalactic HII regions). Generally, O/HIs 1-4 x 10^-4, where the solar value is ~ 7.6 x 10^-4. Interestingly, several very luminous intergalactic HII regions have been observed at moderate redshifts and are found to have normal metallicities for Irrs; for example, B234 at redshift 0.06 (121) and B272 at redshift 0.04 (88) both have O/H ~ 2 x 10^-4. Some very luminous blue compact galaxies and the unusual M82 system, however, may approach or even exceed solar metallicity levels in the gas (36, 132, 200, 257).

Figure 3. (lower panel) The distribution of Irr galaxies as a function of the oxygen abundance in HII regions is displayed. This sample includes low and high surface brightness Magellanic Irrs, luminous Irrs, blue compact systems, and intergalactic HII regions (see text for references). Gas abundances in Irrs, with few exceptions, are similar to the Magellanic Clouds. (upper panel) Dependence of oxygen HII abundances on gas mass fraction is shown in terms of µ = Mgas(Mgas + Mstars)-1. Simple chemical evolution models predict a linear relationship between O/H and ln (1 / µ), which is not found in this sample or in more traditional plots where µ = Mgas / Mdyn (238). All galaxies from the lower panel with sufficient data to enable µ to be estimated are included. Symbols are as follows: x, high surface brightness normal Irrs: , low surface brightness dwarfs; –, luminous and blue compact Irrs.

There are difficulties in measuring abundances of Irrs from emission lines. Often [O III] 4363 is not strong enough to be accurately measured, so some other means must be found for determining the electron temperature, such as the empirical relationship between forbidden oxygen and hydrogen Balmer emission intensities developed by Pagel et al. (266). This can lead to an ambiguity in the derivation of O/H for low-abundance objects (189). For example, consider the very metal-poor object I Zw 18 (120). The [O III] 4363 ratio gives a temperature of 16,300 K, whereas simple application of the relationship determined by Pagel et al. gives 6400 K. The former results in an O/H value of 0.2 x 10^-4 and the latter in a value of 5 x 10^-4. A plot of [O II] / [O III] intensity vs O/H (189) appears to be able to resolve this ambiguity, since ionization levels and abundances are well correlated in HII regions; however, one must keep in mind that the errors in even routine emission-line abundance determinations can be large. For example, of the 45 overlapping observations in the references cited at the beginning of this section, 15 pairs agree within 20%, 5 within 50%, while 6 are off by factors of two or more.

A further complication in using emission lines as abundance indicators in more distant galaxies where individual HII regions are not resolved stems from anomalous inter-HII emission-line ratios found by Hunter (190). In nearby Irrs, the [O II] / [O III] intensity ratio from diffuse emission considerably exceeds the ratio for HII regions within the same galaxies. Thus, integral emission-line flux ratios will differ from those of typical HII regions. The origin of anomalous line ratios is unclear and could involve either shocks or some type of photoionization process. It is also uncertain whether these anomalous emission regions are similar to the diffuse H emission seen in late-type spirals (246) and the Milky Way (280).

Since Irrs do cover a range in gas metallicity, they have proven valuable in efforts to determine the pregalactic helium abundance level, which is an important constraint on cosmological models (see 14, 259, 268, 389). For example, an upper bound on the primordial helium abundance is provided by the most metal-poor intergalactic HII regions, and studies of larger samples of Irrs can yield insight into the variation of helium enrichment by stars as a function of metallicity, which allows an extrapolation to be made to zero metallicity (for excellent reviews, see 210, 216a, 265). Considerable effort therefore has been devoted to enlarging the sample of very metal deficient emission-line galaxies (cf. 215), but only limited success has been achieved thus far in that I Zw 18 remains the most metal-poor emission-line galaxy.

The metallicities of stars in Irrs seem to be consistent with the moderately metal-poor nature of the gas. Star clusters in the LMC have been found to have metallicities of 1/2 to 1/40 of the Sun (cf. 60, 168). The SMC, while on the average more metal poor than the LMC, is not known to contain objects as metal poor as the most metal-poor objects in the LMC or the Galaxy (127, 128), and thus it evidently has a smaller spread in abundances. Finally, JHK colors of Irrs are near those of several-billion-year-old intermediate metallicity systems (349). If gas captures from external sources are a common occurrence, e.g. among star-burst Irrs, then it is possible that gas and stellar metallicity levels may not mesh in some systems. This phenomenon has not to our knowledge been observed, but data on stellar abundances are still very sketchy.

The relatively low abundances of the Irrs tell us that these systems are less evolved than most spirals, in the sense that the interstellar gas is less processed. Less processing, however, does not mean that the returning metals are not mixed throughout the system. Spectrophotometry of individual HII regions (cf. 191, 266, 340, 381) shows that the O/H and N/S abundance ratios are remarkably constant throughout the disks of Irrs. By whatever process, the metals in these galaxies are fairly well mixed over at least the optically prominent regions.

Correlations between the gas abundances and other global parameters would be expected to give us some clue as to why this particular morphological type of galaxy (the Irrs) would be less evolved and better mixed than most spirals. As already mentioned, except for the intergalactic HII regions, which are the most metal poor, the abundances do not separate according to the luminosity of the systems. There also do not seem to be any compelling relationships between metallicity and current stellar birthrate or galaxian spatial dimensions, and thus our hopes for an obvious clue to evolutionary processes go unrewarded.

The closed-system model with the instantaneous recycling approximation provides the simplest and most widely used galaxy chemical evolution model (268, 312, 339, 353). In this model, the metallicity of the gas is given by Z_g = y ln (1 / µ), where y is the stellar heavy element yield and µ is the gas fraction, i.e., mass in gas per unit total mass (stars plus gas); thus a correlation between Z_g and gas fraction should exist. Since the Irrs are evidently well approximated by a single zone, they provide a good test for the applicability of the basic chemical evolution model to real galaxies. The upper half of Figure 3 is a plot of ln (1 / µ) against the abundance parameter O/H for high surface brightness Irrs (189, 191, 226, 340), for low surface brightness Irrs (189, 226, 340), and for luminous Irrs and blue compact systems (4, 120, 209). The parameter µ is computed for all objects in the manner described by Hunter et al. (191), where the gas mass is taken to be 1.34 times the total hydrogen mass and the mass in stars is estimated using the UBV colors to determine the M_star / L_V ratio from Larson & Tinsley's (220) stellar population models. Thus, the total mass is the mass in gas plus stars, rather than the often-used dynamical mass, which could include contributions from matter that does not partake in the chemical evolution. Clearly, no correlation is evident in Figure 3, in agreement with the similar study by Matteucci & Chiosi (238). Thus we see that while simple chemical evolution models provide useful qualitative insights, they do not quantitatively fit the observations.

There are several possible ways to modify the basic model. For example, in using the closed-system model one must consider over what region does the galaxy really behave as a closed system. Complications along these lines could include gas infall from the outer parts of the galaxy (55, 126, 238), gas that does not actively participate in the galaxy's chemical evolution (188), or ejection of metals from star-forming regions (146). In loosely bound galaxies, material ejected due to supernovae could be lost from the system entirely or rain back down on other regions of the galaxy (115, 346). These types of processes may allow us to understand the empirical correlation that exists between metallicity and dynamical mass in dwarf Irrs and extragalactic HII regions (191, 226, 238, 340). Thus it may well be that total mass is equally or more important than relative gas fraction in controlling Z_g. In this case the simplest closed-system models would not apply, and more specific models will be needed instead.