Annu. Rev. Astron. Astrophys. 1981. 19: 77-113 Copyright © 1981 by Annual Reviews. All rights reserved

4.2. Emission Nebulae

Emission-line observations of HII regions (cf. Peimbert 1975) and mature supernova remnants have many advantages over stellar observations as a probe of current abundances in the gas phase of the ISM in remote galaxies, while planetary nebulae serve a similar function for intermediate-age stars for which, however, self-enrichment effects (often of great interest in themselves) need to be taken into account. The advantages are that emission lines due to the abundant elements H, He, N, O, Ne, S, Cl, and Ar (and C in the ultraviolet) are readily detectable at great distances and can be straightforwardly interpreted when a temperature-sensitive line ratio (usually O III 4363 / 5007) can be measured, although for some elements (notably S, Cl, and Ar) there are uncertain ionization corrections. This is usually the case for gas-rich Irregular galaxies and compact extragalactic HII regions, and for the outer parts of Scd spirals, in which the abundances of O, etc. are low (and the ionizing stars hot; see, for example, Shields & Tinsley 1976) leading to high electron temperatures that can be measured. For HII regions in earlier-type spirals and in the inner parts of Scd spirals, however, the abundances are larger and the ionizing stars cooler and more blanketed, and then some form of model fitting is required which involves serious uncertainties because of a shortage of critical data. However, Searle (1971) suggested that the large-scale trends shown by HII regions in the Scd galaxies M33 and M101 ([O III] / H and H/[N II] increase by very large factors as one goes outwards from the nucleus; cf. Aller 1942, 1956) are due to a large-scale radial abundance gradient, with O and N abundances having low values at the periphery and increasing towards the center, and this has been confirmed by Smith (1975), who measured electron temperatures in the outer regions of these galaxies, and in the case of M101) by Shields & Searle (1978) who were able to fit a fairly unique model to the inner HII region S5 by including the [S III] lines near) 9000 and found 12 + log(O/H) = 9.1 + 0.2 for this object, contrasted with a value of 8.13 ± 0.09 for the outermost HII region they observed. This result, together with confirmatory model calculations (e.g. Collin-Souffrin & Joly 1976, Stasinska 1978), has inspired a number of authors to use various strong-line ratios (for a list of data and correlation diagrams see Alloin, Collin-Souffrin & Joly 1979) as a direct indicator of electron temperatures and/or oxygen abundances (one implying the other), e.g. [O III] / H (Jensen, Strom & Strom 1976), [O III] / [N II] (Alloin et al. 1979), and ([O III] + [O II]) / H (Pagel et al. 1979), and a tentative calibration of all three line ratios has been given by Pagel, Edmunds & Smith (1980). The uncertainties in such methods are important. For example, the calibration of the high-abundance end (based on model calculations extrapolated from S5 by Dufour et al. 1980) would be lowered in the presence of internal dust (Sarazin 1976, 1977); scatter can be introduced by nonuniform geometry (Stasinska 1980) and by departures from the underlying assumption that, for giant HII regions at least, the ionizing radiation field is uniquely related to the oxygen abundance. The uncertainties in these "guessing" methods have been discussed by Stasinska et al. (1981); we estimate a standard error for them of about ± 0.2 dex, over and above the systematic uncertainty at oxygen abundances exceeding that of S5.

The application of modern, detailed shock models to spectra of mature supernova remnants (Dopita 1976, Dopita, Mathewson & Ford 1977, Dopita, D'Odorico & Benvenuti 1980) provides a valuable alternative source of data for abundant elements in the ISM because the uncertainties in the shock parameters, while significant, are entirely different from those arising in the analysis of HII regions; in particular, the [N II] / H ratio is quite insensitive to shock conditions and it is easy to recognise the few cases affected by self-enrichment. In Figure 2 we compare oxygen and nitrogen abundances deduced for HII regions and SNRs as a function of radial distance in M33; the agreement for N/H is remarkably good.

Figure 2. Oxygen and nitrogen abundances in HII regions (filled squares) and supernova remnants (open squares) as a function of radial distance from the center of M33. The HII region data are those of Smith (1975), analyzed using his electron-temperature data (for the outer HII regions) and the ([O II] + [O III]) / H calibration of Pagel, Edmunds & Smith (1980); error bars represent either the discrepancy between the two or the estimated error of the calibration. The SNR data are from Dopita, D'Odorico & Benvenuti (1980) with error bars corresponding to the range of results deduced from various shock models.

Planetary nebulae have been observed in several galaxies of the Local Group (see Webster 1978, Ford 1978, Dufour & Talent 1980) and are usually hot enough that the electron temperature can be measured, apart from technical difficulties caused by their smallness at great distances. Peimbert (1978) has classified galactic planetaries into four types: I, massive with He, N enhanced; II, intermediate Population I with N slightly enhanced (but cf. Kaler 1978); III, high velocity with composition similar to II; and IV, halo objects with O and Ne deficient by an order of magnitude. Carbon exceeds oxygen in many planetaries (see, for example, Aitken et al. 1979 and references therein), so that He, C, and N are all likely to have been affected by self-enrichment to varying degrees, although there is no reason to believe that O and Ne are affected (see, for example, Kaler 1978, 1980). Magellanic Cloud planetaries have been discussed by Webster (1978) and Aller et al. (1980), among others, and show enhancements of N and perhaps He relative to HII regions; their usually normal oxygen abundances suggest that the majority of observed systems are quite young. In some cases the nitrogen enhancement is extreme, and a similar object in NGC 6822 has nitrogen enhanced by a factor approaching 200 implying primary production (Dufour & Talent 1980). We feel that planetary nebulae are not suitable objects for the study of abundance gradients in nitrogen.