4. ABUNDANCE PATTERNS IN DWARF GALAXIES

One outstanding question concerning abundances in dwarf galaxies is the variation of abundance within a galaxy. Since star formation in dwarf irregulars is thought to be episodic, and the solid-body nature of the rotation inhibits efficient mixing, it seems plausible that there might be large variations in abundance from region to region. However, the best studied irregular galaxies (SMC & LMC: Peimbert & Torres-Peimbert 1974, 1976; Dufour 1975; Dufour & Harlow 1977; Pagel et al. 1978; NGC 6822: Pagel, Edmunds, & Smith 1980; Sextans A: Skillman, Kennicutt, & Hodge 1989) indicate that the dispersion in oxygen abundance is generally quite small ( 25%) and consistent within observational errors. These observations support the picture discussed in Section 2.3 - that the newly synthesized elements are returned globally and not locally.

Figure 4. (upper) O/H versus N/O abundances at 18 positions in NGC 4214 using wide (12.5") apertures. Solid circles and stars denote regions near Knot 5 which show higher O/H abundances and lower N/O ratios (0.1 dex) than the rest of the surveyed region. Filled triangles/squares illustrate the effect on the derived abundances when the adopted Te / c(H) is artificially displaced by 1000 K/0.1. The solid line is a least squares fit treating O/H as the independent variable. (lower) Log(N/H) versus 12 + log(O/H). No evidence of N/H variations is seen. (From Kobulnicky & Skillman 1996).

However, Kunth & Sargent (1986), in an attempt to explain why no galaxies more metal-poor than IZw18 (2% of the solar value; Skillman & Kennicutt 1993 and references therein) are seen, suggested that H II regions may "pollute" themselves on short (10⁶ yr) timescales with nucleosynthetic products from the current burst of star formation. Although it seems certain that the heavy elements produced inside the short-lived, massive stars must contribute to the enrichment of the interstellar medium, the spatial and temporal scales on which this happens are poorly known.

Understanding element enrichment processes is especially important for interpreting the chemical evolution of low-mass, low-metallicity galaxies which have been used to derive primordial helium abundances and study the effects of supernova-driven galactic winds. Pagel et al. (1986) noticed that galaxies with spectral signatures of Wolf-Rayet (W-R) stars often had higher He and N abundances than other galaxies at similar metallicity, as measured by the O/H abundance. They suggested that N- and He-rich Wolf-Rayet star winds enrich the nebulae on timescales short compared to the lifetime of the H II region (< 10⁷ yr).

Kobulnicky (1997) has made a concentrated observational effort to test whether HII regions are polluted by their exciting clusters. A recent optical spectroscopic study of the Magellanic irregular NGC 4214, which contains multiple starburst knots of different ages and varying W-R star content, revealed no N or He enrichments and no abundance fluctuations in proximity to young stellar clusters (Kobulnicky & Skillman 1996, KS96). Figure 4 is taken from that reference and shows the high degree of homogeneity in the abundances in NGC 4214. The oxygen abundances all fall within a range of about 0.15 dex (including observational errors) and the nitrogen abundances show a very similar range. Additionally, KS96 reanalyzed abundance data from the literature and found no systematic differences between galaxies with W-R star features and those without such features.

Kobulnicky & Skillman (1997) conducted similar observations of NGC 1569 and found similar results (supporting the finding of Devost et al. 1997 based on empirical abundance measurements). Despite all of the evidence pointing to NGC 1569 being a post-starburst galaxy (e.g., Israel & DeBruyn 1988), there is no evidence of abundance anomalies in the vicinities of the aging clusters of massive stars. No localized chemical self-enrichment ("pollution") from massive star evolution is found, even though the data are sensitive to the chemical yields from as few as two or three massive stars.

Kobulnicky (1997) argues that under the assumption that the stellar clusters are depositing their freshly created oxygen locally in the ISM, strong chemical signatures in the surrounding interstellar material should be detected unless one or more of the following are true: 1) Different star forming regions throughout the studied galaxies "conspire" to keep star formation rates and global abundances uniform at all times, 2) ejecta from stellar winds and supernovae are transported to all corners of the galaxy on timescales of < 10⁷ yr, and are mixed instantaneously and uniformly, or 3) freshly synthesized elements remain unmixed with the surrounding interstellar medium (possibly residing in a hard-to-observe hot 10⁶ K phase or, perhaps less likely, a cold, dusty, molecular phase). This reasoning supports the hypothesis that the newly synthesized elements are predominantly returned to the hot ISM where they can travel significant distances before cooling and becoming visible in the warm and cool phases of the ISM. This implies that the instantaneous recycling approximation often used in galactic chemical evolution modeling may not be generally applicable, even for oxygen!

Figure 5. H (contours, from Martin & Kennicutt 1995) and 2200Å continuum (greyscale, from Meurer et al. 1992) image of the central region of NGC 5253. Five 0.86" apertures from our FOS program are marked. Contours are 95%, 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10% of the peak H flux. N is overabundant by a factor of 3 at locations HII-1 and HII-2 compared to UV-1 and other galaxies of similar metallicity. The peak of the nebular emission is 40 pc from the UV-1 star cluster, and we show that UV-1 is unlikely to be the dominant source of ionizing photons. (From Kobulnicky et al. 1997).

Nonetheless, it is interesting to study exceptional cases. The best candidate for localized pollution in an extragalactic H II region is the amorphous galaxy NGC 5253 where a region of enhanced N (Welch 1970; Walsh & Roy 1987; 1989, WR89) and possibly He (Campbell, Terlevich, & Melnick 1986, CTM) coincides with the spectral signature of Wolf-Rayet stars. NGC 5253 contains a very young starburst (Rieke, Lebofsky, & Walker 1988; Beck et al. 1996), making it an ideal laboratory for testing theories of nucleosynthesis and dispersal of heavy elements by massive stars in giant H II regions. In order to investigate the one well-established case of N enrichment in NGC 5253 and its recent star formation history, IMF, and extinction in more detail, Kobulnicky et al. (1997) conducted ultraviolet and optical spectroscopy at five locations in NGC 5253 using the Hubble Space Telescope. In particular, we wanted to measure the carbon abundance in the region of N enhancement, and determine whether C, too, showed signs of localized enrichment that might be indicative of the nucleosynthetic origin of both elements.

Figure 5 shows the locations of the five observations. Slit locations UV-1, UV-2, and UV-3 are centered on three bright, young stellar clusters. Location HII-1 was intended to sample the gas at the location of the suspected N enrichment south of the emission line peak but close to the cluster of WR stars at UV-1, while position HII-2 was intended to sample a "normal" unenriched region nearby.

The abundance measurements at HII-1, HII-2, and UV-1 reveal that the He, C, O, S, and Si abundances are consistent with those in most low-metallicity systems. Surprisingly, N, appears elevated by a factor of 3 (log(N/O) = -0.85) at both locations HII-1 and HII-2 while the N/O at position UV-1 is typical of most metal-poor galaxies (-1.5 < log(N/O) < -1.3). The N/O ratios at locations HII-1 and HII-2 are consistent with one another, and appear even slightly higher than any of those found by Walsh & Roy (1989). At the position of UV-1, despite the presence of WR stars as evidenced by its spectrum, no extraordinary enrichment is evident.

Figure 6. (Upper panel): log(C/O) versus oxygen abundance for 9 galaxies with reliable carbon abundance measurements. The three positions in NGC 5253 from this work are plotted with solid symbols, and are consistent with uniform C abundance throughout. (Lower panel): log(C/N) versus oxygen abundance for 9 galaxies with reliable carbon abundance measurements. The three positions in NGC 5253 from this work are plotted with solid symbols, and demonstrate the remarkable N overabundance relative to C at two locations in NGC 5253. (From Kobulnicky et al. 1997).

In contrast to the elevated N abundances, the He/H abundances appear to be typical of those observed in metal-poor galaxies, although at the upper edge of the distribution (Pagel et al. 1992). C/O ratios, likewise, appear to be typical of similarly metal-poor systems. Figure 6 (upper panel) shows the C/O versus O/H for NGC 5253 along with 10 other points taken from Garnett et al. (1995a) and Dufour (1984) (see Section 4.3.2). N appears to be the only element showing elevated abundances. This results in the deviant positions of HII-1 and HII-2 in Figure 6 (lower panel) which shows C/N versus 12 + log(O/H).

Thus, it appears that at the location of the strongest H emission, the area of the youngest starburst, there is an N enrichment with otherwise normal abundances. This may be the result of pollution from the young cluster (which is not easily visible in the UV image shown in Figure 5). It is possible that the most massive stars have released a few solar masses of N through stellar winds or ejection events. Future high spatial resolution spectroscopy of this region will help us to determine if there is a pattern of pollution.