6.3. Example 2: IZw18

The dwarf emission-line galaxy IZw18 plays an important role in studies of the properties and evolution of dwarf irregular galaxies because of its extreme properties. Searle & Sargent (1972) determined early on that IZw18 had a very low oxygen abundance; since then the O/H value has been refined to the current best estimate of 1.47 ± 0.15 x 10^-5 in the NW HII region (Skillman & Kennicutt 1993, SK93), the lowest value measured in any emission-line galaxy. The colors of IZw18 are very blue, dominated by young massive stars, and, to date, no evidence has been found for an older red stellar population (see Hunter & Thronson 1995). The oxygen deficiency and blue colors lead to the conclusion that IZw18 and similar galaxies are either very young, experiencing their first episodes of star formation, or have experienced only sporadic episodes of star formation over their history (Searle, Sargent, & Bagnuolo 1973).

If the ionized gas in IZw18 is pristine material which has recently been contaminated by Type II supernova ejecta, the heavy element abundance ratios should show values characteristic of massive star nucleosynthesis. In particular, the gas should be relatively deficient in elements produced mainly in low- and intermediate-mass stars, such as carbon and nitrogen, compared to oxygen, which is produced in massive stars alone. Indeed, Kunth et al. (1994) inferred from high-resolution GHRS spectra an oxygen abundance in the neutral gas that was much smaller than in the HII region in IZw18, and concluded that the ionized gas has been enriched by the present starburst (but see Pettini & Lipman 1995 for a criticism of their analysis).

Figure 32. Log C/O vs Log O/H from FOS spectroscopy of IZw18 (filled squares), compared with data from Garnett et al (1995a; unfilled circles) and Kobulnicky et al. (1997; unfilled triangles). Solar value is from Grevesse & Noels (1993). (From Garnett et al. 1997a).

With this background in mind, Garnett et al. (1997a) obtained new HST observations of IZw18. Observations were obtained in two separate components of IZw18. Figure 32 shows our new results for C/O in IZw18 compared with the data from Garnett et al. (1995a; G95a, discussed in Section 4.3.1) plus new observations of NGC 5253 by Kobulnicky et al. (1997, discussed in Section 4.1). Small number statistics still inhibit any interpretation of the measured dispersion in C/O, so we shall rely on comparison with the most metal-poor galaxies to interpret the IZw18 measurements. The three most metal-poor galaxies from the G95a sample have a mean log C/O = -0.85 ± 0.07 (mean error); by comparison, the mean log C/O = -0.60 ± 0.09 in IZw18 is 2.9 above this average (and even higher than expected from extrapolating the trend of C/O vs O/H observed by G95a). Similarly, the mean log C/N = +0.98 in IZw18 is also well above the values determined for the other very metal-poor dwarf galaxies (G95a).

The elevated C/O abundance ratio in IZw18 suggests that the galaxy has experienced enrichment in carbon from an older generation of stars, and that the current burst of star formation is not the first. G95a noted that nucleosynthesis models for 10-42 Mstars by Weaver & Woosley (1993) predict an integrated log C/O = -0.83 for their "best estimate" of the ¹²C(, )¹⁶O nuclear reaction rate, corresponding quite well with the values measured by G95a in their most oxygen-poor galaxies. If this correctly represents the nucleosynthesis contribution from massive stars alone, then the additional carbon observed in two widely separated locations in IZw18 must come from lower mass, long-lived carbon star and planetary nebula progenitors with lifetimes >> 10 Myr.

This is, essentially, the "time delay" described in Section 3.3. Heavy elements that are produced in massive stars, such as oxygen, are injected into the ISM by a starburst early and can enrich the gas on relatively short timescales; supernova-driven galactic winds may also eject much of the oxygen-rich SN ejecta into the galaxy halo and reduce the effective oxygen yield. After the massive stars have died away, lower-mass stars can enrich the ISM in carbon (and nitrogen) in more gentle mass loss events, leading to higher C/O and N/O than one would expect in a young galaxy which has been enriched by massive stars alone. The relative abundances of C, N, and O then provide an indication of the time elapsed since the last major starburst in a dwarf galaxy (Edmunds & Pagel 1978).

Figure 33. Log C/O vs. log age from chemical evolution models for IZw18 (KMM) and blue compact dwarf galaxies (CCPS). The KMM models are labeled according to model number; KMM6a shows abundances from the KMM two-burst model 6 after 990 Myrs, at the beginning of the second burst. For the CCPS models, refers to the strength of the SN-driven galactic wind. The hatched area shows the range occupied by the observed values of C/O in IZw18 and corresponding errors. (From Garnett et al. 1997a).

In Figure 33 we compare the observed C/O ratio in IZw18 with the predictions of chemical evolution models for blue compact dwarf galaxies as a function of galaxy age. The filled squares show models computed by Kunth, Matteucci, & Marconi (1995; KMM) specifically to explain the abundances in IZw18. These are all one-burst models of short age, except for their model 6, which is a model consisting of two bursts, one that occurred 1 Gyr ago and one beginning only 10 Myr ago. We show their predicted abundances for IZw18 at the beginning of the second burst, after 990 Myr. We also show the results of general models for metal-poor blue compact dwarf galaxies by Carigi et al. (1995; CCPS). These models were computed assuming continuous star formation rather than starbursts, and different massive star yields. Both sets of models include the effects of differential (heavy element enriched) winds. The hatched region shows the values of C/O encompassed by the observations of IZw18 and corresponding error bars.

None of the KMM models are able to account for the high C/O in IZw18, suggesting that an additional source of carbon is needed in their models. The CCPS models can explain our observed C/O ratios for galaxy ages of the order 1 Gyr. From these results it could be inferred that IZw18 had an episode of star formation that occurred at least a few hundred million years ago that led to the presently observed levels of carbon and nitrogen. Interestingly, this corresponds roughly to the age inferred from stellar photometry in the companion irregular galaxy NW of the main body of IZw18 (Dufour et al. 1996), which may suggest that an older population may still lay obscured by the light of the luminous present-day burst in the main body.

The agreement in N/O and O/H between the NW and SE components of IZw18 led SK93 to question the validity of the "self-pollution" model for IZw18. The agreement between our two high values for C/O further strengthen this argument. Thus, it seems highly improbable that these two separate star formation events could lead to essentially identical abundances in a self-pollution model. At this point, the simplest explanation appears to be that IZw18 is not a young galaxy, but rather has had one or more previous episodes of star formation. Future deep HST photometry with the HST will be needed to resolve this unambiguously. In the end, the best way to determine the presence of an older generation of stars will be to detect it directly. However, if we are correct, and the ISM abundance is indicative of an, as yet, unseen population, we will be adding an important diagnostic to our tool kit for galaxy analysis.