Giant H II regions similar to those in spiral galaxy disks are also found in dwarf irregular galaxies. These regions have acquired a variety of aliases, all of which refer to essentially the same type of object, dwarf irregular galaxies with spectra dominated by narrow emission lines. Observers making spectroscopic observations of these objects have called them "isolated extragalactic H II regions" (Sargent and Searle 1970, the discovery paper); "H II region-like galaxies" (French 1980); or "H II Galaxies" (Campbell, Terlevich, and Melnick 1986). Other names refer explicitly to the host galaxies, which are generally irregulars: e.g. "blue compact dwarf galaxies (BCDG's or BCG's)", "star-forming dwarf irregular galaxies". Searle and Sargent (1972) called particular attention to the prototype objects, I Zw 18 and II Zw 40, because of their extraordinarily low elemental abundances of oxygen and neon. From their composition, these authors inferred that the current star formation rate in these galaxies is much greater than the past average rate. They concluded that star formation occurred intermittently rather than continuously in these galaxies, and coined the term "bursts" of star formation to describe this phenomenon. Dwarf irregular galaxies experiencing active star formation, and "starbursts" in general, have received a great deal of attention lately. The proceedings of two conferences cover much of the recent work on the subject (Kunth, Thuan, and Van 1985; Thuan, Montmerle, and Van 1987).
4.2. The "I Zw 18 Problem" and the Mass-Metallicity Relation
The very low elemental abundances in these dwarf irregular galaxies imply that these are "young" or "new" galaxies, at least in terms of their net star formation activity and nucleosynthetic evolution. They therefore offer an opportunity to watch the early stages of chemical enrichment, which happened in the distant past in our own galaxy. Furthermore, in these galaxies there also has presumably been relatively little helium produced by stars, so they are attractive objects to study for the determination of the pre-galactic, cosmological component of helium. Both of these factors helped motivate a lively search for the most metal-poor H II regions which could be found.
Surveys such as those of Kinman and Davidson (1981), Kunth and Sargent (1983), and Kunth and Joubert (1985), yielded around 100 galaxies for which abundances could be measured. However, despite extensive searches, primarily using objective-prism techniques (e.g. Kunth, Sargent, and Kowal 1981), for nearly two decades no other galaxies were found with O/H abundances as low as I Zw 18, one of the two original prototypes. In fact, virtually no galaxies were found with O/H < 1/10 solar, and the median value was closer to 1/6 solar, log(O/H) + 12 = 8.0. Kunth and Sargent (1986) considered whether this result could be due to selection effects; this median value happens to fall near the abundance at which the strong optical [O III] lines reach their peak intensity, as discussed in Section 2.3. (Campbell, Terlevich, and Melnick 1986). It has been noted that I Zw 18 itself was not found by objective prism work; it was first identified as a blue compact galaxy. Nevertheless, Kunth and Sargent (1986) concluded that the failure to find other "I Zw 18's" was not a result of selection bias, but rather indicated that extremely metal-poor H II regions are intrinsically rare. In order to reconcile this conclusion with the existence of gas-poor dwarf spheroidal galaxies having metallicities far lower than those of I Zw 18, they suggested a scenario of rapid self-enrichment of the gas in giant H II regions by supernovae from the ionizing cluster. If the supernova products remained concentrated within the H II region rather than becoming diluted throughout the entire system, then the gas metallicity would rise to that of I Zw 18, while most of the stars remained metal-poor.
The case of I Zw 18 still remained problematical, however. For an instantaneous starburst, the heavy elements are not released until the massive stars begin to die, several million years later, and by then the ionizing radiation field required to produce a bright H II region will have declined drastically, because these are the very stars that provide most of the UV photons (e.g. Lequeux et al. 1979; also see Figure 8 in Kunth 1986). If I Zw 18 is still experiencing its first starburst, the stellar population cannot be coeval; that is, the star formation epoch must have a duration longer than several-million-year timescale (Kunth and Sargent 1986). If C and N were enriched relative to O in I Zw 18, compared to their ratios in other dwarf galaxies, then an even more complex star formation history would be required (Dufour, Garnett, and Shields 1988; Pantelaki 1988).
Recall, however, the mean metallicity-galaxy mass relation discussed above (Figure 2). A similar relation is seen for the dwarf irregulars, along with a correlation between the O/H abundance and the gas mass fraction, as expected for the simple model of galactic chemical evolution (e.g. Lequeux et al. 1979; Talent 1980; Kinman and Davidson 1981; Vigroux, Stasinska, and Comte 1987). This suggests an alternate strategy for searching for extremely metal-poor H II regions: namely, choose galaxies known to be of low mass, rather than selecting objects because of their emission lines. This is the approach taken recently by Skillman and his collaborators, who examined a number of dwarf galaxies in the Local Group and found several H II regions with O/H values as low as that of I Zw 18 (Skillman et al. 1988b; Skillman, Kennicutt, and Hodge 1989; Skillman, Terlevich, and Melnick 1989). The success of this strategy is apparent from the histogram of O/H values for their sample as compared to earlier samples (Figure 5). There still appears to be a threshold value near the O/H abundance of I Zw 18, suggesting that the self-enrichment mechanism advocated by Kunth and Sargent (1986) may apply, but this more recent work does fill in the metallicity "gap" between I Zw 18 and the other previously studied dwarf irregular galaxies. These low-metallicity H II regions extend the mass-metallicity relation seen for higher-mass galaxies (Figure 2) down to a regime that overlaps with metal-poor dwarf spheroidal systems that lack a substantial Population II component. There seems to be reasonable agreement between the two sets of objects, allowing for uncertainties introduced by the fact that the stellar "metallicities" refer to the iron-group, while the H II region abundances are determined for oxygen (e.g. Aaronson 1985; Pagel 1987).
|
Figure 5. Comparison of the histograms of O/H abundance for different samples of dwarf irregular galaxies, reprinted from Skillman, Terlevich, and Melnick 1989. |
4.3. Dwarf Galaxies as Tests of the Simple Model
The simple model of galactic chemical evolution, discussed in Section 3.5, was originally designed for dwarf irregular galaxies like I Zw 18 (Searle and Sargent 1972). Thus it is natural to compare their properties with the predictions of the model, such as the expected linear relation between metal abundance, Z, and ln(Mtot / Mgas). In order to make such a comparison, it is necessary to measure both Mtot and Mgas, which is not easy to do for these faint, low-mass systems. Observations are made of the 21 cm H I emission; its intensity yields the mass of interstellar gas; the total gravitational mass can be inferred from the line-widths. Studies of this type have found that the dwarf galaxies tend to fall near the expected relation but with a fairly large scatter (Lequeux et al. 1979; also see Figure 3.3 of Pagel 1987). The fact that I Zw 18 in particular falls very far from this relation, with far too low a gas fraction for its low metallicity, has frequently been noted (Skillman et al. 1988b). The total masses of these galaxies are somewhat uncertain, however, due to the possible unobserved presence of molecular gas, and issues regarding the dynamics of these systems (e.g. Gallagher and Hunter 1984; Skillman et al. 1988a; Hoffman et al. 1989). Furthermore, these small galaxies might not be closed systems, but might experience infall of primordial material or the loss of metal-rich supernovae ejecta which more easily escapes their weak gravitational fields (Larson 1974; Larson and Dinerstein 1975).
4.4. Dwarf Galaxies and the Extragalactic H II Region Sequence
The H II regions in dwarf irregular galaxies can be studied using the
same methods as used
for regions in the disks of spiral galaxies. In general, they seem to
follow the same
patterns as the disk regions. When directly measurable, their electron
temperatures
decrease as O/H increases, following the same relationship (e.g. Figure
1 of
Terlevich 1986).
As expected, the "bright-line" method of abundance analysis also applies
to the
dwarf irregular H II regions. However, the regions with O/H abundances
less than about 1
× 10-4 fall on the lower branch of the O/H vs. line
intensity
diagram described in Section 2.3., where
R23 = {[O II] 3727Å + [O III] 4959+5007Å} /
H
is directly
proportional, as opposed to inversely proportional, to O/H
(Figure 6, reprinted from
Skillman 1989).
|
Figure 6. The bright-line ratio
vs. abundance relation is shown for low-mass galaxies in the
Local Group. The ordinate R23 is the line ratio {[O II]
3727Å + [O III] 4959 + 5007Å) /
H |
Another property that H II regions in dwarf galaxies share with those in
spirals is the
correlation between the degree of ionization (measured from the
O++ / O+ ratio) and the
metallicity. Recall that, in spiral galaxies, the degree of ionization
increases, and O/H
drops, from the center out to the edge of the disk. For dwarf galaxies,
the radial location
drops out of the picture, but the ionization fraction still increases as
the abundance
decreases. The interpretation of this effect is not universally agreed
upon. Many authors
interpret this effect as being due to hotter ionizing stars, and infer
T*, or "Tion" using
nebular models. In this case, one must account for such a variation in
the stellar population. Age effects alone are insufficient (e.g.
Searle 1971),
although they may
account for the observed spread of values, which appear to fall below an
upper envelope (e.g.
Viallefond 1986).
Many authors favor metallicity-dependent variations in the stellar
initial mass function, either by truncation at the top end,
Mupper
(Shields and Tinsley
1976),
or a change in slope (e.g.
Terlevich 1986).
Another school of thought suggests that the
change in ionization fraction is due to a change in U, the ionization
parameter, essentially a
metallicity-dependent variation in the nebular geometry
(Mathis 1985;
Dopita and Evans
1986).
Such a circumstance could arise, for example, if metal-rich stellar
associations
produce stronger winds which more efficiently sweep out the ionized gas,
yielding a nebula with a lower filling factor.
Vilchez and Pagel
(1988)
attempted to circumvent this
ambiguity by constructing a line intensity ratio involving both
O++ / O+ and S++ / S+; they
find that the composite bright-line ratio
= (O+ /
O++) / (S+ / S++) depends only on the
ionizing-star
temperature and is independent of U. These authors apply their formula
to a sample of
dwarf irregular H II regions, and find that the increase in inferred
stellar temperature with
decreasing abundance persists. A third line of argument, less widely
discussed than the
others, is that the change in ionization structure indeed reflects a
softening of the radiation
field, not as a result of a change in initial mass function with
metallicity, but instead
because of changes in the evolution of the cluster stars at
different metallicities. I will
return to this issue in Section 6, below.
4.5. Elements other than Oxygen
4.5.1. Nitrogen. Most spectroscopic studies of dwarf-galaxy H II
regions yield abundance
values for nitrogen as well as oxygen. These objects, which have O/H
values substantially
less than solar, do not follow the approximate N/O
O/H relationship seen for
more metal-rich
regions (see Section 3.3.1. and
Figure 4). Rather, the N/O values for
dwarf irregular
galaxies scatter around a mean value of log(N/O) + 12 = -1.5 (e.g. see
reviews by
Pagel 1985;
Dufour 1986).
Some of the scatter may arise from ionization structure effects: these
regions are highly ionized, so that N+ and O+
contain only a small fraction of their
respectively elements. Another source of observational error is the
difficulty of obtaining
accurate fluxes for [N II] 6548, 6584Å, which are extremely weak in
these objects and are situated on the wings of the extremely strong
H
line (e.g.
Dufour, Garnett, and
Shields 1988).
Nevertheless, it seems clear that N does not behave like a purely secondary
nucleosynthetic product; it must also have a large primary component,
which dominates in the most metal-poor regions (e.g.
Matteucci and Tosi
1985;
Vigroux, Stasinska,
and Comte 1987;
Garnett 1989c).
Results from a recent study of nitrogen abundances in dwarf
galaxies is shown as Figure 7 (from
Garnett 1989b),
which includes, for comparison, the
abundances for several regions in the "secondary" domain.
|
Figure 7. The abundance ratios N/O vs. O/H are shown for a large sample of dwarf irregular galaxies, and also for several relatively metal-rich H II regions (labeled points at upper right). At log (O/H) + 12 < 8.3, (N/O) shows no systematic variation, but scatters around a mean value of about log (N/O) = -1.5. From Garnett (1989b). |
4.5.2. Sulfur. Sulfur abundances determined for H II regions in large disk galaxies were discussed in Section 3.3.2. above. Until recently, relatively few reliable sulfur abundances were available for H II regions in dwarf galaxies other than the Magellanic Clouds (for these, see Dennefeld and Stasinska 1983; Dufour 1986). These regions generally have such high excitation that the [S II] 6717, 6732Å lines are weak and insufficient for measuring accurate total abundances, since a very small fraction of the nebular sulfur is in the S+ ion. The [S III] 6312Å line is also generally weak, intrinsically temperature-sensitive, and often blended with nearby lines of [O I]. In principle, the best method for determining sulfur abundances is to observe the stronger [S III] 9069, 9532Å lines. Recently, Garnett (1989a, b) has observed these lines in about a dozen dwarf galaxy H II regions, and determined reliable S/O abundances using nebular models to obtain ionization correction factors. The results show no convincing evidence for a variation of S/O with respect to O/H. The mean value is log(S/O) + 12 = -1.6, with a scatter of ±0.3 dex, probably at least partly due to observational uncertainties.
4.5.3. Other elements: The abundance of neon is relatively easily obtained for metal-poor, high excitation dwarf galaxies, through the strong [Ne III] blue lines. Neon correlates very closely with oxygen, which is expected since it is believed that neon and oxygen are synthesized in the same population of massive stars (Kunth and Sargent 1983; Vigroux, Stasinska, and Comte 1987). Carbon abundances are harder to obtain, since they require measurement of lines in the satellite ultraviolet. At present, with only the relatively small telescope of the International Ultraviolet Explorer (IUE) satellite available to observe these lines, results are available mainly for very nearby galaxies such as the Magellanic Clouds. (This situation will improve dramatically once the much more sensitive Hubble Space Telescope begins collecting data on extragalactic H II regions.) So far, it appears that N/C is constant for any value of C/H, which is interpreted as indicating a mainly primary origin for both N and C (Peimbert 1986; Dufour 1986).