Highlights

The sample consists of around 30 disk galaxies, all of which have deep H data, with a small sub-sample having full data - H, broad-band and HII region spectroscopy. The study forms Annette Ferguson's Ph.D. thesis (1997) and we shall highlight some results.

HII Regions and Star Formation

 
Figure 1: An H continuum-subtracted image of NGC 6946. The dashed line indicates R.

It is clear from the deep H images that many galaxies - but not all! - have HII regions, and thus very recent massive star formation, well beyond the `optical radius' defined by the isophote. An immediate qualitative impression from the images (see Figure 1, NGC 6946) is that the star formation in the outer disk is exclusively in small, unclustered HII regions of typical luminosity 1-10 times Orion, while star formation in the inner disk () occurs in HII regions that themselves cover a wide range of luminosities, and cluster into large complexes. Some of this effect is no doubt due to the simple fact that there are fewer HII regions in the outer disk, but the impression is compelling that the star formation in the outer disk is different from that in the inner disk. Why?

Fall & Efstathiou (1980) investigated local gravitational instability in finite-thickness isothermal gas disks, deriving the criterion for Jeans' instability, giving fragmentation on scales greater than the disk thickness, , with the surface density and the sound speed in the gas. They expressed this criterion in graphical form and derived a maximum radial extent to fragmentation (their Figure 7). They predicted that gravitational fragmentation in typical disks should be confined to the inner scale-lengths, approximately equal to the optical radius, thus providing an explanation for the edges of active star formation in disk galaxies.

This result of Fall & Efstathiou is rather under-appreciated, perhaps due to its presentation in terms of their model parameters, exacerbated by the sparse observational data at that time with which to compare their predictions. Re-arranging their inequality (38) into more familiar form gives a critical gas density for gravitational instability on these scales:

with the epicyclic frequency. The Toomre `Q' criterion for infinitely-thin shearing gas disks yields essentially the same result for fragmentation:

In this case there is a maximum and minimum unstable length scale, and a fastest rate of growth at scales with wavelength . This fastest-growing wavelength varies with radius, but is typically around 1 kpc. It is probably no coincidence that this leads to bound systems corresponding to about the scale of the largest systems in star-forming disks (Efremov 1995) and indeed of the large HII region complexes in the inner disks studied here. Of course it is not trivial to go from gravitational instability on this large scale, to star formation.

Kennicutt (1989) invoked the Toomre-Q criterion to explain his observed sharp drop-offs in the radial distribution of HII regions as a function of gas density. Echoing the results of Fall & Efstathiou, Kennicutt found that the star-formation/HII region edge was typically close to the optical radius of the disk. We do see star formation beyond the optical radius, but only in small HII regions, and there is an apparent qualitative difference between the HII regions in the inner and in the outer disk. Does this change reflect crossing the Q-threshold? This then raises the question of what is driving the outer disk star formation?

 
Figure 2: (Left) Variation of vs for NGC 7793. Different symbols illustrate the effect that diffuse ionized gas has on the derived relationship. The `hook' at high HI gas densities is most probably due to the neglect of molecular gas, for which data are unfortunately lacking. (Right) Variation of vs for NGC 6946. Note the abrupt steepening of the correlation below  11 M/pc.

However, these interpretations of our data are complicated by the fact that the values of the parameters - all radially-varying - that determine the `threshold' gas density are rather uncertain, and factors of two are important. Recent determinations of quantities such as the real radial variation of the HI gas velocity dispersion give, in some cases, edges to the gravitationally-unstable disk that are well outside even our observations. The rather dramatic change in appearance of the star-forming regions seen in our images would then be unexplained, even given the fact that varies with radius. Further, disks are not the simple one-component systems assumed in this instability analysis, and `Q' for multi-component disks is complicated (e.g. Elmegreen 1995). It is also probably relevant that for most of the galaxies the outer-disk HII regions are in spiral arms.

Various star-formation laws may be tested with the data, and the intention is to derive the best-fitting description of all the data. A preliminary analysis shows that no one model - such as a straightforward Schmidt Law - fits all the galaxies. The variation of massive-star formation in NGC 6946 as inferred from the H observations is shown in Figure 2 (right panel). These data show a steepening of the `Schmidt Law' index from inner to outer disk, as noted previously by Kennicutt (1989), but our data show this continuing with no abrupt end.

Our H data often allow study of diffuse gas, through surface photometry, in addition to identification and counting of HII regions. As discussed by Walterbos and by Rand (this volume) and by ourselves (Ferguson et al. 1996a,b), the line ratios in the diffuse ionized gas, its close association with HII regions and its energy requirements are all best explained if the vast bulk of the diffuse ionized gas (DIG) is photoionized by massive stars. Inclusion of diffuse H emission when estimating the massive-star formation rate can change the derived radial profile of star formation both qualitatively and quantitatively (Ferguson et al. 1996a). Figure 2 (left panel) shows derived azimuthally-averaged star-formation rates in one of our Sd galaxies, NGC 7793, both with and without the diffuse gas, as a function of HI column density.

Chemical Abundances in Outer Disks

 
Figure 3: Chemical abundance gradients (O/H, N/O) derived for NGC 628 and NGC 6946. The dashed lines indicate the solar values.

Spectroscopy of outer HII regions allows chemical abundances to be derived from the emission-line strengths. The chemical abundances are important discriminants of different theories of the temporal variation of star formation rate and gas flows, complementing the H- data that constrain the very recent star formation rate. Again, data for the outer disk is particularly important in discriminating among theories (e.g. Prantzos & Aubert 1995). Further, the chemical abundances of the gas-rich outer disks may be relevant for the interpretation of the metallicities seen in quasar absorption-line systems.

We have obtained spectroscopy of several of our newly-detected HII regions in several galaxies, using the KPNO 4m telescope. The galactocentric radial distances of our target HII regions extend to twice R. Abundances have been derived from strong lines, using the technique of McGaugh (1994) for oxygen, and that of Thurston et al. (1996) for nitrogen. A few of our HII regions in the inner disk have previous elemental abundances derived by McCall et al. (1985), and our observations and results agree well with theirs. The resultant chemical abundances are shown in Figure 3 for 2 galaxies; for these the outer disk is consistent with simple extrapolation of the gradient from the inner disk. Our estimates of the outer [O/H] are typically dex, with [N/O] from solar to one-third of solar. The nitrogen data are consistent with a secondary origin for nitrogen. These values are reminiscent of dwarf irregular galaxies (e.g. Garnett 1990), of damped Lyman clouds (e.g. Lu et al. 1996), and of low surface brightness disks (e.g. McGaugh 1994). We are developing chemical evolution models consistent with the star formation rates we infer from the H and broad-band observations, and augmenting our sample of HII regions with chemical abundances.

© Copyright Astronomical Society of Australia 1997