The Hii Regions of M33. II. A Photometric Catalog of 1272 Newly Identified Emission Regions

The Local Group galaxy M33 (NGC 598) is rich in H ii regions. Early narrowband optical surveys revealed hundreds of them (Aller 1942; Courtès & Cruvellier 1965; Sabbadin, Rafanelli, & Bianchini 1980; Boulesteix et al. 1974; Courtès et al. 1987; and many others). A recent compilation revealed that a total of 1030 H ii regions and 36 supernova remnants were identified prior to 1998 (Hodge, Skelton, & Ashizawa 1999). These objects, especially the brightest ones, have been the subject of a large number of papers, which have used them to study the composition, the abundance gradient, the star formation rate, the modes of star formation, and the velocity field of the galaxy; e.g., see the review published by Sharov (1988). In a previous paper we used this early sample to investigate the luminosity function and size distribution of M33's H ii regions (Wyder, Hodge, & Skelton 1997, hereafter Paper I).

This paper reports the identification of a large number of additional H ii regions (and related emission regions) in M33. Most of the newly recognized objects are fainter and/or smaller than those in previous catalogs. For that reason we use them to explore the faint end of the H ii region luminosity function, which is generally incomplete for most observations of galaxies. As explained in § 4, the luminosity function is interesting because it is the result of the combination of a number of astrophysically important features of a galaxy.

In order to determine a luminosity function correctly, it is necessary to evaluate quantitatively the completeness of a sample. This paper investigates that aspect of the new catalog and reviews the completeness of previous catalogs, using the techniques described in § 5. The corrected total luminosity function is presented in § 6.

In the process of measuring the Hα luminosities of our sample, we have also determined their isophotal sizes. These are used to complete the size distribution curve for the M33 sample, which is discussed in relation to the function found by van den Bergh (1981) in § 7.

In addition to classical H ii regions, the present survey has also identified a large number of more diffuse objects; most of these are probably hydrogen clouds that are excited by the general ultraviolet flux in the disk of M33, which is rich in O and B stars (Humphreys & Sandage 1980). The morphology of the various types of emission regions is discussed in § 8.

The images used to identify objects in this catalog were generously provided to us by Shawn Gordon. They were obtained by W. Blair, C. Smith, R. Kirshner, and K. Long with the Kitt Peak National Observatory (KPNO) 4 m telescope, using a narrowband (34 Å FWHM) Hα filter (centered at 6571 Å to allow for the blueshift of M33) and a wider (145 Å) filter for the continuum, centered at 6105 Å. The original purpose was to discover and study supernova remnants in M33, using a third set of images exposed behind an S ii filter (Gordon et al. 1998). The images were calibrated as described in Paper I.

Figure 1 outlines the coverage of the four sets of images obtained with the KPNO 4 m telescope; the area covers most but not quite all of the main disk of M33. The giant H ii region NGC 588 lies on the west edge of the covered area and several fainter H ii regions lie beyond the boundaries of our search. Thus our catalog is not complete, and a new faint survey of the outer parts of the galaxy should reveal additional H ii regions.

Zoom In Zoom Out Reset image size

Table 1 is a catalog of the newly identified emission regions in M33. Names have been assigned to them according to the precepts of the IAU Commission 5, and these designations are registered with the Commission and the Strasbourg Data Archives. The objects are ordered in increasing right ascension (with a few exceptions where last minute corrections to the catalog were made), and each number is given the prefix "HBW." The positions of the centroids of each object were obtained by reference to the Digitized Sky Survey Catalog and are accurate to a few seconds of arc. Centroids were chosen by eye rather than by some photometric criterion, which would be awkward to define in the cases of complex morphologies.

The Hα fluxes for most of the emission regions were measured by integrating over the area defined by an outer isophote level of 10⁻¹⁶ ergs s⁻¹ cm⁻² arcsec⁻², as measured above the local background. Very faint or diffuse H ii regions had no areas with a surface brightness this bright, and therefore no Hα fluxes are listed for those. The sizes listed in the catalog are based on the area contained within that isophote.

The last column in the catalog gives some descriptive notes. Unresolved emission regions, which are probably mostly compact H ii regions, but which may also include some bright planetary nebulae or emission‐line stars, are identified by the letter "u." Diffuse² regions for which there is no obvious excitation source (and for which boundaries are often somewhat arbitrarily drawn) are indicated by the letter "d." Other notes call attention to unusual morphologies or other notable circumstances.

Charts that identify the emission regions are not provided, as the objects, together with all previously cataloged emission regions, are included in the identification atlas of M33 published elsewhere (Hodge et al. 1999).

In Paper I we explained the motivation behind the determination of the luminosity function of a galaxy's H ii regions. Observationally, the firmest conclusion is that the bright end of the luminosity function is nearly independent of the galaxy's characteristics, with some minor exceptions (Kennicutt, Edgar, & Hodge 1989). The slope and shape of the bright portion of the luminosity function is determined largely by the mass spectrum of star clusters undergoing formation. Except, of course, for the problem of small number statistics near the bright limit, this portion of the function is well established.

The faint end, on the other hand, is greatly uncertain observationally because of brightness limitations and the incompleteness of surveys, as discussed in Paper I. This portion of the luminosity function involves single‐star H ii regions; its shape is a function of the bolometric stellar initial mass function and certain characteristics of the interstellar gas. Because our present list of newly recognized H ii regions essentially doubles the number of faint H ii regions, our data should be useful in more clearly defining the faint end of the luminosity function, making it more amenable to astrophysical interpretation.

Not all of the newly discovered emission regions in Table 1 should be included in a luminosity function of H ii regions. We have omitted all objects that we describe as patches of diffuse emission as they are probably not normal single‐star or single‐association H ii regions. Also, as explained above, many of them have such low surface brightness that their luminosities were not measured.

Figure 2 shows the differential luminosity function for the newly identified H ii regions in Table 1. The data are plotted in logarithmic bins of luminosity. Clearly, although a few fairly luminous H ii regions are present, the bulk of the objects have luminosities in the range where previous luminosity functions were highly incomplete.

Zoom In Zoom Out Reset image size

As we did in our previous paper, we carried out a number of tests to determine the completeness of the present catalog at various flux levels. Incompleteness at the faint end is largely a question of the detection limit, while at brighter luminosities emission regions in crowded areas can be undercounted when a photometric boundary is used as an identification criterion. Two overlapping H ii regions could be counted as one, or small faint H ii regions could be lost if superimposed upon the image of a large, bright region.

Two of us (T. W. and P. H.) independently carried out completeness tests using the field combination method described in Paper I. We used 19 different pairs of fields, superimposed on the computer screen to present simulated, more crowded fields. The areas were chosen from throughout the galaxy, with a preponderance of outer fields in order to have the average effective crowding similar to the original sample. We then used the same criteria used in the initial survey to identify separate emission regions, producing a catalog for each image pair. This was then compared with the lists of H ii regions in previous catalogs, including that in this paper. The number of emission regions recovered was then recorded and all results combined to produce a measure of the completeness of our searches as a function of the brightness of the H ii regions. Areas of diffuse emission were not included for reasons described above.

Figure 3 shows our results. Clearly the new survey reaches to fainter limits than did earlier surveys (the Boulesteix et al. 1974 and Courtès et al. 1987 catalogs), and is finding at least 25% of the H ii regions at the very faint level of 2 x 10³⁴ ergs s⁻¹, our measurement (but not detection) limit.

Zoom In Zoom Out Reset image size

With all data corrected for incompleteness, we have combined the three catalogs (Boulesteix et al. 1974; Courtès et al. 1987, and this one) to produce a complete luminosity function for H ii regions in M33 (limited, of course, to the area of our survey, which excludes a few remote H ii regions). This is plotted in Figure 4, which uses logarithmic bins of 0.2 dex. The distance used for M33 is 840 kpc (Freedman, Wilson, & Madore 1991).

Zoom In Zoom Out Reset image size

The shape of the luminosity function is similar to that derived in Paper I, extending to a limit that is an order of magnitude fainter than previously reached. The peak of the function remains broad and is remarkably symmetrical. The peak is at L = 10^35.8 ergs s⁻¹ and the width (FWHM) is approximately 2.4 dex.

Before this curve can be compared with theoretical models, we need to correct the derived luminosities for the effects of extinction. Unfortunately, measurements of the extinction in all of the H ii regions are not available. However, Viallefond & Goss (1986) have compiled reddening data for 16 of M33's emission regions for which they had high‐resolution radio data and for which published values were available for optical line strengths. Their Table 2 quotes extinction values at Hβ based on both optical data (Hα/Hβ line strengths) and radio/Hβ data. They argue that the radio/optical ratios produce the more reliable extinction values, the optical‐only data being subject to the effects of internal dust.

Concerned that their assembled results might be skewed by the fact that they necessarily were limited to including only bright H ii regions, we used their data to look for a possible correlation with luminosity, but no significant correlation was found. Therefore we propose adopting a mean of their tabulated extinction values as the best correction to be made globally to the luminosity function. The mean extinction derived from the radio/optical ratios is 1.22 ± 0.50 mag at Hβ (this can be compared with the mean Hα/Hβ, taken from both core and global values, of 0.80 ± 0.38). Adopting this value as representative for all of M33's H ii regions means that the curve plotted in Figure 4 should be moved to higher luminosities by 0.16 dex before it can be compared with models (unless, of course, the models include the effects of dust, both internal and external). Note also, however, that we have not corrected the fluxes for contamination by [N ii] emission. Such correction will decrease our values by 10%–20%.

Because this paper is primarily a data paper, we do not here pursue the interesting questions that can be addressed by comparing Figure 4 with various models of star formation regions. We will do so, however, in a future paper in this series. There we will also compare our results with those for other galaxies where the data reached comparably faint limits.

For all H ii for which we measured fluxes, the areas within our specified isophote limit was also determined. These figures have been converted to effective diameters, defined as the diameter of a circle of the same area. These data have been used to plot a size distribution (Fig. 5). For uniformity, we have excluded the emission regions in the southeast quadrant, which had a shorter effective exposure because of thin clouds (see Paper I). Also, all unresolved and diffuse regions were excluded. Figure 5 is shown in order to provide the reader with an idea of the size characteristics of this sample of newly cataloged H ii regions.

Zoom In Zoom Out Reset image size

To effect a comparison with the size distributions of H ii regions in other galaxies, we have combined our new data with the sizes determined for all previously cataloged H ii regions in M33. Since van den Bergh's (1981) first presentation of this sort of data it has been traditional to plot the size distribution cumulatively, as we have done in Figure 6. For comparison we also plot the curve derived in our previous paper. As found by van den Bergh (1981) and by many subsequent investigators, the size distribution is usually exponential, of the form N = N₀e^{- D/D_o} where D is the diameter, D_o is a characteristic scale peculiar to the galaxy, N₀ is a constant related to the total H ii region population of a galaxy, and N is the number of H ii regions with diameters greater than D. Our more complete sample does not change the shape of the M33 curve very much, except to straighten it out somewhat at both the large and the small ends of the distribution. A least squares solution for a linear fit yields a value of D_o = 34 ± 1 pc, in good agreement with our previously derived value.

Zoom In Zoom Out Reset image size

Most of the newly cataloged emission regions listed in Table 1 are normal H ii regions. However, there are also several objects whose morphologies indicate that they may not be examples of traditional H ii regions. A total of 227 objects, 18% of the sample, are classified in Table 1 as diffuse emission regions. Most of these are large, with very low surface brightnesses and do not appear to be clearly associated with massive stars or star groups. We believe them to be interstellar gas that is excited by the general ultraviolet flux in the galaxy, derived from the high density of O and B stars that exist throughout the disk. Their surface brightnesses are generally below 10⁻¹⁶ ergs s⁻¹ cm⁻² arcsec⁻². The adopted borders of these diffuse regions were chosen by eye and do not necessarily have quantitative significance. The average dimension of the cataloged examples, which we quote with the warning that it may have little real significance, is 2 9 ± 05 (∼12 pc).

An additional 168, or 13%, of the newly cataloged objects are unresolved, with diameters less than ∼2^'' (∼8 pc). Many of these are fairly luminous; the brightest have luminosities on the order of 10³⁶ ergs s⁻¹. There are probably three different types of objects among the cataloged unresolved regions. Some are probably compact H ii regions, some are probably planetary nebulae, and some may be emission‐line stars whose Hα emission is bright enough to swamp the continuum images. Spectra of each of these sources would be needed to settle the question of their nature.

We note the presence of three other kinds of unusual morphologies. These are various kinds of linear features, especially among the diffuse regions and some of the luminous complexes. These are identified in Table 1 as either "linear," indicating a straight‐line morphology (of which there are 36) or "arcs" or "filaments," indicating a narrow, curved morphology (of which there are 48). The third category includes the "bubbles" and "rings." These kinds of objects are also found among the brighter, previously cataloged H ii regions and have been the subject of extensive study (e.g., Drissen et al. 1991 and Hunter 1994b). Our catalog includes 18 of these objects, most of which are probably stellar windblown bubbles like N70 in the Large Magellanic Cloud (Skelton et al. 1999) but intrinsically fainter. A spectroscopic study like that published by Hunter (1994a) and Oey & Massey (1994) for Magellanic Cloud bubbles would be of interest.