Proceedings of the Workshop
"The Magellanic Clouds and Other Dwarf Galaxies"
of the Bonn/Bochum-Graduiertenkolleg

CO Emission in Low Luminosity

Star Forming Galaxies

Christopher L. Taylor1, Henry A. Kobulnicky2, and Evan D. Skillman3

1Astronomisches Institut der Ruhr-Universität Bochum
2UCO/Lick Observatory, U. California, Santa Cruz,
3Astronomy Department, University of Minnesota

Received 27th March 1998
Abstract. We report new 12CO 1→0 observations of 5 nearby dwarf irregular galaxies with low oxygen abundances. None of the galaxies are detected. Two of these galaxies have not been previously observed in CO, and for the other three we improve upon the existing upper limits. In the case of Leo A, we do not confirm a previous detection. This non-detection is consistent with other non-detections of galaxies with similar metallicities. Combining our new data with data from the literature, we find strong evidence that the CO emission in dwarf galaxies depends upon the metallicity, and evidence that it may depend upon another property as well. Because it is almost never detected in dwarfs with oxygen abundance 12 + log (O/H) ≤ 8.0, CO is not a good tracer of the molecular content of low metallicty dwarf galaxies.

1. Introduction

Often, when we make observations of CO we are actually not interested in the CO at all, but rather the H2, which is the dominant species in the molecular ISM. Aside from the inherent interest in studying the distribution and kinematics of the molecular gas, star formation occurs in the molecular gas. Progress in understanding star formation and its role in galaxy evolution requires us first to understand the H2. Because H2 has no strong emission lines from which its column density may be easily determined, we use CO as a tracer of the molecular medium.

CO has been more difficult to detect in dwarfs than in spirals, even from the first efforts at observing dwarfs (e.g. Elmegreen et al. 1980). Under the assumption that the CO/H2 conversion ratio derived for the Milky Way is universally applicable, this implied that dwarf galaxies have very little molecular gas (Young et al. 1984; Tacconi & Young 1987). However, the observation that dwarf irregulars (dIrrs) often host giant H II regions suggests that they must contain molecular gas in which the OB stars could form - thus implying their molecular gas is deficient in CO.

Recently a consensus has developed that the Milky Way CO/H2 conversion ratio is not necessarily applicable to all galaxies. Several studies, (e.g. Wilson 1995; Verter & Hodge 1995) have provided strong evidence that the conversion ratio depends upon the galaxy metallicity, so that it increases with decreasing metallicity. Specifically Wilson (1995) found the following relationship between the conversion factor (α) and oxygen abundance (O/H): log (α/αMW) = (5.95±0.86) - (0.67±0.10) [12 + log (O/H)] for galaxies whose oxygen abundances range between 7.98 ≥ 12 + log (O/H) ≥ 9.02. Unfortunately, many dwarf galaxies have oxygen abundances below 7.98 and nearly all remain undetected, so not much is known about the CO/H2 conversion ratio at low metallicities. In an effort to improve this situation, we have obtained 12CO 1→0 observations of 5 nearby, low metallicity dIrrs.

2. Observing Strategy

We carried out our observations on 5 - 11 January 1998 with the NRAO 12 m telescope at Kitt Peak. The 3 mm SIS receiver was used with the filterbank spectrometer and a 1 MHz filter, yielding 256 channels per spectrum, and a channel width of 2.6 km s-1. The receiver was tuned to the central velocity of the H I distribution in each galaxy. Operating at 115 GHz, the half power beam width was 55". System temperatures varied from between ∼300 to 500 K, infrequently rising higher at times. When the galaxy being observed was small enough, observations were carried out in beam switching mode (BSP), with a throw of 2'. If not, absolute position switching (ABS) was employed to ensure a reference position uncontaminated by emission from the galaxy. Each scan had a linear baseline subtracted. Scans for which a linear baseline was adequate were averaged, with each scan weighted by a factor of 1 / Tsys2. Scans which could not be fit with a linear baseline were not used. Final, averaged spectra for three positions in Leo A are presented in Fig. 1. Table 1 gives observed and derived data for the galaxies.
Table 1.
Galaxy Distance 12 + log (O/H) TR*(rms) I(CO) L(CO)
  (Mpc)   (mK) (K km s-1) (104 K km s-1 pc2)
LeoA 1 (BSP) 2.2 7.3 2.7 <0.07 <1.9
LeoA 1 (ABS) 2.2 7.3 3.9 <0.10 <2.7
LeoA 2 (ABS) 2.2 7.3 5.5 <0.14 <3.9
LeoA 3 (ABS) 2.2 7.3 2.9 <0.08 <2.0
SexA 1 (BSP) 1.4 7.5 3.4 <0.09 <1.0
SexA 2/3 (ABS) 1.4 7.5 5.5 <0.14 <1.7
DDO 210 (BSP) 4 7.4 4.8 <0.12 <11
DDO 187 (BSP) 4.4 7.4 2.7 <0.07 <7.6
Pegasus (BSP) 1.2 7.9 5.1 <0.13 <1.1

At least three of the galaxies we observed (Leo A, DDO 210 and Pegasus) have been observed before. In fact Leo A was the only dwarf with 12 + log (O/H)≤7.8 to be detected in CO (Tacconi & Young 1987), despite efforts on several other similar systems. This prompted us to want to confirm this important detection. In addition, recent high resolution H I observations by Young & Lo (1996, 1997) have found several dIrrs to have two components to their atomic ISM, the well known and widespread warm gas with velocity dispersion σ = 9 km s-1, and a new, cold component with σ = 3.5 km s-1, concentrated at the regions of highest H I column densities. We reasoned that this cold component of the H I was the most likely area to find CO emission in these galaxies, if it was detectable. Many previous observations did not have H I maps available to direct the search for CO, and so pointed at the center of the galaxies. However, dIrrs often have H I holes at their centers, making the center an unlikely location for dense molecular gas. For our observations we selected galaxies for which H I interferometric maps were available, and targeted our CO observations at the locations of highest column density in these galaxies.

[Click here to see Fig. 1!]

3. Results

The first result of this work is that we did not recover the detection of Leo A by Tacconi & Young (1987), though our sensitivity was more than adequate. We observed 3 positions in Leo A, two corresponding to the locations with cold, dense atomic gas (Young & Lo 1996), and the third to the position listed by Tacconi & Young, which coincides with the central H I hole. If our result is correct, and Leo A has no detectable CO emission down to our detection thresholds, then there are now no low metallicity dIrrs which have been detected in CO. This has important implications for the dependence of the CO/H2 conversion ratio on metallicity. If we assume that the relation of Wilson (1995) is applicable down to such low metallicities, then the conversion ratio is 11.5 times that of the Milky Way, and our upper limit for Leo A corresponds to an MH2≤2·106 Msun. For other galaxies in our sample this number goes as low as 2·105 Msun, levels at which just a few giant molecular clouds are detectable.

Fig. 2 shows a plot of the distance independent quantity, L(CO) / M(H I) versus oxygen abundance, where L(CO) is computed via: LCO = ICO AS, where AS is the source area (e.g. Sanders et al. 1991). For comparison purposes, we have supplemented our new observations with data from the literature. The normalization by neutral hydrogen gas mass, M(H I), allows us to study the CO luminosities in a sample of diverse galaxies using a quantity that might be termed a "specific CO luminosity". Since most galaxies have not been mapped in CO with an interferometer, we use the telescope beam size at the distance of the galaxy as a consistent first order estimate the source diameter. Taking the definition of dwarf to be MB>-18, there are only 21 dwarfs in the literature which possess all three of CO, H I, and oxygen abundance measurements. Our observations have added two new galaxies, DDO 187 and Sextans A to the list, and have improved the sensitivity of the remaining three.

[Click here to see Fig. 2!]

4. Conclusions

As Fig. 2 clearly shows, there is a dearth of CO detections in dwarf galaxies a metallicities less than 8.0, the only known exception being I Zw 36 (Tacconi & Young 1987). This is strong evidence that the CO emission in a galaxy depends upon the metallicity. With so many lower limits at the low metallicity end, we cannot determine the form of this dependency. Fig. 2 also provides evidence that the CO emission depends upon a second property, because the scatter in L(CO) / M(H I) at a given abundance is quite large. This is most noticeable at 12 + log O/H = 8.2, where there are 4 galaxies with L(CO) / M(H I) scattered over 1.5 orders of magnitude, a range much larger than the errors.

The primary purpose of CO observations often is to understand the molecular gas. In dwarf galaxies, however, it is clear that CO is often not useful as a tracer of the molecular ISM. It is possible that other tracers may be developed in the future, such as atomic and ionized carbon emission (Maloney & Wolfire 1997) or millimeter and submillimeter dust emission.

References

Links (back/forward) to:
First version: 12thAugust,1998
Last update: 28thSeptember,1998
Jochen M. Braun   &   Tom Richtler
 (E-Mail: jbraun|richtler@astro.uni-bonn.de)