7 Summary

We use B-band CCD images to search for the presence of extraplanar dust or dust chimneys in a sample of 10 edge-on, spiral galaxies which we had previously observed in the H${\alpha }$ emission line (Rand 1996). The existence of such features in nearby NGC 891 has been known for some time. However, their overall occurence in quiescent (non-starburst) disks, as well as the manner in which they are formed, has only come under investigation very recently. We quantify the amount of extraplanar dust by identifying (with the aid of unsharp-masking) extinction features which cannot be accounted for by the standard absorption dust lane. This can only be done reliably for 5 of the most edge-on galaxies in our sample ( $i>87^{\circ}$) and we use a scattering+absorption radiation-transfer model (Xilouris et al. 1999) to separate "vertical'' extinction features from the conventional dust disk in these cases. We investigate the possibility that dust is expelled from stellar disks by localized star-formation by comparing the B-band unsharp-mask frames with our H${\alpha }$ images. Our main conclusions from this study are as follows:

1.

Of the 5 galaxies deemed sufficiently edge-on to warrant a search for extraplanar dust, 3 show strong evidence for dust chimneys and isolated high-z extinction clouds. The linear features extend up to 2 kpc above the midplane with some simply terminating at high latitudes and others reconnecting with the disk. In general, the extraplanar features are poorly resolved in our images with the consequence that the opacities we derive ( $\tau_{B} \simeq 0.15$) are too low by a factor of 5 or so. The masses we recover for the individual dust structures are expected to be correct and are of order 10⁵ $M_{\odot}$ (assuming a solar-type gas-to-dust ratio). With this technique, the total amount of dust we find outside the standard dust lane is $\sim$ 1%;

2.

If we combine our results with a very recent similar study by Howk & Savage (1999), we believe that 5 out of the 10 disks, which are sufficiently inclined to allow us to probe the extraplanar layer (i>87), possess high-lattitude dust. We concur with Howk & Savage that the "chaotic'' dust lane in NGC 891 cannot be considered anomalous and vertical chimneys probably occur in half of all quiescent spiral galaxies;

3.

Our use of optical extinction to identify extraplanar dust is heavily biassed towards dust clouds on the nearside of the disk and relies on sufficient background light from the bulge to make attenuation conspicuous. Consequently, we explore the transparent submillimeter thermal emission recently detected from cold dust in NGC 891. Profiles in z-height, at 450 and $850~\mu$m, constrain the quantity of extraplanar dust to <5% of the total galactic dust mass;

4.

We can establish no detailed spatial correlation between dust chimneys and either H${\alpha }$ emission emanating from the disk (HII regions) or the near-halo environment (Diffuse Ionized Gas). This is something of a surprise since far-infrared bright galaxies (such as M 82) are known to entrain optically-thick dust at the working surface of the starburst outflow. The Diffuse Ionized Gas in quiescent galaxies is believed to arise from O-stars in HII regions keeping high-latitude (z=1-4 kpc) gas in a state of ionization. Indeed, for our edge-on sample, brightest Diffuse Ionized Gas emission occurs above areas of the disk experiencing relatively high rates of recent star formation. There is also some possibility that we find dust chimneys occurring more frequently in galaxies with higher overall star-forming activity. The fact that recent star-formation and extraplanar dust are not coupled locally (but might correlate globally) may suggest a difference in timescale between the two phenomena. Chimneys probably take $\sim 10^{7}$ yr to form, regardless of whether they are driven out by radiation of gas pressure. On the other hand, H${\alpha }$ emission traces O-stars and these massive stars are known to possess lifetimes of $\sim 10^{6}$ yr;

5.

We examine H${\alpha }$ and optical images of the nearby edge-on disk galaxy NGC 55. Its proximity (D=1.6 Mpc) offers vastly superior resolution over galaxies comprising our main sample. The most active star-forming region near the centre of NGC 55 appears to have prised dust clouds away from the midplane. However, more generally, we find no obvious correlation between local disturbances in the dust lane and increased H${\alpha }$ emission;

6.

We investigate the possibility that the Milky Way might contain extraplanar dust. It is conceivable that the supershell structures described by Heiles may relate to dust chimneys. However, there is little evidence for high-z dust from observations of local cirrus and reddening measured towards local low mass stars;

7.

Dust chimneys are unlikely to be significant contributors to intergalactic enrichment. Dust and metals detected in clusters such as Virgo and Coma can probably be attributed to phenomena such as starburst superwinds, tidal interactions and ram-stripping. Despite this, dust chimneys must play an important role in the circulation of gas, grains and heavy elements around the interstellar medium. Chemical mixing of this kind allows star-formation to proceed efficiently within spiral disks.

Acknowledgements

We thank the archival services of the ING and AAO telescopes for making CCD data available to us so quickly. We are also grateful to the IRAS archivists at IPAC, Caltech. PBA acknowledges his continued financial support from PPARC.