Astron. Astrophys. 356, 795-807 (2000)

2. Submillimeter maps

ABR obtained the deepest images yet of a nearby galaxy in the submillimeter waveband. They detected emission from of the optical disk, down to a noise-limited surface brightness of 3.5mJy/beam and 13mJy/beam at 850 and m respectively. At both submm wavelengths, the flux exhibits a pronounced peak at the nucleus and secondary maxima either side of the centre (equivalent to 3 kpc for a distance of 9.5 Mpc, adopted by both ABR and XAD, to NGC 891). The secondary maxima may be attributable to a ring structure seen edge-on or limb-brightening associated with spiral arms. We do not present the SCUBA maps here, but refer to ABR for both these images and more technical details vis-a-vis the observing procedure. The submm maps are much more radially-extended than the corresponding emission detected by IRAS, indicating that longer wavelength FIR radiation dominates at larger galactic radii.

ABR made a careful estimate of the ratio of cold dust to warm grains by smoothing all their submm/FIR images of NGC 891 (60,100,450 & m) to the same resolution and then dividing the major axis into spatially-independent radial bins. This uniform treatment of the data was considered of major importance because there had been a tendency, in the past, to compare photometry at different wavelengths without regard for, or knowledge of, differences in the size of the relevant emitting region(s) (e.g. Guelin et al. 1993; Israel et al. 1999; Clements et al. 1993). The flux density emanating from each radial bin was matched by the superposition of 2 greybody curves. Thus:

where is the flux density at wavelength (in Wm^-2Hz^-1), is the geometrical cross-sectional area of an interstellar grain, D is the distance to NGC 891 and is the submm emissivity. and are the number of emitters (grains) in the warm and cold dust components respectively whilst and represent the corresponding blackbody intensities. is sometimes refered to as the emission efficiency and can be considered as the ratio of the emission cross-section to the geometrical cross-section of the grain (Spitzer 1978; Whittet 1992).

ABR found that all radial bins could be fitted well with =15-20K  ¹ and =30-37K, and increased from about 26 at the nucleus to 54 at . The ratio is fairly secure. However , the exact dust masses pertaining to and depend critically on the absolute level of which, as we shall see, is highly uncertain. At this stage, it is important to emphasize that over 90% of the m emission arises, at all radii, from the cold (15-20K) dust component and that this material, in turn, constitutes the bulk of galactic dust. In other words, our m SCUBA map indicates, almost completely, how galactic dust is distributed within NGC 891. Moreover, if we could avail ourselves of a well-determined value for the grain emissivity at m, , our SCUBA image would yield, to a good approximation, the total dust mass in the disk.

The submm/FIR emissivity, , is a poorly known quantity and virtually the only sources of information stem from Hildebrand (1983) and Draine & Lee (1984), which are cited almost exclusively in the literature. The measurement of Hildebrand (1983) is based on Kuiper observations (55 and m) of a single Galactic reflection nebula, NGC 7023, where attenuation of ultraviolet light from the central star is effectively used to constrain the amount of dust present. In order to account for the observed FIR flux densities, the emissivity at m was estimated to lie in the range . Although this represents little better than an order of magnitude determination, the emissivity of Hildebrand has been applied by FIR observers almost religiously since its derivation nearly 2 decades ago. This is at least partially attributable to a dearth of measurements in the interim years, in particular observations applicable to the diffuse interstellar medium (ISM) where grains may well differ from their particulate counterparts in reflection nebulae (Pendleton et al. 1990; Mathis 1990). We note that Casey (1991) has repeated the Hildebrand procedure for 5 Galactic reflection nebulae (and derived an emissivity approaching the upper limit of the Hildebrand determination), but this work has been seldom cited in the literature. Draine and Lee (1984) have used a pot pourri of laboratory measurements and astronomical data to construct dielectric functions appropriate to graphite and silicate interstellar grains. In this way, they have derived a FIR emissivity close to the mean value given by Hildebrand. Finally, we note that Bianchi et al. (1999) recently determined an emissivity value of for the diffuse dust in the Milky Way at m. Their technique relied on the tight spatial correlation observed between interstellar extinction and the FIR radiation detected by COBE. This allowed them, in a similar fashion to Hildebrand, to constrain the amount of dust responsible for the FIR emission. Depending on the exact behaviour of Q with wavelength (discussed below), the Bianchi et al. value is somewhere towards the higher end of the Hildebrand range.

In the following sections, we attempt an independent determination of the submm/FIR emissivity in NGC 891 by using extinction observed in the disk to constrain the amount of grain material present. This approach differs from previous treatments of extragalactic submm/mm observations where, to one extent or another, workers have tended to make use of dust properties in the Milky Way in order to infer quantities in external galaxies. For example, Guelin et al. (1993) use a Galactic emissivity to turn their 1.3mm observations into a dust column density. Assuming a solar gas-to-dust ratio they then quantify how much (molecular) gas is present in the system. In a slightly different approach, Neininger et al. (1996) use 21cm emission to infer a dust column density (again via a Galactic gas-to-dust ratio) and then compare this quantity with their observed submm/mm emission in order to derive the corresponding emissivity. Our current method is based on the technique pioneered by Hildebrand (1983). In essence, the dust column density is measured by means of optical/ultraviolet extinction and this quantity is then compared to the observed FIR/submm emission in order to infer the long wavelength emissivity. In deriving we make no direct use of Galactic properties as such. Whilst we follow the same general method as Hildebrand, we believe that the submm window produces more robust values of emissivity than fits made to the emission near the greybody peak (Hildebrand fits near m). The reason for this is that, for any observed level of FIR/submm flux, the implied dust mass will vary at least as strongly as for emission near the peak but only in the Rayleigh-Jean tail (where T is the grain temperature). If the dust mass is fixed a priori by the amount of extinction, the corresponding uncertainties are then transfered to the derived values of emissivity. Thus will be much more dependent on the grain temperature than and, as a consequence, far more uncertain (see Hughes et al. 1997 for an indepth discussion).

Before we are able to embark on this process for NGC 891, we have to be sure that the submm, which traces the bulk of the dust in this galaxy, corresponds closely to the grain material responsible for the absorption lane. To this end, we briefly describe the photometric fit made to NGC 891 by XAD which yields directly the optical depth at each point in the disk.

© European Southern Observatory (ESO) 2000

Online publication: April 17, 2000
helpdesk.link@springer.de