Annu. Rev. Astron. Astrophys. 1991. 29: 59-88 Copyright © 1991 by Annual Reviews. All rights reserved

3. DUST SCATTERED GALACTIC RADIATION AS THE PRIMARY SOURCE OF THE FAR ULTRAVIOLET BACKGROUND

Radiation from early stars scattered by Galactic dust is a candidate for the primary source of the diffuse far ultraviolet background. Indeed, with our present knowledge, it is tempting to employ the term ``obvious candidate,'' but this does not reflect the near consensus outlook of workers in the early 1980s. There were two widely held objections to starlight scattered by dust as the source of this radiation: (a) the intensity of the radiation was believed to be uniform across the sky, yet dust was believed to be sparse to nonexistent at high Galactic latitudes, and (b) dust was believed to be strongly forward-scattering in the far ultraviolet, and hence even if dust were present at high Galactic latitudes, radiation from hot stars in the plane would not be scattered back to our midplane position.

Although a fair amount of optical work has since been carried out which indicates that dust is widely distributed at high Galactic latitudes, results from the IRAS satellite have been sufficiently dramatic that this point is no longer contentious (Low et al. 1984). However, even now it is not clear if high latitude dust is omnipresent or just widespread. (I return to this point at the end of this section.)

The second factor, the character of the scattering properties of dust in the far ultraviolet, is still a matter of dispute. Substantial work by a number of investigators has been carried out in an effort to determine these parameters. Many of these investigations used the instrumentation on the OAO-2 or the IUE satellite, but a number employed specialized instruments specifically designed for studies of dust. In the analysis of their data, most workers employ the formalism of Henyey and Greenstein (1941) in which scattering is characterized by two parameters: the albedo ( = 0 implies complete absorption; = 1 implies complete reflection), and the scattering asymmetry factor of the grains (g = 0 implies isotropic scattering; g = 1 implies total forward scattering). In Table 2, I list the results obtained by various workers on this topic.

It is readily apparent that there is no systematic agreement in these results. The reasons for these differences are not entirely clear, but some specific problems can be cited. Studies of individual dust clouds are made difficult because of uncertainties in the scattering geometry of the cloud/star system, e.g., is the cloud in front of, or behind, the illuminating star? Large-scale studies of the diffuse Galactic background are difficult because of the problem of separating diffuse scattered light from starlight in view directions at low Galactic latitudes. Measurements at low latitudes are crucial since they fix the albedo, which then allows for a solution for the scattering asymmetry factor. For example, the analysis of the extensive D2B data set (Joubert et al. 1983) was limited to | b | > 30°. Consequently, the albedo could not be determined but had to be assumed, and the result derived for the asymmetry factor is crucially dependent upon the assumed albedo. Simple errors can lead to incorrect results. Fix, Craven, and Frank (1989) obtained a large value for the asymmetry factor by fitting their plane-to-pole data set with standard formulae; unfortunately, the formulae used were not appropriate for their case since they neglect an optical depth term; this leads to an incorrect result. Finally, it should be noted that some of the results in Table 2 rely on the relative calibration of several instruments, which introduces a systematic uncertainty in the derived values.

Table 2. Dust properties in the far ultraviolet (FUV)

Workers	Target	Results
Andriesse, Piersma, and Witt (1977)	NGC 1435	g 0.25
Jura (1979)	NGC 1435	g 0.2
Carruthers and Opal (1977)	Orion	``Albedo is high''
Witt and Lillie (1978)	Orion	higher in FUV than at 2175 Å bump; scattering more isotropic in FUV than at longer
Bohlin et al. (1982)	Orion	0.5 < < 0.7
Tanaka et al., Onaka et al. (1984)	Orion	0.3 < < 0.6, 0.2 < g < 0.5
de Boer and Kuss (1988)	Orion	g 0.6, and g similar across FUV band
Donas et al. (1981)	M101	g lower in FUV than at visible
Witt et al. (1982)	NGC 7023	0.6; g 0.25
Witt, Bohlin, and Stecher (1986)	CED 201, IC 435	similar in FUV and at 2700 Å if g constant
Lillie and Witt (1976)	DGL	0.4 < < 0.6, 0.7 < g < 0.9
Anderson, Henry, and Fastie (1982)	DGL	Either g > 0.9 or < 0.2
Joubert et al. (1983)	DGL	0.6 < g < 0.7 (assumed = 0.5)
Jakobsen et al. (1984)	DGL	x (1-g) 0.16
Onaka (1990)	DGL	x (1-g) 0.07
Fix, Craven, and Frank (1989)	DGL	g > 0.9
Hurwitz, Bowyer, and Martin (1991)	DGL	= 0.16 ± 0.03, g < 0.2 (80% confidence)

The Berkeley UVX instrument had particular advantages for the investigation of the ultraviolet scattering properties of dust. Because of the specialized characteristics of this instrument, observations of diffuse radiation could be made over a large range of reddening, free of the interfering effects of starlight or instrumentally scattered radiation (Hurwitz, Bowyer, and Martin 1991). The resultant data clearly display saturation effects at higher values of hydrogen column density (and hence larger reddening). (See Figure 2 for an example of this effect.) The high Galactic latitude observations of most workers can be reproduced almost equally well with a combination of either a high albedo and high g or a low albedo and low g. However, the observations of Hurwitz et al. include low latitude targets, which by necessity include a relatively high density of stars. By exploiting the imaging properties of the Berkeley UVX experiment, these stars were identified and their signals removed from the data. The resultant diffuse background at low Galactic latitudes was surprisingly faint and was consistent only with a low value for the albedo, independent of g. If the albedo of high latitude grains is similar to that found for the low latitude regions, then g must necessarily be low. A detailed analysis of these data gives best-fit results of 0.16 for the albedo and zero for the scattering asymmetry factor; the range of allowed values is listed in Table 2. These results are quite surprising and suggest that the grains scattering the ultraviolet radiation are of different size and character than those producing scattering in the visible and emission in the infrared. Although these results are unexpected, they appear to be well founded (though it can be reasonably argued that this author is not the most objective judge of this statement, and it must be emphasized again that this field is acknowledged to be fraught with difficulties and uncertainties). In any case, these results are the product of a single investigation and were obtained from a limited, and perhaps atypical, region of the sky.

The question of the extent of Galactic dust at high latitudes is of intrinsic interest and will have an impact on a variety of observational programs. It might be expected that data from IRAS and COBE could resolve this issue, but this is not the case, at least at present (e.g., Smoot 1991). The IRAS data are inconclusive because of uncertainties in the zero point offset of the instrument; data from COBE do not suffer this problem, but both instruments are subject to uncertain contamination from zodiacal dust at the Galactic poles.

To examine this question, Martin, Hurwitz, and Bowyer (1990) compared two far ultraviolet spectra obtained with the Berkeley UVX instrument at high Galactic latitudes. Studies in the far ultraviolet have the advantage that zodiacal light is not present shortward of ~ 2000 Å. (The exact wavelength cutoff depends upon the detailed observational parameters, but for view directions well off the plane of the ecliptic, the zodiacal light cannot be detected below 1900 Å [Tennyson et al. 1988].) The continuum from the view direction that includes a small but measurable dust component has the same shape as but a higher intensity than that obtained from the lowest intensity view direction. The authors therefore conclude that dust is also the source of most of the continuum emission in the lowest intensity view direction. Since this region seems typical of the lowest hydrogen column density view directions in the Galaxy, Martin et al. conclude that at least some far ultraviolet scattering dust is likely to be present everywhere.

The far ultraviolet intensity-to-N_HI is different in the direction of lowest N_HI from its value at higher N_HI. Martin et al. prefer the explanation that it is the dust associated with ionized gas that contributes to the excess far ultraviolet intensity in the low N_HI target, but they cannot rule out alternative, but somewhat forced, scenarios.

Hurwitz, Bowyer, and Martin (1991) also suggest that dust is present in all view directions in the Galaxy. In their derivation of the far ultraviolet dust scattering properties, one of their independent variables is the residual dust scattering at the poles. They are able to rule out the hypothesis that there is no dust at the poles at the 68% confidence level; unfortunately, the data are not sufficiently definitive to rule out this hypothesis at a higher level of significance.