2.5. Should there be a universal dust model?

ZDA discovered a total of fifteen viable dust models that satisfy the extinction, IR emission and abundances constraint in the local ISM. Is there any way to discriminate between these models? COMP grain models differ from BARE ones, because a significant fraction of their dust particles are fluffy composites containing voids. Composite particles have therefore an effective electron density that is significantly smaller than that of the bare particles. Consequently, observations of X-ray halos can, in principle, discriminate between the different classes of viable dust models [21] since X-ray halos are primarily produced by X-rays scattering off electrons in large grains. However, X-ray halos sample a very limited fraction of the general ISM. So even if an X-ray halo would favor one model, it would not preclude the viability of others in different regions of the ISM, since dust properties exhibit significant variations along different lines of sight.

Some of the observational evidence for such variations are:

• the variations in the steepness of the FUV rise and the strength of the 2175 Å bump [26]
• the richness of mineral structures and ices seen in the evolved stars that are absent in the diffuse ISM [63]
• variations in the elemental depletion pattern in the hot, warm, and cold phases of the ISM [61]
• variations in the PAH features in different regions of the ISM [5, 56]
• variations in the strength and width of the silicate features [14]

These variations probably result from the existence of a large variety of dust sources producing dust with different composition and mineral structure, and from the fact that the ISM is not homogeneously mixed. Furthermore, grain processing in the ISM by thermal sputtering, grain-grain collisions, grain coagulation, and accretion in clouds plays an important role in producing large spatial variations in dust properties [37]. Such variations are manifested in the observed UV-optical extinction and IR emission from galaxies. Evidence for spatial variations in the extinction was provided by Keel & White [40], who analyzed the extinction properties of dust in spiral galaxies that are partially backlit by an elliptical one, and by Clayton [7], who reviewed extinction studies of the Magellanic Clouds and other nearby galaxies. There are too many observations showing the spatial variations in the IR emission spectrum from galaxies to list here, but many can be found in the special issue of The Astrophysical Journal Supplement Series (volume 154) reporting the first results from the Spitzer satellite.