6.1. In the infrared
Now that we have a consistent model of the disk, we can make some predictions, in order to test the two component model we have presented.
First, the ice has some characteristic features at 3.1 and around 50 m. The feature at 50 m should be detectable by the ISO LWS and SWS). Fig. 18 shows an infrared spectrum covering the range 5-100 m. An wide emission feature very characteristic of ice can be seen, around 45-50 m. In the 10 m spectrum, the mantle of refractory organics creates an emission feature around 6-7 m. These features should be easily detectable in LWS and SWS spectra. Finally, we predict also that characteristic crystalline olivine features should be seen in the mid-infrared spectrum at 24, 28 and 35 m.
As noted before, the ISO infrared camera ISOCAM should essentially (in spatial extension) image the scattered starlight at 10 m when pointing at Pictori s. It should also reveal a change in the particle composition characterized by a rapid decrease of the thermal emission (distributed on four 1.5 arcsec pixels of radius), replaced (but with an order of magnitude lower) by the scattered starlight.
In Fig. 19 we show a predictive 20 m profile, that should be compared to future 20 m imaging observations of the Pictori s. At a wavelength of 20 m (by comparison to 10 m), we are sensitive to colder (more distant from the star) dust, thus allowing to probe more accurately the disk in the range 50-100 AU. As shown in Fig. 19, the 20 m would confirm the change of composition at 85-100 AU by a very steep decrease of the flux in that region, because of a sudden jump in the temperature of the grains since iced grains are colder than core-mantle grains. These data should also resolve the discrepancy between our models and Backman et al. measurements at 20 m that we attribute to decreasing atmospheric transmission (which depends critically on the weather) in the 20 m window leading to an underestimation of the flux.
6.2. In the visible
From the scattered flux we have deduced from our observations and our two components model, one notes that the scattered flux does not decrease when approaching the star. One must use a scattering model to compute the dust density and thus see clearly the inner gap in the dust. This can be understood from the superposition of two effects. First of all, the scattered flux is dominated by the iced outer component down to a very small distance from the star. Secondly, the flux received from the star decreases as r-2 (r beeing the distance to the star) a single dust grain, so, if the dust density increase in the central part of the disk (where the void is) is less fast than the decrease of that flux, one will observe a scattered flux slowly decreasing with r. Another significant effect is due to the scattering phase function: in all the physically realistic cases, the phase function shows an approximatively similar shape with a pronounced narrow maximum for scattering angles around zero. This means that the amount of scattered light, proportional to that phase function, increases rapidly for particles when they draw near the imaginary line joining the star and the observer and so they will dominate for small distances to the star. This increase can compensate a decrease of the particle density.
Thus, the simulations made using our density shows a scattered light profile that does not drop sharply when approaching the gap, but rather a fairly slow and continuous decrease.
We must be very cautious with the modeling of the scattered flux. At present, no rigorous way exists to reproduce theoretically the scattering due to porous particles, except using the Discrete Dipole Approximation (Draine, 1988) which requires a huge amount of computer resources to calculate the physical quantities (we are for the moment limited to a particle size of about 1 m). But with the everyday increasing power of the new computers, this should not remain a limit in the near future.
© European Southern Observatory (ESO) 1997
Online publication: April 6, 1998