Dear Readers,
I am hesitant to set this precedent, but I have agreed to put out a special issue of the newsletter to get one more white paper in press before next week's SIRTF science working group meeting. This November 6, 1998 issue will be given the number 5a.
The deadline for submissions to the next issue of the newsletter is still Friday the 13th of November.
Doug Kelly, editor (dkelly@as.arizona.edu)
In the appendix we summarize the results in this science area from the recently-concluded ISO Science conference. The summary of the summary is that the imaging/photometry reported at the meeting does not dramatically change the picture produced by IRAS - of order 25% of main sequence stars in the solar neighborhood have debris disks down to a level of order 100 zodis. The time scale for dissipation of the disks is hundreds of millions of years. A generally new idea emerging from the conference is that in many astrophysical environments - particularly circumstellar, both pre-main sequence, main sequence, and post-main sequence - crystalline silicate minerals with a variety of spectral signatures are seen. These signatures cover wavelengths from ~10 to ~40um. Their study could easily become as much of a cottage industry as PAHs are in the hydrocarbon world. This could be the subject of a whole separate Legacy program, but it does give added urgency to any spectroscopy we would do on disks. Parenthetically, a really obvious question which SIRTF ought to answer is: given that the crystalline silicates are common in all circumstellar environments, why aren't they seen in the diffuse ISM, or are they?
It goes almost without saying that a study of the debris disk phenomenon is a good thing to do with SIRTF. By debris disks, we mean disks associated with main sequence stars which are beyond the T Tauri or "Young Stellar Object" phase. Clearly these debris disks bridge neatly into the Protoplanetary disks on the younger side and into the Sun's zodiacal cloud on the highly evolved side. A characteristic of these disks is that the lifetime of their particles against expulsion from the system or against destruction from PR drag is shorter than the lifetime of the star. Thus it is generally assumed that they are replenished by grinding up of larger bodies within the disks - analogous to asteroids, comets, planitesimals, or whatever. Thus the disks are taken as signposts of planetary formation - or planetary system formation. Note that in addition to the great intrinsic scientific interest of this phenomenon, SIRTF studies may have considerable programmatic value, because they will help to define the parameters of NASA's forthcoming TPF [Terrestrial Planet Finder] mission.
A MIPS/SIRTF Program can be broken down into several categories of exploration:
Note that the importance of a. and c. above cannot be overstated, because these will be the systems which it is possible to resolve and image spatially. This provides important "ground truth" for the interpretation of the spectral energy distributions of stars with excesses - of which SIRTF can obtain an almost unlimited number. We think that establishing this fundamental set of "touchstone" disk properties should be a focus of the MIPS GTO program, although it will also require coordination with the other instrument teams - with IRS for certain.
To keep things manageable, we suggest that the MIPS program focus on programs 2 and 3b/3c above, leaving 1 to the MIPS-IRAC cluster team and 3a/3d to coordination with other GTOs. This proposal is roughly consistent with our submission to the November MIPS team meeting [dated 11/17/97] in which we had three programs:
Taken together, these programs are designed to give insights into:
The number which we measure most directly with MIPS is FE(lam), the excess flux over the stellar photosphere at wavelength lambda. For precision, we define this as:
FE(lam) = FD(lam) divided by FP(lam), that is, it is the disk flux divided by the photospheric flux at that wavelength. A system for which the disk is as bright as the photosphere has FE(lam) = 1. On the other hand, the quantity of greatest interest is the fractional luminosity radiated by the dust at all wavelengths, which is equal to the fraction of the starlight absorbed by dust and thus a fairly direct indicator of the amount of dust. Let us call that quantity
TAU = power radiated by dust/power radiated by star
For reference, TAU = 3e-3 for Beta Pic, 2e-5 for Vega, and 8e-8 for the zodiacal material in our own solar system. We should take as a target for the MIPS/SIRTF program to get down to within 10 zodis, or TAU = 1e-6 in round numbers on a routine basis. I have plotted TAU vs. FE(100) for several of the classic Vega stars from the IRAS survey. There is a roughly linear relationship, and for Vega, FE(100) = 17. Therefore, by analogy with these stars, an A star with TAU=1e-6 should have FE(lam) ~ 1 for wavelengths between 70 and 160um (the longer MIPS bands). Thus we can use the extrapolated stellar photosphere as a figure of merit for the observability of an excess. This is a fairly rough criterion because I haven't figured in the dependence on the stellar temperature - cooler stars will probably have lower FE(lam) for the same TAU, whereas colder dust would lead to a larger FE(lam) for the same TAU. But it is a good place to start.
From the expected SIRTF sensitivities [5-sigma, 500-s] I find the following limiting magnitudes for detection of +photospheric emission from a star [based on zero magnitude = 40Jy @ 10um and a Rayleigh-Jeans extrapolation]:
24um, limiting magnitude = 11.2
70um, limiting magnitude = 7.3
SED mode, limiting magnitude = 5.3
160um, limiting magnitude = 3.3
For stars of this brightness having FE(lam) = 1 in the MIPS bands, the 500 sec integration would lead to a 10-sigma detection of the [photosphere + dust]. Thus these are not unreasonable magnitude targets for an initial survey, but we could go for fainter visual magnitudes due to the infrared- visual color index and the ability to integrate longer than 500 sec. So one might start with the following initial limiting visual magnitudes:
24um: mv = 12
70um: mv = 8
SED: mv = 6
160um: mv = 4
What this means, for example, is that for stars brighter than mv = 6 it would be reasonable to search for a 10 zodi disk using the SED mode, etc.
We would carry out measurements of a selected sample of stars [say ~ 250 total] selected by various criteria to a level of 10 zodis [based on refinement of above criterion]. We would work on stars of spectral types A-M - say 50 stars of each spectral type. Wherever possible, these could be nearby stars to do double duty with program 3a, and in any case we presumably would select the nearest and brightest examples for each of our categories. Table 1 lists statistics of single stars within 15 pc derived form the Hipparcos and Gleise catalogs. Binaries closer than 10" have been removed from this tabulation; some would be added in as part of an investigation of importance of binarity.
These 50 could then be broken down into several subgroups sorted by age, metallicity, rotation, binarity, x-ray emission, etc. We suggest a "rolling wave" observing strategy based on the above numerology. That is, stars brighter than 4th magnitude would be measured in all four MIPS channels [160, SED, 70, 24], whereas stars between 4th and 6th magnitude would be measured with all but 160um, etc. This strategy may need a little rethinking and should probably be stretched to enhance our sensitivity to lower temperature dust than that associated with the A stars used to define my templates. In any case, stars with photometric evidence for lots of dust could become candidates for follow on observations - IRS spectroscopy, photometry in the remaining MIPS bands, super-resolution imaging, etc. If there are still but a few stars with kinematically detected planetary companions (3d), they could reasonably be included in this program. If there are lots of them, perhaps they need their own program. Table 1 shows that stars in this Statistics of Dust Disks program could do double duty in the "nearby star" program - their proximity to the Sun would make them good candidates to search for brown dwarf and faint companions.
A good figure of merit or success criterion for our survey is the detectability of the stellar photosphere. To assess this, Figure 1 shows the photospheric emission based on a simple black body extrapolation for stars of spectral types A-M within 15 pc, taken from Table 1. Each star appears at five positions within the vertical line for that spectral type. Sensitivity values are given for the five relevant SIRTF modes:
Figure 2 shows predicted ratio of dust to photospheric emission (FD/FP) for two wavelengths. This ratio is given by Fdust/Fstar = alpha (Ts/Td)^4 B(Td)/B(Ts). The figure is drawn for 70 K dust with alpha =Ldust/Lstar=10-6. This shows that the FD/FP depends on the stellar temperature, for fixed TAU, as stated earlier. Thus for cool stars it will be considerably harder to get down to TAU = 1e-6 than for hotter ones. It also suggests that knowledge of the predicted stellar photospheric emission to better than ~10-20% will be necessary if we are to explore this region of parameter space effectively.
Here the program is pretty straight forward - beat away on these systems with the full power of SIRTF, including super-resolution imaging at all MIPS bands [if indicated] and - in particular - low and high resolution spectroscopy with IRS. Note that an object like HR4796 is too small to be resolved by MIPS but is a great candidate for spectroscopy, while nearer systems like Vega and Beta Pic should get the full treatment. We should probably limit the size of the sample to about one dozen objects, as each is several hours of SIRTF time and a potential ApJ paper by itself. These are by definition bright systems for which signal/noise for imaging, photometry, and spectrophotometry is not an issue and for which super-resolution is feasible.
We should also ensure that low spectral resolution IRS spectroscopy of the prominent disks is carried out to match up with the MIPS SED information and to look for a wide variety of solid state features from minerals and ices. There will undoubtedly be strong interest among the IRS team in observing these objects, so we will need to coordinate this program with them. Note that the low resolution spectrophotometry is a means of exploring both the structure [gaps, density gradients, etc] and the composition of the disk, as the symmetry of the situation should allow the spectrum to be readily inverted.
We should consider IRAC imaging at 3.6um in some cases to search for a scattered light component. The idea is that the higher resolution potentially achievable with SIRTF could lead to an improved image, and also that the extent of the disk might be greater in scattered light than in thermal emission, at least in the SIRTF bands.
For Vega, with TAU = 2e-05, we can estimate that the total scattered light is 2e-06 times the stellar radiation, taking a grain albedo of 0.1, corresponding to rather efficient scattering. This is not an unreasonable assumption, given that the grains are known to be large in these disks. The corresponding flux, given that Vega has zero mag at 3.6um, is 300uJy, which would be readily detectable from IRAC even if spread over an annulus on the sky of inner radius 3", which is the size for the Vega disk inferred from the models reproducing the thermal infrared behavior. However, the real issue is the brightness of the diffracted and/or stray light from the central star, which makes this particular example look a little bit unfavorable [actually a whole lot unfavorable]. On the other hand, the Beta Pic disk extends over many arcsec in visible scattered light. So perhaps another way of looking at this is to see whether there is a case where the scattered light could be seen from SIRTF in the near infrared but not be visible from HST or from a well-optimized ground-based telescope [by construction, the contrast of the disk to the star is independent of wavelength for this scattered light component].
A. Robberto reported on results of Beckwith et al GTO program.
They studied 5 clusters within 100-200pc:
2602
Chameleon
Alpha Per
Pleiades
NGC7092
For each, they looked at 10-30 stars, to limits of 10mJy@25um and 100mJy@60um, 3-sigma. Their experiment was sized to detect 0.1 Earth masses of dust. The result was no detection in systems older than 10 millions years, and that only 5 of 77 stars were detected at 60um. This argued for early dissipation of the disks, which was very controversial. It was noted during the discussion that x-ray emission can also be used as an indicator of stellar youth.
In Chameleon they saw a number of systems with silicate emission, perhaps attributable to an atmosphere or warm region above the accretion disk..
B. Dominik reported on results of Habing et al GTO program.
They did a volume-limited sample, D<25pc, excluding O,B,M,variable, binary stars, etc. and limited themselves to cases where the predicted 60um photospheric emission was 50mJy. This led to a sample of 84 stars, and they observed another 7 for TBD reason. Results are as follows:
Stellar Type Definite Detection Maybe Detection A 6/15 9/15 F 6/28 11/28 G 4/26 6/26 K 3/22 6/22
To summarize, ~20% have definite detections, and ~35% have maybe detections. Their detection criterion was a 2-sigma excess above the photosphere, which is claimed to be ok for a statistical study [not sure if this is for the "maybe" or the "definite" column]. It seems likely that there is a selection effect which makes the [more luminous] A stars easier to detect, so the trend with stellar type may not be real...but it is grist for SIRTF's mill.
The assertion was made that there are age estimates for 95% of the stars in their sample. On this basis, they estimate that the dissipation time scale for a disk is ~400 Myr, with "lots of variation". This is reportedly the same time scale as estimated for the Kuiper Belt (Stern and Colill). It is also very the same number derived by Becklin et al from their GTO program on disks.
The Becklin et al sample consisted of F and G field stars with excess [IRAS] at 60um, stars in open clusters (Alpha Per, Coma Berenices, Hyades, Pleiades, Ursa Majoris), and a group of classical and weak-line T Tauri stars. They analyzed the data by making composite spectra of the objects in a given category [i.e. a given cluster or the field stars together] and derived a range of dust densities [actually effective optical depth to starlight - the fraction of the total power that comes out in the infrared] for each such category. Combining this with data on other objects ranging from HR4796 to the sun, they derive a time scale for disk dissipation which is also ~400Myr, but also shows considerable variation. Some rather old stars show strikingly large optical depths.
C. Lagage reported on spectroscopy and imaging of the Beta Pic system. Spectroscopically, a silicate emission feature is seen, and the signature of crystalline silicates [rather than amorphous silicates which dominate in the ISM] was noted. In imaging at ~13um, they were able to resolve the disk within a few pixels of the star [1.5" pixels]. In a poster talk, Pantin et al reporta spectrum of Beta Pisk going out to 45um which shows a crystalline ice peak at 43um.
D. Helen Walker et al reported some imaging data on a few stars - Vega, Beta Pic, Fomalhaut, Epsilon Eri, HD169142, and HD142666, several of which were "resolved".
My editorial comment is that the imaging/photometry reported at the
meeting does not dramatically change the picture produced by IRAS - of
order 25% of main sequence stars in the solar neighborhood have debris
disks down to a level of order 100 zodis. The time scale for dissipation of
the disks is hundreds of millions of years. A generally new idea emerging
from the conference is that in many astrophysical environments -
particularly circumstellar, both pre-main sequence, main sequence, and post-
main sequence - crystalline silicate minerals with a variety of spectral
signatures are seen. These signatures cover wavelengths from ~10 to
~40um. Their study could easily become as much of a cottage industry as
PAHs are in the hydrocarbon world. This could be the subject of a whole
separate Legacy program, but it does give added urgency to any
spectroscopy we would do on disks. Parenthetically, a really obvious
question which SIRTF ought to answer is: given that the crystalline silicates
are common in all circumstellar environments, why aren't they seen in the
diffuse ISM, or are they?
To Top