Dear Readers,
Welcome to the second edition of the MIPS/IRAC GTO newsletter. This episode features three new white papers! Michael Pahre describes an IRAC program to study the spectral energy distributions and evolution of normal galaxies out to Z=1, George Rieke discusses a MIPS program to study the evolution of galaxies in clusters, and Charles Lada and Erick Young discuss the great discoveries that are possible with SIRTF by doing surveys of circumstellar disks. In addition, there is a brief editorial by me explaining the purpose of the newsletter and a note from Steve Willner describing the structure of the IRAC GTO science team.
If you have comments on any of the papers in this newsletter, you can send them to me and they will be included in the next newsletter. The deadline for submissions to the next newsletter is Friday, October 2.
Doug Kelly, editor (dkelly@as.arizona.edu)
I first heard about the GTO newsletter about a month ago, a week before the rest of you. I agreed to serve as editor and to write a letter of announcement. Like many of you, I thought the newsletter sounded like a great idea. But what would the newsletter be about? I asked George Rieke that question. He said that since IRAC and MIPS had agreed to coordinate science programs and time was running fast, the newsletter should serve as a forum for the rapid exchange of ideas to help define those programs. It could also serve as a forum for the introduction and discussion of new ideas for possible GTO science. Now that the newsletter is started, it is up to you, the members of the IRAC and MIPS science teams, to make this effort a success. As can be seen in this issue, we are off to a good start. I encourage all of you to participate in the discussion and to make sure that your favorite GTO program gets some airtime. In that way, you can be sure that the newsletter evolves to serve the needs of you and your science team.
To TopThis is an administrative note about how IRAC is planning for GTO observations. We have organized into three teams covering broad areas of research:
Other GTO's -- MIPS, IRS, general -- who are interested in collaborating in these areas are invited to contact the appropriate IRAC team leader. Mike Pahre, in this newsletter, describes one project that the Galaxy team is considering.
Instruments used: IRAC (MIPS under discussion [see below])
Total time: 25 hrs
The deep, Distant Galaxies GTO programs for IRAC and MIPS [see newsletter #1 agreement by G. Rieke] will produce dramatic images of galaxies at z>2 undergoing their formation stages. While such images will be impressive and exciting, it will be important to place the global properties of such galaxies in the context of the "normal" galaxy population at more modest redshifts. Without such a "comparison" galaxy sample, it will be difficult to assess the changes in the comoving number density of luminous and sub-luminous galaxies, the evolution of the global star formation rate, and the weak (or passive) evolutionary history of the oldest stellar populations. It will also be important to study the evolution of the SEDs of nearby galaxies in order to develop IR photometric redshift techniques---especially those based on the 1.6um "bump" in the rest-frame SED.
For such a comparison to be worthwhile it is necessary that: (1) the nearby galaxies be observed, selected, and analyzed in a manner that is similar to the distant galaxies (in the respective rest frames); (2) the redshifts of the nearby galaxies be known; and (3) the redshift distribution of the nearby galaxies be broad enough that the redshifts 1.6um feature moves through several imaging filters (i.e., 0<z<1).
Imaging with IRAC of "pencil beam" redshift survey fields---already studied by other researchers, and documented in the literature---will provide large galaxy samples which satisfy these requirements. Fields include those from the CFRS (Lilly et al. 1995), the Hawaii survey (Songaila et al. 1994), the Caltech survey (Cohen et al. 1996a, 1998), the CNOC2 survey (Carlberg et al. 1998), and the HDF Flanking Fields (Cohen et al. 1996b). In ~25 hours, these fields can be imaged with IRAC to moderate depths sufficient to detect passively-evolved galaxies two magnitudes fainter than L* at z=1, thus providing the SEDs of normal galaxies throughout the LF at 0<z<1.
Construction of flux-limited galaxy samples at 3.5<lambda<8.0um, along with a follow-up redshift survey to "fill-in" the remaining unidentified galaxies in such surveys, will allow the LF and its evolution to be measured during the last half of a Hubble time. Hierarchical galaxy formation models make specific predictions about the evolutionary history of the IR LF even at these modest redshifts (e.g., Kauffmann & Charlot 1998). Measurement of the IR LF and its possible evolution therefore constrain cosmological and galaxy formation models. Adding far-IR observations will allow for the calculation of the lower-redshift, global star formation rate per unit comoving volume.
It will also be crucial to the understanding of the far-IR background--- both in discrete sources and a DC background flux---to have a complete inventory of the far-IR properties of nearby, normal galaxies. The detection of an optical background light (Bernstein 1997) and the sum of the flux detected in galaxies at 2.2um with the Keck Telescope (Djorgovski et al. 1995; Moustakas et al. 1997; Bershady et al. 1998) suggests that two to three times more flux has been detected from normal galaxies than may have been included in the simple models used for comparison with the detection of the far-IR background (Dwek et al. 1998); the full accounting of normal galaxies is therefore crucial for the interpretation of a possible population of high redshift, dusty protogalaxies as a source of the far-IR background.
While no MIPS observations have been considered yet, they would provide important information on the global SEDs of "normal" galaxies that is a context for the Distant Galaxies, IR Ultraluminous Galaxies, and IR Background observing programs. It is the interest of the IRAC GTO panel to assess the possible interest of the MIPS GTO panel in developing a cooperative research program on normal galaxies at 0<z<1 which would benefit other GTO programs of both panels, in addition to providing exciting results by itself.
Bershady, M., Lowenthal, J. D., & Koo, D. C. 1998, ApJ, 505, 50
Carlberg, R., et al. 1998, in Large Scale Structure in the Universe, in press
Cohen, J. G., Hogg, D. W., Pahre, M. A., & Blandford, R. 1996a, ApJ, 462, L9
Cohen, J. G., Cowie, L. L., Hogg, D. W., Songaila, A., Blandford, R., Hu, E.
M., & Shopbell, P. 1996b, ApJ, 471, L5
Cohen, J. G., Hogg, D. W., Pahre, M. A., Blandford, R., Shopbell, P., &
Richberg, K. 1998, ApJS, in press
Djorgovski, S., et al. 1995, ApJ, 438, L13
Dwek, E., et al. 1998, ApJ, in press
Kauffmann, G., & Charlot, S. 1998, MNRAS, 297, L23
Lilly, S. J., Tresse, L., Hammer, F., Le Fevre, O., & Crampton, D. 1995,
ApJ, 455, 108
Moustakas, L. A., Davis, M., Graham, J. R., Silk, J., Peterson, B. A., &
Yoshii, Y. 1997, ApJ, 475, 445
Songaila, A., Cowie, L. L., Hu, E. M., & Gardner, J. P. 1994, ApJS, 94, 461
A variety of science objectives could be addressed by a SIRTF program on galaxy clusters. For example, what is the nature of the intergalactic medium in clusters? Does it contain dust? What are its energetics? How do galaxies in clusters undergo cosmic evolution? Does the cluster environment noticeably change the time evolution of galaxy luminosity and if so, how? In particular, the increased potential for galaxy interactions and the possibility of stripping of interstellar gas by the intracluster medium both suggest that galaxy evolution may be strongly modified. This hypothesis appears to be supported by the relatively high incidence of S0 galaxies and of peculiar dwarfs in rich clusters, but further insights can be achieved by looking at far infrared luminosities, particularly for distant clusters where detailed morphologies become difficult to obtain, even with HST.
These questions are subject of active interest. Dwek et al. (1990) developed a model of the Coma cluster to explain observations of a modest level of extinction toward background galaxies, presumably due to intracluster dust. To accompany the perported extinction, they predicted relatively strong far infrared fluxes through heating of dust by the hot gas between galaxies; to avoid exceeding the IRAS detection limits, they had to place this dust in a previously undetected large halo around the visible cluster. If this model is correct, it would imply a previously undetected large mass of gas in the outer regions of the cluster, as they point out. Hu (1992) found that clusters with cooling flow galaxies were detectable in the IRAS data and concluded that the fluxes originated from the intracluster medium. Wise et al. (1993) tried to detect 56 clusters in the IRAS data, succeeded with only 4-5, and concluded that two of the detections were at levels above those expected from individual galaxies and hence could be due to an intracluster medium. Recently Stickel et al. (1998) have reported a detection of a surface brightness in the Coma cluster of 0.1 MJy/sr at 120microns, within about a factor of two of the prediction by Dwek et al. (1990). Stickel et al. assume that the theoretical prediction is confirmed by these ISOPHOT measurements. Thus, the possibility of large amounts of gas associated with the cluster appears to be confirmed.
However, we have just finished a paper (Quillen et al. 1998) which re- examines the far infrared emission by the member galaxies of the Coma cluster. We combine IRAS and new ISOCAM data with H alpha imaging and radio data from the literature. The infrared, H alpha, and radio estimates for the Coma galaxies give very consistent estimates of the overall star formation rate. Through standard ratios of H alpha and/or radio to far infrared, we have generated independent estimates of the integrated infrared luminosity of the cluster galaxies that agree to within a factor of two or better with the estimate from the infrared observations. The estimated far infrared surface brightness of the cluster due to its member galaxies we find to be more than half, and perhaps all, of the surface brightness measured by Stickel et al. with ISOPHOT. Thus, the emission of the diffuse intergalactic medium must be at least four times smaller than the theoretical prediction of Dwek et al.(1990). Their model could probably be adjusted to produce lower fluxes (with further increase in the previously unseen body of gas), so it is still unclear whether Coma has diffuse emission at an interesting level. Quillen et al. show that most other clusters may also have a significant contribution from their member galaxies, so to probe them we need observations of greater sensitivity and resolution than available from IRAS - and it does not appear that ISO will supply them.
This situation can easily be clarified by SIRTF measurements. The combination of high sensitivity and reasonably good angular resolution, plus the ability to map rich clusters efficiently at 24, 70, and 160 microns are the keys. MIPS scans over these clusters will unambiguously detect the individual emitting galaxies. The resolution of ~ 6" at 24 microns will identify which galaxies are emitting significantly, while the longer bands will constrain their spectral energy distributions. The good dc stability that should be achieved by MIPS, plus the high sensitivity, will allow detection of any diffuse intragalactic emission to low levels. For example, 0.1 mJy/sr corresponds to 2.5 mJy in a MIPS 160 micron beam and to 0.7 mJy in a 70 micron beam. Although these values are near the confusion limit at the latter wavelength, they will be strongly detected above photon noise in both bands (1 sigma in about 1 minute) and by cleaning the scans of discrete galaxies and smoothing, a diffuse component should become easily visible.
Another view of the far infrared emission of rich clusters was taken by Kelly & Rieke (1990). They assumed that the cluster far infrared emission was entirely due to the sum of the outputs of the member galaxies. By summing the IRAS data on ~70 optically selected high redshift clusters, they achieved a high weight detection of a "typical" rich cluster at an average z = 0.4. They compared this detection with nearby clusters and hence estimated the integrated evolution of the cluster far infrared outputs. For the nearby clusters, they estimated a luminosity function from sources then available. For each cluster with high sensitivity IRAS data, they normalized the luminosity function to the IRAS detections of the brightest galaxies, then integrated down to low luminosities to get the total for the cluster. The result was an average luminosity for nearby clusters that was similar to the average for the clusters at z = 0.4, and hence contrary to expectations of possibly faster-than-"normal" evolution for cluster galaxies due to accelerated rates of interactions and mergers, for example. In Quillen et al. (1998), we show that the low end of the cluster galaxy luminosity function is probably incorrectly estimated in the references used by Kelly & Rieke, and hence that they overestimated the luminosities of the nearby clusters by a factor of ~ 3. Using a luminosity function with slope similar to the field galaxy function, and also to the optical luminosity function in rich clusters, it is shown that the nearby clusters are likely to be significantly fainter than the ones at z = 0.4. The nominal luminosity evolution over this range of z is (1+z)^(~3.9), but large uncertainties remain in determining the nearby cluster luminosities and hence in this rate of evolution.
It is ironic that we seem to know more about the luminosity of clusters at z = 0.4 than we do about ones at z < 0.05. The problem with the nearby clusters is that IRAS had too limited angular resolution and sensitivity to make a good census of their member galaxies. In terms of integrated cluster measurements, the clusters are so large on the sky that the IRAS confusion noise overwhelms the cluster signal. This situation is made clear by the multi-aperture extractions of Wise et al. (1993), who found that the cluster detectability decreased rapidly with increasing aperture size, even where the aperture was still within the Abell diameter.
The same MIPS scans that would be used to search for diffuse cluster emission would be well suited to making a nearly full census of the infrared emission from the member galaxies also. For example, the detection limits discussed above for Coma would reach to galaxy luminosities of about 10^8 L_o, well below L* and hence capable of defining the luminosity function well and constraining the galaxian luminosity from the cluster accurately. Assuming that there are no serious complications from emission by an intracluster medium at a higher level than the sum of the galaxy contributions, these measurements would allow determination of overall cluster luminosity evolution from z = 0.4, just by comparing with IRAS. Additional measurements of clusters at higher redshift could extend evolutionary studies back to earlier epochs, to an era where strong evolution is apparent in field galaxies and hence a meaningful comparison can be made with the cluster evolution. The cluster-averaged flux at z = 0.4 is about 30 mJy at 60 microns, so the MIPS sensitivity would easily allow extension of cluster studies to much greater redshifts.
In summary, a relatively modest MIPS program on nearby and distant galaxy clusters could provide unique information both on the presence of a substantial intracluster medium, and on the possibly different pattern of galaxy evolution compared with the field.
Dwek, E., Rephaeli, Y., & Mathur, J. 1990, ApJ, 475, 565Introduction and Background:
Understanding the origin of the solar system has long been one of the fundamental objectives of astrophysical research. In the last few years two major breakthroughs have convinced many researchers that significant progress toward the realization of this goal is now possible. First, extrasolar planets have been indirectly detected toward a number of nearby stellar systems including 51 Pegasi, 70 Virginis, and 47 Ursa Majoris (Mayor & Queloz 1995, Marcy and Butler 1996, etc.). This has provided truly compelling evidence that the sun is not unique in having planetary companions. How often planetary systems accompany stars in the galaxy is difficult to ascertain with present knowledge. However, by analogy with the solar system, it is thought that such planets would likely have formed from a spatially thin circumstellar disk of material. Theoretical considerations suggest such disks should accompany a star during its formation as a result of angular momentum conservation in collapsing gas (e.g., Adams, Lada and Shu 1987). The second major breakthrough in this area of research, the discovery of circumstellar disks around numerous newly formed and forming stars, has suggested that the physical circumstances that lead to planet formation may be a common occurrence and, indeed, a natural consequence of the star formation process (e.g., Beall 1987, Rucinski 1987, Beckwith et al. 1990, Lada and Adams 1992).
Circumstellar disks can have many observational manifestations, both direct and indirect, the most compelling being resolved images of disks. Such images are extremely rare but have been obtained at submillimeter-wavelengths through VLBI techniques and at optical wavelengths in the one region, the Orion Nebula, where the HST is sensitive enough to image such objects in silhouette against the bright background of the HII region (McCaughrean & O'Dell 1996). Despite the fact that it is currently very difficult to resolve them, circumstellar disks can nonetheless be quite easily detected. Because they contain substantial amounts of dust they readily emit and reflect light at infrared through millimeter wavelengths. Consider that if the earth were to be crushed into particles the sizes and masses of interstellar grains, the surface area of the material would increase by 14 orders of magnitude. Thus, a protoplanetary disk is much easier to detect than the planetary system that ultimately forms from it. Current detectors are sensitive to very small amounts of dust, and circumstellar disks with total masses comparable to the minimum mass needed to produce all the planets of the solar system can be detected at infrared and millimeter wavelengths around stars in the closest star forming regions. Consequently, infrared surveys are the best tool for determining the frequency of circumstellar disks and ascertaining how common are the conditions for planet formation.
Figure 1 shows the energy distributions expected for a three solar luminosity K5 pre-main sequence star at the distance of Taurus (or Ophiuchus) with and without an optically thick circumstellar disk. For wavelengths longer than about 1 micron the source energy distribution of the star with the disk exceeds that of the naked star. In this model almost all the excess infrared emission is starlight absorbed and reprocessed by the disk. The infrared excess systematically increases with wavelength, along with the ability to recognize the signature of disk emission around the star. For example, as is illustrated in the diagram, the excess emission produced by the disk at a wavelength of 30 microns is more than 10x that produced at 2 microns. Whereas such factors as disk orientation, stellar surface temperature, and inner disk holes make detection of the disk excess emission difficult at wavelengths shorter than about 3 microns, at longer wavelengths the presence of disk excess is essentially unambiguous. Indeed, ground-based observations have shown that only about 1/2 the stars with circumstellar disks produce a significant enough infrared excess to be detected at 2.2 microns. Moreover, around the youngest stars, starlight scattered by residual protostellar envelopes can also produce infrared excess at wavelengths shorter than 2 microns, making the association of near-infrared excess with the presence of a circumstellar disk more ambiguous. An accurate census of circumstellar disks in star forming regions requires sensitive observations at mid and far infrared wavelengths. Such observations are difficult to obtain from the ground.
SIRTF offers an unparalleled capability to detect circumstellar disks in star forming regions. For example, Figure 1 displays the SIRTF sensitivity at 8 and 30 microns. The two horizontal lines represent the 50 sigma detection limits at these two wavelengths. Equivalently, SIRTF is capable of detecting the naked photosphere of a 3 million year old, pre-main sequence star in Taurus which has a mass near that corresponding to the hydrogen burning limit. In fact, SIRTF is capable of detecting the photospheres of such young stars across the entire spectrum of stellar mass in star forming regions which are as distant as 1 kpc from the sun! Clearly then, SIRTF can detect both optically thin as well as optically thick disks around young stars over a relatively large volume of space.
Proposed Scientific Program:
We propose to use SIRTF to obtain an extensive census of protoplanetary disks in young star clusters and selected star forming regions. The goal of this survey will be to determine the frequency distributions of these disks as functions of age, stellar mass and content in young stellar clusters and to use this information to determine the lifetimes of circumstellar disks and thus directly constrain the duration of planet building activity and the likelihood of planet formation around young stars.
As an illustration of how we will use such observations to constrain disk lifetimes and evolution, we show in Figure 2 some related results obtained from ground-based 2.2 micron surveys of a few nearby clusters. Here we plot the fraction of near-infrared excess sources in the cluster vs the cluster age. One immediately observes that the likelihood of finding infrared excess sources in a cluster is dependent on the cluster age. At about 6 million years it appears that stars with near-infrared excess are no longer present in young clusters. This appears to place a global limit on the lifetimes of the inner disks for stars formed in young clusters and thus places a direct constraint on the timescale allowed for planet building in the inner regions of circumstellar disks.
However, this constraint is strictly an upper limit because of the fact that the duration of star formation is likely comparable to the ages of these clusters. For example, if the duration of star formation is 5 million years in each cluster and stars form at the same rate over this period, the results in Figure 2 would also be consistent with a disk (or more accurately infrared-excess) lifetime of only 1-2 million years. Because the duration and history of star formation may vary from one cluster to another and the duration of the disk phase may vary between individual stars within a cluster, a more detailed analysis of such data is required to derive any more meaningful constraints on the evolution of protoplanetary disks. However, because of the ambiguities in using 2.2 micron data to trace disk emission, existing data like that in Figure 2 cannot be used for this purpose. It is first necessary to obtain observations of clusters at longer wavelengths which can more accurately and completely trace their disk populations.
We propose to use SIRTF to image ~ 20 young clusters spanning ages between 1-70 million years to determine the lifetimes and investigate the evolution of protoplanetary disks. SIRTF will provide us with such complete inventories of circumstellar disk populations in clusters that we will be able to explore aspects of disk evolution not possible with any other type of observation. For example, we will be able to determine how disk lifetime varies with stellar mass. Do lower mass stars have longer lived disks than higher mass stars? If so what are the corresponding implications for the production of planets as a function of stellar mass (e.g., is it more difficult to form planets around higher mass stars?). With such sensitive observations of a large sample of clusters we will also be able to explore the effect of stellar environment on disk lifetimes. For example, does the presence of O stars in a cluster significantly abbreviate disk lifetimes due to photoevaporation, as observations of the Trapezium cluster now seem to suggest? Because good color-magnitude diagrams exist for many of the clusters we plan to observe (e.g., NGC 2362, NGC 2264, IC348, etc) we can combine these with pre-main sequence tracks to investigate the evolution of disks as a function of both age and mass within individual clusters as well as across clusters of differing age. Moreover, because of the great increase in sensitivity afforded by SIRTF we anticipate being able to follow the evolution of disks to much later ages than shown in Figure 2. Evidently after only a few million years (or less) disks evolve to the point where they become too optically thin to be detected at near-IR wavelengths from the ground. But these optically thin disks may remain detectable by SIRTF for long periods of time, perhaps up to 50-100 million years or more. We may be able to follow the evolution of disks into a new regieme, that of post planet formation when the disk begins its evolution toward a Beta Pic like object. Thus, we anticipate that SIRTF will fundamentally alter our understanding of the evolution of circumstellar disks and in doing so place strong constraints on the question of the ubiquity of planetary systems in the galaxy.
It is important for us to obtain multi-wavelength images of the clusters we select for study in order to define the SEDs of the candidate disks. This will be particularly important for measuring the infrared excess produced by evolved, optically thin disks and more distant sources. So we expect to utilize both MIPS and IRAC for this project. In particular, observations with IRAC will be necessary for some cluster cores where crowding may cause confusion even at 30 microns. We will typically need to survey a region about 1/2 degree square toward each cluster. Observations of control fields will be needed in some cases as well. Our sample of clusters will include all the nearby, well studied young clusters as well as a few more distant clusters selected on the basis of age, stellar content, and richness.
In preparation for our proposed SIRTF program we plan to obtain observations at 10 and 20 microns with the WIRE satellite. Erick Young and Charles Lada are co-investigators on an approved WIRE AI program (PI: Elizabeth Lada, University of Florida) to perform large scale surveys of nearby GMCs and young clusters. Our WIRE program was allocated a significant amount of priority 1 orbit segments. We will take advantage of the wide-field capability of WIRE to obtain complete inventories of circumstellar disks and young stellar objects within the large extended regions of nearby GMCs, something that would be too time consuming for SIRTF to do. This, in combination with our proposed SIRTF observations will enable a comparison of disk properties for stars formed in and out of clusters. In addition, WIRE observations will be obtained for as many young clusters as possible. The WIRE data will thus be used to help select the clusters for inclusion in our SIRTF survey. For the most part the WIRE observations of clusters will be severly limited by source confusion due to the instrument's relatively poor angular resolution. Nonetheless, the WIRE observations we obtain for embedded clusters will play a critical role in our detailed planning of the SIRTF survey.
References:
Adams, F. C., Lada, C. J. & Shu F. H. 1987 ApJ, 321, 788