Enough questionswhat do we know and how do we know it

The disk dispersal time or disk lifetime is often asserted in the literature as "about 10 Myr." This estimate is certainly good to an order of magnitude, but the justification for this number (or any other specific number) is weak at best, given the data in hand. Some inner dust/gas disks have disappeared within 1 Myr—by the time the star becomes optically visible. Some inner dust/gas disks last at least 10 Myr. In at least one case, that of our own solar system, some theorists have expressed the need for the gas disk to

Property Disk geometry Mean excess in SED;

disk fraction Accretion rate on to star

Dust mass

Dust mineralogy; size distribution

Observational Diagnostic Interferometry; SED modeling Broad-band photometry

Ultraviolet/optical spectrophotometry

Millimeter/sub-millimeter photometry

Mid-infrared spectroscopy

Example Study Eisner et al. 2004, 2005b Hillenbrand et al. 2006;

Mamajek et al. 2004 Muzerolle et al. 2000;

White & Hillenbrand 2004 Carpenter et al. 2005;

Wyatt et al. 2003 Kessler-Silacci et al. 2005; van Boekel et al. 2005

Table 1. Dust disk properties measurable as a function of stellar age survive 100 Myr or longer in order to form the outermost gas giants. As astronomers, we want to understand the mean and the dispersal time of young primordial dust/gas disks. In this section, I will review disk diagnostics, appropriate subject samples, and the difficulties involved in assessing stellar ages. In the next section, I will summarize what is known about disk evolutionary trends.

4.1. Disk diagnostics

To look for evidence of disk evolution in action, we need to consider carefully the diagnostic potential of any particular observable parameter. Many are available, however, the information obtained varies widely between different tracers of disk evolution. This is due, in part, to the variation in observational sensitivity (for example, as a function of wavelength), and in part to the varying efficacy of different disk tracers. In addition, the precision and accuracy of stellar ages—that other, often under-scrutinized (or even ignored) axis in any disk evolution diagram—needs to be critically assessed.

Deferring sensitivity considerations for the time being, what can we hope to measure as a function of stellar age? Resolved disk images, as discussed in my introductory comments, certainly have led to a wider appreciation of the convincing case for primordial "proto-planetary" disks. In fact, it is not an over-statement to say that the stunning images from ground-based interferometers (millimeter) and from the Hubble Space Telescope (optical and near-infrared) were responsible for transforming the field of star formation from a following of dedicated and knowledgeable disciples to high profile science. However, the reality is that few such spatially resolved images exist at present. The study of most young disk systems relies, for the most part, on so-called indirect measurements, such as broadband photometry and high resolution optical or near-infrared spectroscopy.

Table 1 details several properties of quantitative interest for young circumstellar disks and the observational diagnostics used to measure them. These are generalized properties and each can be broken down into a more detailed set of specific physical characteristics. There is an increasingly large literature on these topics, and I list only a few example studies. In most categories there is some limited evidence for at least modest evolution from primordial disk conditions. Conclusions in the area of evolution are typically based on samples of young disks ranging from small (a few) to moderate (tens to a few hundred) in size.

As mentioned above, the focus of my discussion will be on dust-disk diagnostics. In particular, I will focus on disk detection as revealed through infrared excesses—observed emission in excess of that expected from a stellar photosphere. Various levels of sophistication may be employed in the application of this technique, ranging from fully assembled spectral energy distributions covering several decades in wavelength, to two-

color diagrams which cover only a limited portion of the excess spectrum, to statistical study of the disk fraction (frequency of objects in a given age bin with convincing evidence of a disk) or mean excess (magnitude or strength of the excess). Full spectral energy distributions covering ultraviolet to millimeter wavelengths have been available for only small samples of well-studied young stellar objects, making statistics difficult to assemble. Two-color diagrams are widely available, enabling statistical studies, but are more difficult to interpret without detailed knowledge of: 1) more of the spectral energy distribution, 2) the intrinsic spectral energy distribution in the absence of reddening, which can be prevalent towards young star-forming regions, and 3) the properties of the underlying star. Disk fraction and mean excess techniques account for both reddening and intrinsic stellar colors, but are based on partial spectral energy distributions. The discussion below will focus on these last techniques.

Due to several decades of ground-based work, combined with the large and uniform 2MASS photometric database at 1.2, 1.6, 2.2 ¡m now available, data for infrared excess investigations are abundantly available at near-infrared wavelengths. Ground-based work at mid-infrared wavelengths has been more limited in both scope and sensitivity; previous space-based platforms were revolutionary at the time, but somewhat similarly limited in sensitivity (IRAS) and scope (ISO). Thus, our understanding of disk statistics in the 3100 ¡m wavelength regime is not as well developed as in the 1-2 ¡m regime. The Spitzer Space Telescope is currently accumulating sensitive data between 3-70 ¡m, enabling the construction of mid-infrared spectral energy distributions. These observations focus on many of the historically favored objects, though over the next few years, blind-imaging surveys of star-forming regions and young open clusters are also being conducted and will provide needed statistics.

4.2. Stellar samples

Once a technique is adopted, a sample must be chosen. In order to establish trends, robust, complete, and unbiased samples must be established over an appropriate range of ages. For the problem at hand, this includes the youngest revealed protostars through the ages characteristic of star-forming regions still associated with molecular gas (<12 Myr), and continuing to the entire period of terrestrial and gas giant planet formation (thought to be ^100 Myr for our own solar system), as depicted schematically in Figure 1.

Young star clusters would appear ideal for these sorts of studies because they provide the needed statistics. Furthermore, clusters have attractive attributes such as the relatively uniform distance, age, and chemical composition of their members—all of which minimize analysis complication. Young star clusters can therefore, in principle, provide the samples required to compare disk properties such as the mean and dispersion in disk lifetimes as a function of stellar mass (within a cluster), and as a function of stellar age or chemical composition (between clusters). However, careful investigation reveals that the young star samples identified to date are lacking with respect to some important issues.

First, known targets for investigations of disk evolution can be segregated into the following four coarse age groups: <1 Myr (embedded or partially embedded star forming regions), 1-3 Myr (optically revealed stellar populations still associated with molecular gas), 5-15 Myr (association members in gas-poor "fossil" star-forming regions), and finally the punctuated ages (55 Myr, 90 Myr, 120 Myr, and 600 Myr) of the nearest populous open clusters.

Of concern is that the age distribution is not uniform for known samples of young stars over the 1-100 Myr age range for disk evolution. Ample numbers (many thousands) of young stars associated with regions of recent star formation have been identified through surveys of molecular cloud complexes. Because of the intense focus on stellar census data for these young regions, such stars dominate the total numbers, and thus bias the available statistics of the overall young star age distribution towards the 1-3 Myr youngest age category.

The statistics decline dramatically at ages older than about 5 Myr and out to about 50 Myr, due to a lack of large identified samples with known ages in this "young-intermediate" age regime. There are no open clusters or large associations in the 5-50 Myr age range within 150 pc or so of the Sun, save for the Sco OB-2 association at the upper distance and lower age limit. Field stars 5-50 Myr old are extremely hard to identify, since it is only with detailed observations (not, e.g., in wide-field photometric surveys) that they stand out from much older field star populations. They may be revealed through signatures of youth, such as common proper motion with kinematically young groups, enhanced Li I absorption, Ca II H and K core emission, and x-ray activity. In fact, due to our location in the Galaxy near a ring of moderately recent star formation ("Gould's Belt"), finding stars in this age range should be relatively easy. Yet within 150 pc or so, current samples of 5-50 Myr-old stars only number a few tens, consisting of members of the TW Hya, Beta Pic, Eta Cha, and Tuc/Hor moving groups. Continued correlation of large-scale kinematic and activity databases with sufficient spectroscopic following is beginning to address this deficiency. However, the present lack of ample numbers of young stars in the 5-50 Myr age range serves to increase the error bars in disk-evolution diagnostic statistics right where the most interesting "action" of disk evolution may be taking place.

At ages older than 50 Myr, there are again ample samples due to the proximity of several nearby open clusters. Specifically, the IC2602/IC2391 pair, Alpha Per, Pleiades, and Hyades clusters, all within 200 pc and well-studied, are benchmark points in any evolutionary diagram involving either stellar or circumstellar properties.

4.3. Stellar ages

A comprehensive discussion of stellar ages is beyond the scope of this review. Suffice it to say that there are large number of age diagnostics, most of which are poorly calibrated in the young pre-main sequence age range of interest here. The most commonly used measure of stellar age in the <1-30 Myr age range is location in the Hertzsprung-Russell (HR) diagram compared to theoretical predictions of luminosity and temperature evolution as a function of time. HR diagrams (HRDs) are shown in Figure 2 for a number of current and recently star-forming regions, as well as young open clusters in the solar neighborhood.

HR diagrams can be used to infer a mean age and an apparent age dispersion as a function of effective temperature for each cluster. One issue to consider is whether the age spreads inferred for stars in young star clusters from their observed luminosity spreads indeed correspond to age ranges rather than observational errors, the default assumption. Luminosity spreads do decrease with time (e.g., consider the Orion Nebula Cluster versus the Alpha Per Cluster in Fig. 2). However, the errors in converting from observables to luminosity are the largest in the young pre-main sequence phase, just where the apparent luminosity spreads are the largest. Thus the conversion from luminosity to age—and the implied age spreads—are confusing. Understanding the age spreads, or the lack thereof, are important for the purposes of studying evolutionary diagrams. One needs either to consider all stars in a single cluster to have the mean age of the apparent distribution, or to individually assess ages and adopt an age for each star. This issue has not been satisfactorily addressed at the young ages of interest here, thus the evolutionary timescales for young circumstellar disks have large random uncertainties, depending on whether potentially real age spreads are accounted for in the analysis or not.

Figure 2. HRDs for well-studied star-forming regions and young clusters. Data were placed using the temperature scale, color scale, and bolometric corrections described in Hillenbrand & White (2004) and a wide variety of literature for the fundamental data. Pre-main sequence evolutionary calculations are those of D'Antona & Mazzitelli (1997, 1998) for isochrones of 0.1, 1.0, 10, and 100 Myr (solid lines) and masses 0.03, 0.06, 0.08, 0.1, 0.2, 0.4, 0.8, 1.5, and 3.0 Mq (dashed lines).

log Teff/K

Figure 2. HRDs for well-studied star-forming regions and young clusters. Data were placed using the temperature scale, color scale, and bolometric corrections described in Hillenbrand & White (2004) and a wide variety of literature for the fundamental data. Pre-main sequence evolutionary calculations are those of D'Antona & Mazzitelli (1997, 1998) for isochrones of 0.1, 1.0, 10, and 100 Myr (solid lines) and masses 0.03, 0.06, 0.08, 0.1, 0.2, 0.4, 0.8, 1.5, and 3.0 Mq (dashed lines).

Another caution is that theoretical pre-main sequence evolutionary calculations, upon which age estimates from the HR diagram rely, have significant uncertainties in their predictions. First, there is variation between various theory groups of 20-100% over certain mass and age ranges (see comprehensive discussion in Baraffe et al. 2002). Second, pre-main sequence calculations thus far do not favor well in comparison to observational constraints. Specifically, they collectively under-predict stellar masses by 30-50% (Hillenbrand & White 2004). Further, they under-predict low-mass stellar ages by 30-100% compared to lithium-depletion boundary estimates and over-predict high-mass stellar ages by 20-100% compared to post-main sequence evolutionary calculations. Because of this lack of theoretical validation of the age calibration of pre-main sequence isochrones, the evolutionary timescales for young circumstellar disks have large systematic uncertainties.

Was this article helpful?

0 0

Post a comment