Zodiacal Dust as a Constituent of Planetary Systems

The presence of dust around a mature, main sequence star is a reminder that the star was born by accretion through in a dense protostellar disk of gas and dust and that any planets that the star possesses were born out of that same material. The dust around a mature star is not itself primordial, since radiative forces will remove small dust grains in a few thousands to a few millions of years. However, the parent bodies of the observed small dust grains - large numbers of either asteroids or comets - are the remnants of primordial material and the debris with them provides hints to the origin of planetary systems. Additionally, this solid material is important in the evolution of life, since the cold outer reaches of a planetary system are rich in water and volatiles needed for the creation of habitable environments in the inner reaches of the system. Thus, we want to study the exozodiacal clouds of nearby stars for their intrinsic scientific importance using a variety of observational facilities, including TPF-I/Darwin and TPF-C. Yet the brightness of the exozodiacal cloud is also a potential problem for the detection of terrestrial planets due to increased photon noise and confusion between planets and structures in the zodiacal cloud.

Thus, the incidence, distribution and composition of material in the habitable zones of nearby stars is important for the design of TPF-I or TPF-C and for the feasibility of studying individual targets. After a brief discussion about the impact of photon noise from exozodiacal emission on both visible light and mid-IR instruments, we summarize recent observational results, discuss theoretical expectations for the level of exozodiacal emission, and describe future observational programs relevant to determining the level of exozodiacal emission around typical TPF targets. We will show that TPF-I can operate in the presence of zodiacal clouds 10-20 times as bright as our own without undue difficulty and that systems with this amount of emission, or considerably less, are likely to be common among mature solar type stars.

Zodiacal Emission as a Source of Noise

For a cryogenic nulling interferometer operating in an orbit near 1 AU the three dominant noise sources are: (1) stellar light that leaks past the interferometric null because of the finite diameter of the star; (2) emission from the local zodiacal dust; and (3) emission from the exozodiacal dust in the target system that leaks past the interferometer. At short wavelengths (< 8 µm), the stellar leak may dominate all other noise sources while at wavelengths longward of 20 m, emission from the interferometer itself may become important. But over a broad range of wavelengths, the balance between the three noise sources controls the photon-noise-limited noise floor. In the infrared case, the brightness of the planet itself is small (<1%) compared with the backgrounds and can be ignored. Similarly, detector read noise and dark current can be ignored for broad band detection.

Figure 1. Upper limits (or detections) to zodiacal emission.

For a particular temperature, smaller grains are located further from their parent star than the cooler, larger (blackbody) grains and thus emission from small grains is less effectively blocked by the central null of interferometer. Small grains thus produce more noise than large grains for a given total exozodiacal brightness. When the surface density of the exozodiacal dust is 10 times that of our solar system, corresponding to a 20-fold brightness increase, the S/N is reduced by a factor of ~2-3, necessitating an increase in integration time by a factor of ~4-9 to recover the original S/N. Since many hours of integration time are needed to detect an Earth in the presence of zodiacal cloud similar to our own, and days to carry out a spectroscopic program, it is clear that studying systems with 10-20 times the surface density will be difficult.

A more detailed examination of the effects of the exozodiacal emission on the detectability of planets would include inclination effects as well as the possibly confusing effects of structures within the zodiacal cloud. It is already well known, for example, that observing an inclined exozodiacal disk though a simple Bracewell interferometer produces a nulled output signal that mimics that of a planet (Angel and Woolf 1997). This is one of main drivers for more complex modulation schemes such as the Oases (Angel and Woolf 1997), Dual Chopped Bracewell and X-array designs (Lay 2005), and composite designs (Velusamy, Beichman, and Shao 1999). As these studies have shown, with sufficient angular resolution and uv-plane complexity, structures such as wakes and gaps can be distinguished from the signatures of planets. This necessity, along with the need to distinguish multiple planets in a system, is one of the key drivers for observing with TPF-I using a variety of baselines up to at least 100 m. Good angular resolution and diversity in uv-plane coverage are an important aspect of properly characterizing planetary systems.

A simple order of magnitude estimate of the brightness of structures in the zodiacal cloud shows that this confusion will not be a problem in clouds like our own, but could become one in brighter clouds. Additional modeling and observation of zodiacal clouds in the range of 1-100 times the brightness of our cloud will be necessary to assess the importance of this effect.

Summary of Current Observational Results on Exozodiacal Disks

Results from the Spitzer Space Telescope have greatly advanced our understanding of the incidence of exozodiacal clouds as a function of age, spectral type, and metallicity. At 70 µm Spitzer is sensitive to levels of exozodiacal emission from ~35-75 K dust with roughly 5-10 times the expected level of emission of our own Kuiper Belt. A wide variety of Spitzer programs have found the following characteristics of exozodiacal emission:

• Approximately 14±3% of mature, solar type stars (F5-G5) have detectable 70 µm zodiacal emission (Bryden et al 2006). This rate is somewhat higher among A and early F stars (~25%) and smaller for stars later than K (<4%) (Su et al 2005; Beichman et al 2006b).

• Emission at 10 µm, corresponding to dust in the habitable zone and thus most relevant to TPF, is rare at the Spitzer sensitivity level at mature stars. However, the unfavorable contrast ratio at this wavelength means that Spitzer can detect emission only at a level ~1,000 times brighter than our own zodiacal cloud (Figure 1). Initial estimates based on IRAS (Mannings and Barlow 1998; Fajardo-Acosta, Beichman and Cutri 2000) and ISO (Laureijs et al 2002) observations were that <2% of systems have detectable disks at 10 µm while the largest Spitzer sample studied to date of 150 stars suggests a rate of less than ~1% (Beichman et al 2006b; D. Ciardi, private communication). While a few individual objects, including A stars like beta Pic and beta Leo and the 2-4 Gyr old K0 star HD 69830 (Beichman et al 2005b), show bright emission from small grains in the habitable zone, stars like this are very rare.

• Zodiacal emission is both more intense and more frequent (up to 30% at 24 µm) at ages less than ~150 Myr (Rieke et al 2005; Siegler et al 2006) but at stellar ages greater than about 1 Gyr the incidence of exozodiacal emission shows little dependence on age.

• Despite the detection of exozodiacal emission in association with planet-bearing stars (Beichman et al 2005a), there is little dependence on the incidence of exozodiacal emission on metallicity despite the clear dependence on the incidence of planetary systems on metallicity (Beichman et al 2006b).

• Relatively featureless spectra are seen toward most disks using the Spitzer/IRS spectrometer (Jura et al 2004; Chen et al 2006; Beichman 2006a) with a few dramatic exceptions (Beichman et al 2005). Disk emission becomes detectable around 25-30 µm in Spitzer/MIPS 24 µm photometry and Spitzer/IRS spectra implying temperatures of 75 K and an inner edge to the emitting region around 7-14 AU.

Theoretical Implications of Current Observations

While it is true that we cannot yet observe zodiacal disks at the level of our own cloud, particularly in the habitable zone, the existing observations begin to rule out some disk "luminosity functions" that are much brighter than our own. Observations seem to rule out a log-normal distribution of zodiacal brightness centered on a level that is 100 brighter than our solar system level. Future observations will be needed to constrain similar log-normal distributions centered on levels of emission more similar to our own system's.

Theory based on the evolution of debris disks as revealed by ISO (Dominik and Decin 2003) and Spitzer (Wyatt et al 2006) suggest that the predicted level of zodiacal emission drops well below 10 times that of our system by the time the star reaches a few Gyr. While our theoretical understanding is far from complete, present and future observations of disks, should give us confidence that the expected level of emission will be at or below the desired ~10-20 exozodiacal level needed for the detection of terrestrial planets around many nearby stars.

Prospects for Future Observations

It will take observations with facilities other than Spitzer to push to lower levels of zodiacal emission. The Herschel telescope will measure cold Kuiper Belt disks to solar system levels while ground based interferometers such as the Keck Interferometer (KI) and Large Binocular Telescope Interferometers (LBTI) will spatially suppress the stellar component to measure definitively the 10 µm exozodiacal emission that arises in the habitable zone and that might cause problems for TPF.

The Keck Interferometer (KI; Colavita et al 2006) is currently implementing a nulling interferometry mode at 10 µm specifically targeted at observations of exozodiacal emission around nearby main sequence stars. In this mode, the central star is placed on a destructive fringe, allowing detection of the much fainter surrounding emission while rejecting intense photospheric emission. The size scales probed by 85-meter baseline in this mode are 25 to 200 mas, corresponding to the habitable zone for many nearby main sequence stars. Initial observations using this mode have been made and the final detection level is expected to be 100 times the level of our solar system with a sensitivity limit of 2 Jy for the target star. At this sensitivity limit and within the declination range of the telescope, there are 53 main sequence stars with A through K spectral types which can be observed with KI, including: 10 A stars, 18 F stars, 10 G stars, and 6 K stars. The KI observations will be sensitive to dust in the habitable zone at a factor of 10 lower levels than the Spitzer observations. The sample of available stars will determine the frequency of disks at the 100 solar system level and will test the theoretically predicted spatial distributions. In a few years after implementation of the KI nuller, the Large Binocular Telescope Interferometer (LBTI; Hinz et al 2003) will become operational. LBTI also works at 10 µm and eliminates the stellar flux through interferometric combination, but its unique design of co-mounted telescopes on a 15-meter baseline gives it a large field-of-view with ~100 mas resolution. LBTI is expected to push to sensitivity levels a factor of 10 below KI's and complement KI by looking at material at ~2x greater distances from the star.

With Spitzer and Herschel measuring excesses from 200 to 20 µm down to near solar system levels, and with KI and LBTI pushing to near solar system levels at 10 µm, we can confidently expect to understand the statistical properties of TPF targets well enough to know whether or not there exists, as theory suggests, a population of stars with low levels of zodiacal emission. We will also have detailed information on many individual targets, particularly at northern declinations accessible to KI and LBTI.