EXTRASOLAR ZODIACAL EMISSION - NASA PANEL REPORT Panel Members Dana Backman F&M College, chair Paul Kalas MPIA, Heidelberg Chas Beichman Caltech-IPAC Hal Levison South-West Research Inst. Pierre Bély STScI Bill Reach Caltech-IPAC Chris Burrows STScI George Rieke Univ. of Arizona Mark Colavita NASA-JPL Dave Sandler Thermotrex Inc. Stan Dermott Univ. of Florida Harley Thronson NASA-HQ Christ Ftaclas Michigan Tech. Univ. Wes Traub Harvard-CfA I. Executive Summary The zodiacal dust cloud is the most prominent sign of the presence of planetesimal-sized objects in our solar system. Although its total mass is relatively tiny, the thermal emission and scattered light from a portion of the zodiacal cloud a few x 0.1 AU across is comparable to that from a terrestrial planet. Our best knowledge of the zodiacal cloud is of the material near Earth's orbit, at heliocentric radii from about 0.9 to 2 AU. The detailed structure of the cloud is only now being explored. The relative contributions of various dust creation and destruction mechanisms are not yet well understood. What can be said at present is that most of the dust near 1 AU is from main belt asteroid collisions and erosion, comets are an important secondary source, and Poynting- Robertson (PR) radiation drag inducing drift toward the Sun is the most important local removal mechanism. Simple back-of-the-envelope arguments presented below indicate that the observed zodiacal dust density is probably about the norm (within a factor of ~2) for the past Gyr. This is quite uncertain, however. Individual asteroid collisions, giant comets, and comet showers may produce large enhancements lasting 107 yr or more. We cannot easily derive the density of the local zodiacal cloud in the past because the main process (aside from collisions) that removes the dust's planetesimal parent bodies is the "chaotic" influence of planetary perturbations. Early in the history of the solar system the zodiacal cloud could have been 106 times as dense and bright as at present, based on an estimate of the maximum plausible original mass in the asteroid belt. Models of the construction of a planetary system indicate, however, that the presence of an asteroid belt is not guaranteed. Thus, there is apparently a large possible range in mass of original asteroid populations in planetary systems otherwise resembling ours, along with an unpredictable rate of removal of asteroids and comets and the likelihood of occasional substantial transient increases in dust injection rate. The above concerns lead to the main conclusion of this report: it is impossible to predict the density of dust at terrestrial temperatures in extrasolar planetary systems, even assuming the arrangement and characteristics of major planets resemble those in our system. The necessary presence of comets might imply a very uncertain general lower limit of 1/10 the solar zodiacal dust density, and IRAS measurements indicate few cases among nearby main sequence stars of more than several x 100-zodi clouds of warm dust. Thus, our solar system's zodiacal cloud, which would pose some problem for external detection of the Earth, is probably near the low end of the possible range of system densities although the distribution within that range is not even guessable at this point. It can be stated safely that the average zodiacal density should steadily decrease over Gyr time scales in each system due to a decrease in the mass of dust parent body populations. This implies that planet-search instruments should be directed first at older systems. A planet-search interferometer should have precursor studies to identify those systems with the least dust emission. Dust emission at 1- to 100-zodi levels is not easily detected by unresolved photometry because the corresponding fractional dust excess over stellar photospheres would be only 10-4 to 10-2, smaller than the uncertainty in stellar mid-IR spectral energy distributions. Thus, high spatial resolution is necessary to separate exozodiacal emission from stellar brightness. The optical and IR brightness contrasts between planets and dust are comparable because the albedo and temperature of the dust are like those of planets. Finding planets and finding exozodiacal clouds will be of similar difficulty for spatial resolutions of order 0.1 AU and dust densities like those in the solar system's zodiacal cloud. Possible instruments to carry out the necessary preliminary exozodiacal dust surveys include the Keck, Magellan, LBT, and VLT interferometers working on the ground in the thermal-IR, and HST plus coronagraph in the visual and near-IR. NGST at visual and near-IR wavelengths with coronagraph and "slow" adaptive optics to compensate for its modest optical smoothness could also be sensitive to exozodiacal clouds. Ground-based IR interferometers can probably reach the 100- or 10-zodi level for systems within 10 pc using simple and already-understood Bracewell (chromatic) interferometry. Levels down to 10- to 1-zodi appear feasible with implementation of achromatic interferometry in a dual-nuller configuration. A space-based exozodiacal mapper would be a useful intermediate step between exozodiacal detection from the ground and planet detection from space. Maps of exozodiacal emission could exploit expected dynamical effects of extrasolar planets on their surrounding dust clouds to infer the presence, orbit locations and orientation, and even masses of the planets. II. Introduction It takes very little mass in the form of small grains in a circumstellar environment to produce substantial scattered light and thermal infrared emission. The surface brightness of the solar system's zodiacal cloud, with a total mass equal only to that of a single small asteroid, would make it difficult to detect Earth directly from neighboring stars at both optical and IR wavelengths. On the other hand, the presence of small dust grains with lifetimes much shorter than the age of the solar system would instantly indicate to an external observer that at least planetesimal-sized bodies exist in our system as reservoirs for the dust material. And if planetesimals, then why not planets? Furthermore, large masses embedded in the dust make their presence known through their gravitational effects on the cloud's morphology. Thus, a planetary system's dust can both obscure and indicate the presence of planets. This report will attempt to summarize the current state of knowledge about our solar system's dust component, relate it to dust discovered around some nearby main sequence stars, and compare methods for detecting and characterizing extrasolar zodiacal clouds on the road to finding extrasolar earth-like planets. III. Solar System Dust Interplanetary dust particles (IDP) are evident in the optical and IR surface brightness of zodiacal light easily observed from the Earth. The total IR-emitting (effective) cross-sectional area of grains between 1 and 2 AU from the Sun is about 2 x 1020 cm2. The grains characterizing the IR emission near Earth's orbit have typical sizes of 10-100 µm. The surface area given above thus corresponds to a mass of roughly 1018 - 1019 g for a density of 3 g cm-3, equivalent to a single solid body only 5-10 km in radius. The dust in the 1-2 AU annulus has an areal surface density (vertical optical depth) of ~ 1 x 10-7, increasing slowly toward the Sun. These characteristics will be referred to here as a "1-zodi" cloud (Good 1994; Backman et al. 1997). Table 1 gives the COBE DIRBE team's results (Kelsall et al. 1998) for albedo and emissivity of the zodiacal grains at 1-3 AU. The albedo in the near-IR is about twice that in the visible. Absolute emissivity should equal 1 - Albedo. Table 1: IDP Albedos and Emissivities l, µm Albedo Relative Emissivity 1.25 0.20 2.2 0.26 3.5 0.21 1.66 4.9 1.00 12 0.96 25 1.00 60 0.73 100 0.65 140 0.68 240 0.52 The bolometric luminosity ratio Ldust/Lsun is ~ 2 x 10-8 for the 1-2 AU annulus and is estimated to be about 1 x 10-7 for the entire zodiacal cloud. The latter value depends strongly on the dust density and density gradient close to the Sun. The luminosity of the hottest dust may not be a crucial issue in interferometric detection of exozodiacal systems and extrasolar planets because a central null intended to block stellar emission will also block the inner reaches of exozodiacal clouds. COBE and IRAS observations have little to say about the amount of dust at r < 0.9 AU. There are, however, constraints on the variation of dust density from 1 AU to 0.3 AU which imply that the volume density varies roughly as r-1.3 (Leinert et al. 1981). If the geometry of the cloud is a "wedge" (Figure 1; constant opening angle subtended at the Sun, scale height 'h' proportional to 'r'), then that volume density gradient is equivalent to face-on surface density varying as r-0.3. We do not know the position of the inner cutoff of the dust distribution, although it may correspond simply to the vaporization temperature of silicate grains. Observations close to the Sun during eclipses show the zodiacal dust merging smoothly into the F-corona, continuing in to 3 solar radii (0.015 AU) with volume density slope flattening slightly to r-1.0 (Mann et al. 1996). The details of dust density beyond 2-3 AU are also not well known. There is a result from Pioneer 10 observations that the radial decrease of zodiacal light intensity is about r-2.5 in the visible, equivalent to an r-1.5 volume density law (Hanner et al. 1976). Table 2 gives estimated polar zodiacal light surface brightnesses as if viewed from various positions out to 3.5 AU based on Pioneer, Helios, and COBE DIRBE data. Table 2: Polar Zodiacal Surface Brightness versus R Sn, MJy sr-1 R, AU 4.9 µm 12 µm 25 µm 0.5 6.5 46 51 1.0 0.33 11 18 1.5 0.04 4.0 9.2 2.0 ~0 1.9 5.3 2.5 ~0 1.0 3.4 3.0 ~0 0.57 2.3 3.5 ~0 ~0.35 ~1.6 A patch of the zodiacal cloud in our solar system at 1 AU with diameter 0.3 AU (viewed face-on) would have the same brightness as the planet Earth. This would be roughly true for both IR thermal emission and optical scattered light to the extent that the material has approximately the same albedo and temperature as Earth. Detecting earth-like planets around other stars in the face of this emission thus requires spatial resolution with this scale as well as strategies for distinguishing moving emitters from "fixed-pattern" brightness. III.A. Sources and Sinks for Planetary Dust} The sources of IDP include at least: asteroid collisions, comet activity and collisions in the inner solar system, comet collisions in the Kuiper Belt (KB), and ISM grains. The main physical processes affecting IDP are: expulsion by radiation pressure, inward Poynting-Robertson radiation drag, solar wind pressure (including significant electromagnetic effects), sublimation, mutual collisions, and gravitational influence of planets. No rigorous accounting has been made of the relative strengths of the grain sources, nor of the effects of the various destruction and dynamical processes on the equilibrium spatial distribution of the dust. Difficulties arise not simply because of complex cloud structure but also from problems in comparing multi-wavelength data sets which do not measure precisely the same lines-of-sight simultaneously and are often subject to large calibration error. Very different models of the spatial distribution of the material have produced acceptable fits to visual and IR data (e.g. Giese and Kniessel 1989). Much work remains to be done combining observational data with a priori models of the spatial distribution based on dynamical evolution of the grains. III.A.1. Processes Affecting Grains The direct radiation "overpressure" factor b = Frad /Fgrav is a function of grain size, density, and albedo (Burns et al. 1979). Values of b greater than 0.5 result in expulsion from the solar system of a grain released from a parent body in a circular solar orbit. Such values obtain for silicate grains smaller than roughly 1 µm radius. The time scale for expulsion is only of order the orbit period, i.e. essentially instantaneous compared with other time scales. Thus, this process effectively produces a lower limit to the grain size commonly found in solar orbit. Grains large enough not to be expelled spiral toward the Sun under the influence of PR drag. The PR time to de-orbit a black 30 µm-radius (typical) grain with density 3 g cm-3 starting from 1 AU is about 105 yr (Burns et al. 1979). This time scale is proportional to grain size for a given grain composition and proportional to the square of the initial orbit radius. The drift speed increases as r-1. In a "wedge" geometry, equilibrium between a constant dust creation rate and destruction via PR drag would produce a surface density proportional to r0, approximately as observed in Earth's vicinity. The dust mutual collision time scale is of order the orbit period divided by the cloud optical depth perpendicular to the symmetry plane (e.g. Backman and Paresce 1993). This is about 107 - 108 yr for the local zodiacal cloud, much longer than the PR drag time scale. This further indicates that local evolution is controlled mainly by PR drag. Those particles which do collide can produce fragments smaller than the radiation pressure "blowout" size and that material then exits the system rapidly. III.A.2. Dust Sources Approximately 50% (15-75%) of IDP in the 10-100 µm size range come from the asteroid belt, and 10% (5-25%) of the asteroidal dust comes from 3 known asteroid collision families representing ongoing interactions of debris from significant collisions in the past (Dermott et al. 1996a; Durda and Dermott 1997). The scale height of dust near the Earth indicates at least some of the local dust is cometary (Liou et al. 1995). The relative contributions of these and other sources are not well known. The Trojan asteroids (Jupiter Lagrangian population) may also be a significant source of collision debris drifting past Earth toward the Sun because their encounter times are orders of magnitude shorter than for main belt asteroids. The Trojan encounter velocities are not as large, however, so their fragment size spectrum will probably have a different shape than for main belt debris and the proportion of grains in the size range producing substantial mid-IR emission is unpredictable. Comet dust grains almost certainly have different characteristic sizes, compositions, and albedos from asteroid dust, and thus may behave differently in response to the various forces discussed above. Perihelion passage of a large comet such as P/Halley releases grains with surface area of order 10-5 of that in the entire zodiacal cloud. If the PR destruction time scale of typical cometary grains is assumed to be roughly 105 yr (as for asteroidal grains) then P/Halley may contribute 1% of the equilibrium zodiacal dust population during its (maximum) life of 105 yr ~ 103 orbits. Lower and upper limits on the Kuiper Belt dust parent body (comet nuclei) population (Cochran et al. 1995 and Backman et al. 1995, respectively) imply that the surface density of collisionally- produced KB dust is comparable to that in the inner solar system zodiacal cloud. These grains should drift toward the Sun via PR drag. However, for them to make a noticeable contribution to the inner solar system dust population, they must survive: a) collisions with interstellar grains, which may be especially important for larger and more slowly-drifting KB grains that are exposed to bombardment for longer intervals, and b) gravitational perturbation by the Jovian planets that can trap grains in resonances and/or eject them from the solar system; again, this should be more important for slowly-drifting large grains. Both effects are the subject of present modeling efforts (e.g. Liou et al. 1996). Gauging their efficiency is important because without these effects the relative densities indicate that a significant fraction of the inner solar system dust could actually be originally from the KB (Flynn 1996; Backman et al. 1997). Interstellar grains are expected to be a significant source of the smallest particles in the outer solar system. Ulysses detected ISM dust in the outer solar system coming from the direction of the solar motion through the galaxy (Grün et al. 1994). In-situ measurements indicate that ISM dust surface density is only of order 0.1% of IDP density at 1 AU (Grogan et al. 1996). III.B. Fluctuations in the General Zodiacal Cloud Density The following argument leads to a conclusion that the present state of our solar system zodiacal cloud is probably comparable to its "normal" state over the past ~ 109 yr: (1) Most of the present dust in the terrestrial-temperature zone is asteroidal rather than cometary. This is supported by Galileo and Ulysses observations, characteristics of the Earth's resonant ring, and properties of particles caught in Earth orbit by the LDEF (Long Duration Exposure Facility). (2) Collisions of 10 km-radius bodies should occur in the main belt approximately every 107 yr ("particle-in-a-box" approximation, assuming 104 such objects spread uniformly between 2.0 and 3.5 AU). One such collision could completely recreate the present zodiacal cloud. An example of a model history of the zodiacal cloud density (actually, of one family of collision debris) showing a general decline punctuated by large transient increases is presented in Figure 2. (3) The presence of 3 obvious asteroid collision debris families indicates that complete dispersal of one of these families occupies a time span very roughly 3 x the 107-yr interval between their initiation, much longer than the 106-yr time scale for clearing 10-100 µm grains from the main belt by PR drag. This is consistent with the expectation that dust from a major collision is released slowly via grinding of large co-orbiting fragments rather than immediately in the initial collision. (4) Although the 3 known collision families are directly responsible for about 10% of the general dust population, fragments from those families may be responsible indirectly for much more of the zodiacal cloud via erosion of other asteroids ("erosional cascade"). If that is actually the case, the general cloud density could be connected directly to the number and density of active collision debris families. (5) Zodiacal cloud enhancements by significant collisions last long enough to overlap in time so the overall normal state of the cloud is probably determined by the average of a number of collision families in various degrees of relaxation. (6) A crude estimate might then indicate that the normal variation about the present state would be 3 ± ÷3 major collision families active at a given time (with associated general belt asteroid erosion), i.e. variation by a factor of ~ 2. (7) Increases in the zodiacal cloud density by factors of 10 or more due to exceptionally large asteroid collisions, giant comets, or comet showers probably occupy only a small fraction of the total time. Thus, the tentative conclusion is that the present zodi cloud density and brightness is probably the normal state, within approximately a factor of 2, for the present asteroid and comet population. III.C. Long-term Evolution of the Zodiacal Cloud The history of the zodiacal dust cloud over the age of the solar system is likely to have been closely connected to the history of the main asteroid belt. It is safe to state that the total mass of the asteroid population has steadily decreased and will continue to decrease due to mutual collisions and planetary perturbations. A "bulge" in the asteroid size distribution (Zellner 1979) provides evidence that the original population of the asteroid belt had a size spectrum concentrated between 50-100 km radius. Many if not all of the presently abundant asteroids smaller than that size range are not primordial but are collision fragments. The total main belt asteroid population now has a collisional evolution time scale (population e-fold time) of a few x 1010 yr. The fact that this is significantly longer than the age of the system means that the evolution of the population is not "controlled" by mutual interactions but by something external, i.e. the planets. Many calculations have shown that asteroids in secular resonance with Jupiter (in the Kirkwood gaps) will eventually be thrown into planet-crossing orbits, some to exit the solar system. The winnowing of the asteroid belt is not just by collisions but also by planetary perturbations, so it is "chaotic" and the density of the asteroid belt and associated zodiacal cloud cannot be easily derived for past epochs. The present total mass of main belt asteroids is estimated to be 5 x 10-4 M‰. It is possible that the original asteroid belt mass was 1000 times larger than present, i.e. 0.5 M‰. The zodiacal dust released collisionally by that population could have been as much as 106 denser than the present cloud because collision rates depend on the square of the number of colliding bodies. Thus, the zodiacal cloud might have been enormously brighter earlier in the history of the solar system without a large belt mass relative to the major planets. A simple implication of this is that planet searches should be directed toward older stellar systems to avoid high dust density. III.D. Signs of Planets The large-scale structure of the zodiacal cloud is determined by planetary perturbations. The plane of symmetry of the cloud is inclined to the ecliptic and slightly warped, and the Sun is offset from the cloud's center of rotational symmetry (Dermott et al. 1996b). These offsets might be discernible in some circumstances in other systems and allow inference of the existence of planets. COBE observations (Figure 3) have recently confirmed that the zodiacal cloud near Earth has a marked trailing/leading asymmetry due to the trapping of dust particles with small orbit eccentricities into orbital resonances with Earth (Figure 4) (Dermott et al. 1994; Reach et al. 1995). There is a ring of dust that co-revolves with Earth around the Sun; Earth resides in a cavity in this ring and a cloud of enhanced dust density permanently trails the Earth in its orbit. The dust number density in the trailing feature peaks about 0.2 AU behind the Earth and is estimated to add between 10% and 40% (Dermott et al. 1994 and Reach et al. 1995, respectively) to the local smooth cloud density over an ecliptic plane area several x 0.1 AU diameter. Additional work is needed to better define the true size of the wake signals, but the Earth's wake has an IR signal at least 0.1x that of Earth. Each planet probably has such a wake. The width and relative enhancement of a wake is determined by the orbital period of the planet, the typical grain PR drift velocity, and the mass of the planet. The absolute wake density probably remains proportional to the smooth cloud density, within some limits, as the latter varies with fluctuations in intensity of grain injection processes. Wakes in exozodi clouds could be mistaken for planets, depending on spatial resolution, but it is important to note that a wake can be a prominent indicator of the presence of a planetary mass. The wakes should be distinguishable from planets via spectroscopy. IV. Other Stars At least 15% of nearby normal main sequence stars of all spectral types have cold dust populations with spatial scales corresponding to our system's Kuiper Belt (30-100 AU) (Backman and Paresce 1993) (Figure 5). These are sometimes called "Vega / b Pic" disks after two nearby especially prominent prototypes. The IRAS detection limit for dust around nearby stars corresponds to a fractional dust luminosity limit of Ldust/Lstar > 10-5, about 100x the value for our zodiacal cloud and also 100x a model upper limit on the amount of dust in our KB (Backman et al. 1995). There is no obvious correlation within the present small-number statistics between stellar type and average amount of dust. Only a few main sequence stars have warm circumstellar dust (i.e., at terrestrial temperatures) detectable with IRAS sensitivity (Aumann and Probst 1991). Three examples, b Pic, z Lep, and 51 Oph, have Ldust/Lstar values for warm dust in the range 10-5 to several x 10-4 (Fajardo-Acosta et al. 1998). It is crucial to note that some models of the construction of the planets in our solar system indicate that the presence of an asteroid belt is not guaranteed. Numerical experiments by Wetherill (1992) yield one or more large bodies instead of an asteroid population in 50% of runs. This is counter to the common intuition that an asteroid belt must form just inside the orbit of the first ice giant due to gravitational disturbances by that planet. These model systems otherwise resemble ours, with a set of terrestrial planets and a set of Jovian planets. Thus, complementing the point raised in section III.C that planetary systems can have much larger asteroid and asteroidal dust populations than does our system, present knowledge indicates that a system with an earth- like planet could also have much less terrestrial-temperature dust than ours does. Recent discovery of nearby planetary systems containing "hot Jupiters", Jovian-mass or larger objects at 0.05 to 2.5 AU from the stars, indicates the variety of planetary systems that is apparently possible (Butler et al. 1997). Of course such systems will show up first and most easily in radial velocity searches, so their true prevalence is unknown. However, recent theoretical work by Lin et al. (1997) and Tremaine (1997) explain these objects as results of processes of large-planet migration in the protoplanetary disk caused by drag from either remnant gas or dense planetesimal populations. In "hot Jupiter" systems, terrestrial material (planets, asteroids, and dust) would likely have been erased by the inward migration of the large planets. In contrast, a KB-like zone of planetesimals seemingly is guaranteed beyond the planets in the region where the protoplanetary disk density was too low and encounter times too long to support construction of planets. It seems possible that cold dust like in the "Vega / b Pic" systems may be detected eventually around most main sequence stars (Dominik et al. 1998). KB-like systems with Ldust/Lstar < 10-5 would be expected to send grains via PR drag toward their central stars because the mutual grain collision time scale would be longer than the PR time scale. However, as noted above, few main sequence KB-systems have significant grain populations at temperatures above 150 K, indicating that central "voids" are common. This can be explained as being due to a combination of: a) erosion of inbound grains by interstellar dust, b) dynamical influence of large outer planets preventing grains from drifting close to the primary stars, and c) present inability to detect zodiacal systems with Ldust/Lstar much below 10-5 such that KB-supplied inner zodiacal clouds could have escaped notice to date. Thus, it seems that the amount of cold dust in a system cannot be used to predict the amount of hot dust because the presence of hot dust is not guaranteed even if there is a planetary system, and the cold dust may not be able to drift close enough to the star to become hot dust. In terms of the planet-finding problem, observational and theoretical studies of both the "Vega / b Pic" KB-like dust disks and of the dust population in our outer solar system should help us discover whether KB-like systems are necessarily remnants of the planet-formation process and thus signposts of planetary systems, and whether central "voids" require the influence of planets for their maintenance. IRAS and ISO have not been able to detect hot dust at the 1-zodi level around nearby stars because their low spatial resolutions prevent separation of dust emission from stellar photospheres. The best these instruments can manage in the face of uncertainty about photospheric fluxes is to discover nearby dust systems down to the few x 100-zodi level. SIRTF, on the other hand, will have detector arrays with sub-diffraction pixels. With this resolution and higher sensitivity it should be better than its predecessors at detecting thin dust around some of the nearest stars (Backman et al. 1997). Unfortunately a Centauri, the nearest system (1.3 pc), lies at a galactic latitude of -0.7o and is projected against very bright interstellar background emission, so it may not be easily investigated. V. Optical Detection Techniques V.A. Zodiacal Scattered Light A 1-zodi cloud would have a face-on surface brightness of about V = +22 mags arcsec-2 at 1 AU from the primary star. This would be entirely scattered/reflected light from the primary. Figure 6 gives the zodiacal surface brightness at 1, 2, 3, 4, and 5 AU from the Sun (upper to lower curves, respectively) as a function of wavelength. The reflected sunlight component from about 0.4 to 4 µm is modeled from the observed visible zodiacal light, and the thermal infrared emission from 5- 20 µm is based on the IRAS model. The transition from scattered light to thermal emission occurs at about 3.5 µm for our local zodiacal cloud. The near-IR brightness is about the same or slightly brighter than in the visible because the grains are grey or slightly red. The surface brightness of exozodiacal light (extended emission) with a given optical depth at a chosen angular distance from a primary star will be proportional to the stellar apparent brightness. Exozodiacal optical surface brightness will not be easily distinguished from background and instrumental sources of scattered light without detecting a radial "edge" or azimuthal asymmetry in the exozodiacal cloud. Based on experience with galaxy surface photometry, detection is probably possible to exozodiacal surface brightness as low as 6 mags arcsec-2 below the background, which will require a dramatic reduction in the normal background for the relevant wavelengths, apertures, and angular scales. Exozodiacal cloud detection will be much more difficult than, e.g., finding galaxies around quasars (Ftaclas 1998). V.B. HST and NGST The superior surface accuracy of HST plus advanced-technology coronagraphic cameras expected in the near future can probably reach the 10-zodi level. An 8-m NGST with coronagraph and "slow" adaptive optics might detect the visual-l surface brightness of a 1-zodi cloud at 10 pc in a diffraction-limited beam if the instrument can compensate for NGST's expected modest surface accuracy compared with HST. The envisioned NGST instrumentation includes an adaptive mirror for "static" wavefront correction over a very small field (1 arcsec diameter). The post-correction background brightness would consist of residual diffraction and scatter from actuator noise (wavefront sensor errors, position errors, etc.). A critical concern is the unknown angular-scale power spectrum of that post-correction background. If it is "white noise" (Figure 7a), calculations show that the adaptive mirror needs to be corrected to an accuracy of about 2 Å. This requirement could be relaxed in some cases of background spatial noise that is not "white" (Figure 7b). Note that the exozodiacal detection problem at visual wavelengths with an 8-m aperture is more difficult than directly detecting Jovian planets and comparable to detecting individual terrestrial planets (Ftaclas 1998). VI. IR Interferometric Detection VI.A. Zodiacal IR Emission The most relevant feature of the exozodiacal measurement problem in the IR is the strong 10 µm foreground. The preferred wavelength band for these measurements is around 10 µm because of: a) favorable contrast ratio between the exozodiacal signal and the star, b) presence of a terrestrial atmospheric window, and c) existence of spectral features of substances such as O3 and CH4 that might indicate a non-equilibrium atmosphere and the presence of life (ExNPS 1996). However, the 10 µm exozodiacal signal will be deeply embedded in foreground terrestrial and local zodiacal emission. Figure 8a shows the face-on optical depth of the smoothly varying portion of the solar system zodiacal dust as a function of radius from near the Sun to 100 AU (Traub et al. 1996) for the COBE model parameters (Kelsall et al. 1998). The COBE model spatial distribution t^ ~ (0.71 x 10-7)(rAU-0.39) was assumed valid from near the Sun to 100 AU. That figure also shows typical temperatures in the cloud assuming emissivity = 1 (cf. Table 1). The inset scale gives angular radii for an observer at a distance of 10 pc; thus the Earth lies at a radius of 100 mas and the solar radius is 0.44 mas. Figure 8b displays the corresponding face-on surface brightness (also called specific intensity or spectral radiance) in frequency units at wavelengths of 8, 10, and 12 µm. Note the steep fall-off with radius caused by weak blackbody flux at 10 µm for temperatures below about 300 K. The IRAS zodiacal model has a steeper radial gradient, t^ ~ (1.12 x 10-7)(rAU-0.80), than the COBE model. The COBE and IRAS models are both constrained by observations relevant to material between about 0.9 and 2 AU and are similar in that region. Extrapolations of these models down to a few solar radii is evidently dangerous: they differ there by almost a factor of 10 in surface density. Also, to the extent that much of the dust originates in the asteroid belt or in solar- activated comets, there is probably a substantial break in density beyond 3-5 AU. Figure 9 shows the integral power density from the Sun's surface outward for the IRAS model. Note that 50% of the power comes from the region inside 0.1-0.2 AU, depending on the wavelength, and 90% of the total power is included inside a radius of 0.5-1.5 AU. For the COBE model (not shown) the corresponding radii are 0.2-0.5 AU for the 50% points and 1.0-2.0 AU for the 90% points. An exozodiacal dust-detecting interferometer needs to be sensitive in this range of radii. VI.B. Bracewell Interferometry - Magellan As a specific example of a potential exozodiacal interferometer, Figure 10a displays the antenna pattern of the 6.5-m Magellan I telescope and Figure 10b the fringe pattern formed by the 60-m separation of Magellan I and II. The single-telescope "beam gain" is rotationally symmetric, being just the diffraction pattern of a centrally obscured circular aperture. The two-telescope "fringe gain" is a linear pattern of interference fringes projected onto the sky, where the fringe orientation is perpendicular to the projected baseline between the telescopes and additionally averaged in azimuth. The product of these two gains can be directly multiplied point-by-point across a model exozodiacal map to obtain the signal in a single detector element. The fringe phase is adjusted to give a central null on the star, blocking most of the star's glare. Note that the single-telescope diffraction-limited beam extends out to about 200 mas, or about 2 AU radius at 10 pc, which is sufficient to accept essentially all of the model dust emission at 10 µm. Also note that (in star-nulling mode) the inner interference fringe has a transmission peak at about 10 mas or 0.1 AU, which is an excellent match to the 50% power points of the dust emission noted above. The product of the IRAS model dust emission at 10 pc and the antenna pattern of the Magellan interferometer results in reduction of total IRAS model power observed at 10 µm from about 0.20 mJy to about 0.06 mJy due to the interferometer gain pattern. Finally, note that stellar leakage through the central null of an ideal sinusoidal interference pattern varies as the square of the ratio of the projected star diameter to fringe period. Wavefront perturbations determine a Strehl ratio (on-axis relative power), the departure from unity of which governs how much additional star light leaks into the central diffraction-limited mode. Here the combined leakages are a factor of a few greater than the dust emission signal, showing that the initial contrast ratio of star/dust ~ 10,000 can be reduced to a more manageable value in the range 1-10. A 3s detection of the partly-nulled 10 µm flux given above for a 1-zodi cloud at 10 pc is possible in about 2 hours assuming smooth telescope mirrors, a cooled interferometer, and noise dominated by fluctuations in telescope thermal photon emission (Traub et al. 1996). VI.C. IR Space Interferometer A thermal-IR interferometer located in space would have obvious advantages in sensitivity to exozodiacal clouds and extrasolar planets over ground-based facilities (Bély et al. 1998). For a space mission where telescope diameter is especially critical to cost, the background power needs to be well known in advance for the success of the mission because noise depends on the ratio of (background intensity) to (4th power of telescope diameter). Figures 11a and 11b compare the signal/noise ratios for exozodiacal dust detection by interferometers located at 1 and 5 AU from the Sun, respectively. The set of curves illustrate performance by a range of sizes for the 4 unit telescopes in a baseline Terrestrial Planet Finder (TPF) instrument design (ExNPS 1996). An important performance criterion for exozodiacal dust and extrasolar planet detection is that telescope noise should be less than local zodiacal noise. By equating the corresponding surface brightnesses and solving for the temperature, an upper limit for the telescope temperature in terms of the telescope emissivity and local zodiacal brightness can be found. For a nominal value of emissivity the telescope temperature must be Ttel £ 69 K if the spacecraft is at 1 AU or £ 54 K if at 5 AU. There is no clear need to cool a space interferometer's telescopes much below these values (Traub et al. 1996). VI.D. Achromatic Interferometery - Keck The Keck Interferometer will have the following unique features relevant to the detection of the zodiacal emission around other stars: High Strehl (phasing): Adaptive optics at each aperture to provide accurately phased wavefronts. Phase-referenced operation (cophasing): Phase referencing, using high-bandwidth fringe tracking on an off-axis star or on the target star at a different wavelength, to cophase the interferometer and provide high sensitivity and high path-length stability. Interferometric nulling between the two apertures reduces the dynamic range between the central star and the exozodiacal signal to enhance detection. Additional nulling interferometers at each aperture can be used to calibrate the background and the residual leakage of the central star. With this approach, detection of exozodiacal signals as small as 10- or even 1-zodi should be possible. The Supplement section to this report compares achromatic single and dual-nulling operation of an instrument with the baseline characteristics of the Keck Interferometer. VII. Panel Recommendations (1) Support should be given for implementation of one or more ground-based large-aperture IR interferometers capable of surveying main sequence stars within 10 pc to detect exozodiacal clouds to a sensitivity level of 10-zodi or better. This will allow identification of the least dusty systems as a precursor to IR searches for earth-like planets. (2) A space-based exozodiacal mapper less sensitive than a planet-search instrument could find cloud asymmetries and planetary dust "wakes" that would indicate the presence and locations of planets and discern the orbit plane orientation in systems toward which the planet-search instrument could then be directed. (3) Extensive observational and theoretical studies of the structure and history of our local zodiacal cloud should be made to: a) allow better interpretation of observations of exozodiacal clouds, and b) better characterize foreground emission that will interfere with exozodiacal mapping and extrasolar planet searches, especially allowing determination of the best location in our solar system for a planet-search instrument. (4) Some support should be given to projects aimed at determining the age of field stars of solar type because the average dust surface brightness should decrease with time in each system. Searches for earth-like planets should be directed first toward older systems. S. Technical Supplement Regarding IR Interferometry S.1. Assumptions about the Instrument and Target System Table S1 displays assumptions about a facility with characteristics like the planned Keck Interferometer on Mauna Kea on which sensitivity estimates below are based. Table S1: Instrument assumptions Aperture d 10 m Baseline D 85 m Center l lo 10 µm Fractional bandwidth Dl / lo 0.3 Background temperature T 273 K Etendue AW 1 l2 Emissivity e 0.5 AO Strehl ratio at lo S 0.984 Effective efficiency h 0.1 The value for etendue in Table S1 assumes a single-mode filter as discussed in section S2, and the adaptive optics Strehl ratio assumes 200 nm rms wavefront errors. The effective efficiency includes adaptive optics system transmission, starlight relay transmission, diffraction losses, dewar internal losses, coupling losses into the single-mode filters, and detector quantum efficiency (assumed to be 0.75). A further efficiency factor of 0.5 is included to account for use of a nulling beam combiner for the exozodiacal signal. Table S2 gives the signal and signal/noise ratios if the observation target is assumed to be the portion of an exozodiacal cloud located 1 AU from a sun-like star at a distance of 10 pc. The background flux 'B' and the flux from the central star 'S' in two apertures is 7 x 1010 and 1.5 x 108 electrons s-1, respectively. Table S2: Fluxes and Signal/Noise Ratios without Nulling Exozodi density (solar zodi units) 100 10 1 Exozodi flux Z (2 apertures, e- s-1) 1.3 x 106 1.3 x 105 1.3 x 104 Exozodi photometric S/N (t = 104 s) 500 50 5 Exozodi excess over star (Z/S) 1% 0.1% 0.01% Exozodi-to-background ratio (Z/B) 2 x 10-5 2 x 10-6 2 x 10-7 From a strictly photometric perspective, detection of the excess from a 1-zodi cloud is possible. From a practical perspective, two points need to be addressed: 1) the small relative size of the exozodiacal excess relative to the star (0.01% for the 1-zodi case), and 2) the small relative size of the exozodiacal excess relative to the mean IR background (2 x 10-7 for the 1-zodi case). The latter point is probably the most challenging. An approach to problem 1) involves a nulling beam combiner between the two apertures to suppress the light from the central star, increasing the relative size of the exozodiacal excess. An approach to problem 2) adds additional nullers at the individual apertures. These nullers serve as source choppers to provide an accurate calibration of the background using only OPD (optical path differential) modulation. S.2. Single-nuller Configuration Figure S1 is a block diagram of an interferometer configuration using an achromatic nulling beam combiner to combine light from the two apertures. The nulling combiner is similar in principle to a Bracewell interferometer that combines the light from the two apertures out-of-phase in order to provide starlight cancellation. However, the achromatic nulling configuration uses polarization rather than a physical path delay to generate an achromatic 180o phase shift, allowing nulling over a wide spectral band. The delay lines in the figure maintain equality between the two arms of the interferometer. The depth of the achieved null is limited by the size of the source as well as various instrumental and residual atmospheric errors, as discussed in the next section. S.3. Limitations to Null Depth Finite star diameter: Null depth is limited by the relative spatial extent of the object. For star diameter q = 1 mas and lo/D = 24 mas, the null depth is dq = 0.002. Optical path jitter: Deep nulls require a high degree of pathlength stability between the two arms of the interferometer. The amount of variation is a function of the atmospheric parameters and the bandwidth of the interferometer fringe tracker. For median seeing at Mauna Kea and a 1 kHz closed-loop bandwidth the null depth dO = 1.2 x 10-5. Note that there will be adequate photon flux from a bright central star for at least a factor of 10 faster fringe tracking than this example, and aggressive tracker algorithms can provide deeper nulls for the same sampling rate. Wavefront aberration: Corrugations in the stellar wavefront caused by imperfect optics reduce the interferometer fringe visibility and hence the null depth. Relative to the combined intensity of the two beams, the null depth dS = 1 - S, where S is the Strehl ratio of an individual telescope with its AO system. Without further compensation, this limits the null depth dS to 0.016. However, it is possible to do significantly better by using a single-mode spatial filter to remove wavefront aberrations. Scintillation: Unequal intensities between the two beams of the interferometer produce imperfect fringe visibility. We can approximate the instantaneous scintillation as the instantaneous change in the Strehl ratio. For the example above with S = 0.984, the estimated scintillation standard deviation would be 0.008, which would set the null depth to 1.6 x 10-5. This yields a total null depth random component of dL ~ 3 x 10-5 when combined with the path jitter from the 1 kHz fringe tracker described above. Table S3 repeats the calculation in Table S2 but with the nulling in place. The background is still assumed to be 7 x 1010 electrons s-1. We assume systematic leakage dq = 2 x 10-3 and a total random leakage dL = 3 x 10-5. As a result, the components of starlight leakage will be dqS = 3 x 10-5 and dLS = 5 x 103 electrons s-1, respectively. We now find that the exozodiacal excess is readily detectable compared to the star. Table S3: Fluxes and Signal/Noise Ratios with Nulling Exozodi density (solar zodi units) 100 10 1 Exozodi flux Z (2 apertures, e- s-1) 7 x 105 7 x 104 7 x 103 Exozodi photometric S/N (t = 104 s) 250 25 2.5 Exozodi excess over starover star (Z/systematic) 200% 20% 2% Exozodi excess over star (Z/leakage) 10000% 1000% 100% Exozodi-to-background ratio (Z/B) 1 x 10-5 1 x 10-6 1 x 10-7 The exozodiacal signal is reduced in Table S3 relative to Table S2 by a factor of 2 to represent some of the near-star dust emission being nulled along with the star. Important problems remaining to be considered (section S.4) include calibrating the background and the amount of random leakage through the null. S.4. Dual-nuller Configuration Figure S2 shows a configuration that results when additional nullers are added to the individual apertures. In this configuration, each aperture is divided in the pupil plane and its light fed into nulling combiners. The effective baseline for these interferometers is approximately D' = 5 m. The output of these aperture nullers is then fed into the interferometer nuller already described. A second interferometer nuller is added to use the second nulled output of each aperture nuller. Four delay lines are shown in the system, one preceding each input of the aperture nullers. The nullers themselves maintain fixed internal pathlengths for background stability. By adjusting the optical delay lines, all combinations of aperture nulling "on" or "off" and interferometer nulling "on" or "off" can be achieved. As discussed above, the interferometer nuller cancels most of the starlight, leaving residuals dq (finite stellar diameter) and dL (into which are lumped the leakage terms attributable to path jitter and scintillation). Part of the dust emission also gets nulled, leaving residual aZ where a ~ 0.5. With their much smaller baseline, the aperture nullers will cancel all the star light [strictly, a factor of (D/D')2 more than provided by the interferometer nuller] as well as some of the exozodiacal emission, leaving residual eZ. Signals provided by the various nulling combinations are listed in Table S4. Table S4: Signals for Various Nulling Combinations Mode Aperture Nuller Interferometer Nuller Signal (a) off on 2S + Z + B (b) off on dqS + dLS + aZ + B (c) on off dLS + eZ + B (d) on on dLS + eZ + B Switching among these combinations requires only changes in pathlength delay, not pointing. It is accomplished by motion of a PZT transducer by 5 µm and can occur quickly. If only OPD modulation external to the nulling interferometers is used the background does not change; thus, this approach can be used as a source chopper in order to calibrate the background. This procedure is essential to allow detection of faint exozodiacal signals. The same approach can also be used to calibrate the random star leakage. For changes in the instrument background and star leakage slower than the chop rate, the difference between modes (b) - (d) is a "white noise" process which should average out with integration time as 1/÷t. In addition, since S >> Z, mode (a) can be used to calibrate S, which, along with knowledge of the star's diameter, can be used to calibrate the exozodiacal excess. This dual-nuller approach therefore should allow the detection of a weak exozodiacal signal embedded in a strong background. References Aumann, H. H. & Probst, R. G., 1991, Ap. J., 368, 264 Backman, D. E. & Paresce, F., 1993, in "Protostars and Planets III", ed. E. H. Levy & J. I. Lunine (University of Arizona Press), 1253 Backman, D. E., Werner, M. W., Rieke, G. H. & Van Cleve, J. E., 1997, in "From Stardust to Planetesimals", ed. Y. J. Pendleton & A. G. G. M. Tielens, ASP Conference Series, 122, 49 Backman, D. E., Dasgupta, A. & Stencel, R. E., 1995, Ap. J. Lett., 450, L35 Bély, P. Y., Burg, R., Petro, P., Beichman, C. A., Wade, L., Gay, J., Baudoz, P. & Rabbia, Y., 1998, Experimental Astronomy, in press Burns, J., Lamy, P. L. & Soter, S., 1979, Icarus, 40, 1 Butler, R. P., Marcy, G. W., Williams, F., Hauser, H. & Shirts, P., 1997, Ap. J. Lett., 474, L115 Cochran, A., Levison, H. F., Stern, S. A. & Duncan, M. J., 1995, Ap. J., 455, 342 Dermott, S. F., Jayaraman, S., Xu, Y.-L., Gustafson, B. Å. S. & Liou, J.-C., 1994, Nature, 369, 719 Dermott, S. F., Grogan, K., Gustafson, B. Å. S., Jayaraman, S., Kortenkamp, S. & Xu, Y.-L., 1996a, in "Physics, Chemistry, and Dynamics of Interplanetary Dust", ed. B. Å. S. Gustafson & M. S. Hanner, ASP Conference Series, 104, 143 Dermott, S. F., Jayaraman, S., Xu, Y.-L., Grogan, K. & Gustafson, B. Å. S., 1996b, in "Unveiling the Cosmic Infrared Background", ed. E. Dwek, AIP Conference Proceedings, 348, 25 Dominik, C. and the HJHVEGA Consortium, 1998, in "ISO's View of Stellar Evolution", ed. R. Waters, C. Waelkens & K. van der Hucht (Kluwer) Durda, D. D. & Dermott, S. F., 1997, Icarus, 130, 140 ExNPS, 1996, "A Road Map for the Exploration of Neighboring Planetary Systems", ed. C. A. Beichman, JPL Pub. 96-22 (JPL) Fajardo-Acosta, S. B., Telesco, C. M. & Knacke, R. F., 1998, Astron. J., in press Flynn, G. J., 1996, in "Physics, Chemistry, and Dynamics of Interplanetary Dust", ed. B. Å. S. Gustafson & M. S. Hanner, ASP Conference Series, 104, 171 Ftaclas, C., 1998, these proceedings Giese, R. H. & Kniessel, B., 1989, Icarus, 81, 369 Good, J., 1994, in "IRAS Sky Survey Atlas Explanatory Supplement", ed. S. L. Wheelock et al., JPL Pub. 94-11 (JPL), Appendix G Grogan, K., Dermott, S. F. & Gustafson, B. Å. S., 1996, Ap. J., 472, 812 Grün, E., Gustafson, B. Å. S., Mann, I., Baguhl, M., Morfill, G. E., Staubach, P., Taylor, A. & Zook, H. A., 1994, Astron. & Ap., 286, 915 Hanner, M. S., Sparrow, J. G., Weinberg, J. L. & Beeson, D. E., 1976, in "Interplanetary Dust and Zodiacal Light" (IAU Colloquium 31), ed. H. Els\"{a}sser & H. Fechtig (Springer- Verlag), Lecture Notes in Physics, 48, 29 Kelsall, T. N. et al., 1998, submitted to Ap. J. Leinert, C., Richter, I., Pitz, E. & Planck, B., 1981, Astron. & Ap., 103, 177 Lin, D. N. C., Bodenheimer, P. & Richardson, D. C., 1996, Nature, 380, 606 Mann, I., 1996, in "Physics, Chemistry, and Dynamics of Interplanetary Dust", ed. B. Å. S. Gustafson & M. S. Hanner, ASP Conference Series, 104, 171 Liou, J.-C., Dermott, S. F., & Xu, Y.-L., 1995, Planet. Space Sci., 43 (6), 717 Liou, J.-C., Zook, H. A. & Dermott, S. F., 1996, Icarus, 124, 429 Reach, W. T., Franz, B. A., Weiland, J. L., Hauser, M. G., Kelsall, T. N., Wright, E. L., Rawley, G., Stemwedel, S. W. & Spiesman, W. J., 1995, Nature, 374, 521 Traub, W. A., Carleton, N. P. & Angel, J. R. P., 1997, in "Science with the VLTI Interferometer", ed. F. Paresce (Springer-Verglag), 80 Tremaine, S., 1997, Bull. American Astron. Soc., 29 (5), 1207 Wetherill, G. W., 1992, Icarus, 100, 307 Zellner, B., 1979, in "Asteroids", ed. T. Gehrels (University of Arizona Press), 783 Figure Captions Figure 1: Model cross-section of the solar system zodiacal dust cloud showing a "wedge" shape consistent with grain drift inward from the asteroid belt under the influence of PR drag. Densities at the latitude extrema are enhanced, reflecting orbits inclined to the mean plane executing simple harmonic motion along the vertical axis. The inner cutoff represents termination of the model calculation rather than a real discontinuity in the cloud. Figure 2: Model evolution of the total cross-section area of dust and other objects in a debris family generated by collision of two large asteroids at t = 0. The declining trend represents steady removal of particles by PR drag. The transient increases result from later collisions of large fragments within the swarm. Figure 3: COBE 25 µm surface brightness of the zodiacal dust at ecliptic latitude 0o as a function of Earth's orbital longitude. The upper and lower curves represent observations pointing directly behind and in front of the Earth along its orbit, respectively. The clear difference reveals the presence of a "wake" of dust trapped in orbital resonance behind the earth. The sinusoid pattern around the year results from Earth's vertical motion relative to the symmetry plane of the cloud which is not quite identical to the ecliptic. Figure 4: Model of the motion of a particle trapped in a resonance with the Earth, relative to a coordinate frame revolving with Earth. The axis units are AU. The resonance holds the particle in a ring at ~ 1 AU, temporarily interrupting its journey via PR drag from the asteroid belt into the Sun. Figure 5: IRAS 25/60 µm versus 12/25 µm color-color diagram for A, F, G and K main sequence stars in the Bright Star Catalog surrounded by planetary material. The excluded region contains the colors of free-free (plasma) emission. The color of a pure photosphere with no circumstellar dust would be off the diagram to the upper right. The trajectories trace colors of disks structurally identical to the ones around b Pic and a PsA but with varying total amounts of dust in contrast to the stellar photospheres. Figure 6: Spectral energy distribution of the zodiacal cloud surface brightness as a function of wavelength and of distance from the Sun. The nested curves are for dust at 1 AU (highest) to 5 AU (lowest) from the sun. Each SED has a minimum in the near-IR at the transition from scattered solar light to intrinsic thermal emission; the transition moves to longer wavelengths for cooler dust. Figure 7: Zodiacal light surface brightness at visual wavelengths compared with solar and planetary diffraction patterns seen from 10 pc by an NGST-class telescope. a) Solar diffraction reduced by "slow" adaptive optics in the case of system noise with a "white" spatial spectrum. b) Same, but for a plausible case of "non-white" noise. Figure 8: Thermal model of a 1-zodi cloud around a solar-type star viewed from 10 pc based on the COBE zodiacal model. a) Disk vertical optical depth (t^) and temperature versus distance from the central star, compared with a possible profile of the b Pic disk. b) Thermal emission surface brightness versus distance from the star for several wavelengths around 10 µm. Figure 9: Integral power versus distance from the star for a 1-zodi cloud around a solar-type primary viewed from 10 pc, based on the IRAS zodiacal model. Figure 10: Magellan telescope diffraction and fringe pattern sensitivities versus distance from a central star viewed from 10 pc. a) Single telescope diffraction. b) Interferometer fringes from both Magellan telescopes combined, employing central Bracewell nulling. Figure 11: Signal/Noise ratio for a space IR interferometer detecting Earth at 10 µm from a distance of 10 pc in 104 seconds integration. The family of curves shows sensitivies for a range of sizes of 4 unit telescopes along a 75-meter array baseline. The horizontal axis is density of the exozodiacal cloud in which the earth-like planet is embedded, in multiples of our 1-zodi cloud. "Struct = 0.0" means these models assume the exozodiacal cloud is perfectly smooth and structureless. a) Case of an interferometer located at r = 1 AU. b) Interferometer located at 5 AU. Figure S1: Block diagram for achromatic single-nulling on a 2-telescope interferometer. Figure S2: Block diagram for achromatic dual-nulling (once at each aperture plus once for the 2- telescope interferometer as a whole).